AI Visibility Current edition · October 2026 Evening & Occasion Wear

999 responses, 5 models, 393 entities: Trendyol leads, Carmen is second, and the shelf stays mostly domestic even when the question does not ask for Turkish brands

Mert Can Elkaya Published 5 October 2026 56 min read Herm.io AI Visibility Database (direct LLM querying)
In short

Five AI assistants were asked 40 Turkish evening and occasion wear (abiye) questions up to five times each (999 responses). Trendyol leads the general board and Carmen is second. On the 20 questions that ask for Turkish brands, domestic entities take 100.00% of qualifying mentions and foreign entities 0.00%; on the 20 origin-neutral questions, 83.21% and 16.79%. Origin is on file for every qualifying mention on both. Visibility measures findability, not quality.

Key facts
  • 999 responses: 40 Turkish questions × 5 models × up to 5 repeats, across 200 question-model cells, collected on 2 October 2026.
  • Leader Trendyol: 42.44% mention rate (424 / 999), 100.00 AI Visibility Score, named by 5 of 5 models.
  • 393 entities tracked (292 domestic, 61 foreign, 40 with no origin on file); 38 qualify on the general cut.
  • AI Visibility Score = 0.45 × Mention + 0.3 × Position + 0.25 × Breadth, scaled within each cut.
  • 213 cited domains, counted in question-model cells; trendyol.com is cited in 84 of 200 cells (42.00%).
Total Responses
999
Questions Analyzed
40
Entities Tracked
393
Qualified Entities
38

Vertical: Evening & occasion wear (abiye / davet giyim), Turkey
Method: A single point-in-time study of 999 responses — five large language models (Claude, Gemini, ChatGPT, Perplexity, Grok), each asked 40 Turkish-language questions up to five times
Collection date: 2 October 2026
Edition: 2 of a quarterly series. Edition 1 (July 2026) asked the same 40 questions; two of its five models were replaced for this edition.

Key terms: AI visibility, AI Visibility Score, evening wear brands, occasion wear brands, abiye / davet elbisesi, Turkish (yerli) evening-wear brands, LLM brand recommendations, ChatGPT/Gemini/Claude/Perplexity/Grok brand recommendation, Generative Engine Optimization (GEO).

This report reproduces the real Turkish-language questions consumers ask AI assistants about evening dresses, abiye and occasion wear, and measures how often, in what order, and across how many models five large language models name each brand, retailer and marketplace. Entities are ranked separately within six cuts of the same questions — all questions, origin-constrained, origin-neutral, and three question types — defined in Chapter 2. It is a single point-in-time study and should be read within the limitations in Chapter 11.

How to read this report: visibility is not quality

This report measures one thing: how often, how prominently, and across how many AI assistants a brand is named when people ask about evening and occasion wear in Turkey. A high score means these systems currently have a lot of information available about a brand and surface it readily. It does not mean the brand is better, more elegant, cheaper or higher quality. A low score, or a zero, means a brand is currently less visible to these systems, not that it is inferior. AI visibility reflects information availability and discovery, not product quality, fit, price or trustworthiness. Readers and shoppers should not treat an AI recommendation, or this report, as a verdict on whether a brand is “good” or “bad.” This study does not measure the accuracy, quality, or sentiment of any recommendation.


1. Executive Summary

One-sentence takeaway: Trendyol leads the general board and Carmen is the most visible entity that is not a marketplace, on a shelf that is entirely domestic when the question asks for Turkish brands and mostly domestic when it does not — domestic entities take 100.00% of qualifying mentions on the 20 origin-constrained questions and 83.21% on the 20 origin-neutral ones (foreign 0.00% and 16.79%), with origin on file for 2,485 of 2,485 and 1,769 of 1,769.

  • Trendyol leads; Carmen is second. Trendyol, a marketplace, tops the general board at 100.00, named in 42.44% of all 999 responses (424) and by all five models; it holds the cut’s highest mention rate and its highest MRR, 0.4766. Carmen (93.77) is named in 38.44% (384 of 999). The ZKS (79.74), ADL (73.23) and Abiyefon (69.13) complete the top five. ADL is named slightly more often than The ZKS, in 26.03% of responses against 25.83%, but later: its MRR is 0.3279 against The ZKS’s 0.4347.
  • 38 of 393 tracked entities qualify on the general cut — 35 domestic and 3 foreign — with an average score of 52.64. The foreign qualifiers are Mango (#14, 53.01), Zara (#22, 47.58) and Oleg Cassini (#28, 42.29). Three marketplaces — Trendyol (#1), Modanisa (#11) and Hepsiburada (#13) — rank alongside the brands, because they are what the assistants name. With the three removed, 35 entities remain and Carmen heads the brand-only board.
  • The question changes how domestic the shelf is, not who leads it. On the 20 origin-constrained questions (499 responses), domestic entities take 100.00% of qualifying mentions and foreign entities 0.00%, with origin on file for 2,485 of 2,485. On the 20 origin-neutral questions (500 responses), domestic entities take 83.21% and foreign entities 16.79%, with origin on file for 1,769 of 1,769; 5 of the 30 entities in that share are foreign. Origin is assigned by brand home this edition and by a different rule in July, so no origin share here is compared with July’s.
  • The assistants agree on Carmen, not on how long an answer is. Carmen is in every model’s top three, and it is the most-named entity for Gemini 3.6 Flash (47.50%) and Perplexity Sonar (37.50%). Trendyol is the most-named for ChatGPT 5.6 Luna Pro (35.68%) and Grok 4.3 (87.50%), and The ZKS for Claude Haiku 4.5 (27.00%). Gemini 3.6 Flash names 12.38 entities per answer and Grok 4.3 11.40, against 4.44 for Claude Haiku 4.5.
  • Eleven qualifying ranks lean on one model. Each draws at least 60% of its general-cut mentions from a single model. Gemini 3.6 Flash, which joined the panel this edition, supplies 85 of İpekyol’s 109 mentions and 57 of Oleg Cassini’s 72. ChatGPT 5.6 Luna Pro, also new, supplies all 50 of Berneva’s. Grok 4.3 supplies 111 of Hepsiburada’s 178.
  • Grounding was near-complete for four models and partial for one. 83.58% of all 999 responses were web-grounded. Claude Haiku 4.5, Perplexity Sonar and Grok 4.3 grounded 100.00% of their 200 answers and ChatGPT 5.6 Luna Pro 97.99% of its 199. Gemini 3.6 Flash grounded 20.00%: 40 of its answers report a search, and the other 160 returned no grounding metadata at all — which is not evidence that it answered from memory.
  • Against July, The ZKS rose most and Iconas Line fell most — on a panel that changed. On the same 40 questions, with July’s answers counted the way this edition counts and with Gemini 2.5 Flash-Lite and GPT-4o-mini replaced by Gemini 3.6 Flash and ChatGPT 5.6 Luna Pro, The ZKS moved from 6.82% of 997 responses (68) to 25.83% of 999 (258), +19.01 points, and Iconas Line from 14.64% (146) to 6.11% (61), -8.54. The ZKS is in the top ten of all three models that answered in both editions (Chapter 8).
  • trendyol.com is the most widely cited domain. It is cited in 84 of 200 question-model cells (42.00%), ahead of carmen.com.tr (69 cells, 34.50%) and thezks.com (59, 29.50%). Claude Haiku 4.5 returned exactly 5 citations on every one of its answers, so its domain counts are capped and do not compare with the other models’.

Why it matters. Shoppers increasingly ask AI assistants which evening dress to buy and where to buy it, and whether a brand appears in those answers is becoming a discovery channel in its own right. July’s edition set the baseline; this one repeats its 40 questions word for word, which makes a first reading of movement possible — with the confounds stated beside it, because two of the five assistants answering are not the ones that answered in July. The clearest finding is a structure rather than a movement: a marketplace leads, a group of domestic occasion-wear specialists sits behind it, and the shelf stays mostly domestic whether or not the question asks for Turkish brands. A reminder: the numbers below describe visibility and information availability — not which brands are best.


2. Methodology

One-sentence takeaway: Five models answered 40 Turkish questions up to five times each, giving 999 responses across 200 question-model cells; 393 entities were scored within each cut, and since July two models, the counting rule, the origin rule and the brand-only view have changed.

2.1 Scope and models

The study queried five large language models through their APIs with identical Turkish-language questions. Each of the 40 questions was put to each model up to five times, giving 200 question-model cells and 999 responses analyzed, all collected on 2 October 2026. Four models returned 200 responses each; ChatGPT 5.6 Luna Pro returned 199. The questions are in Turkish and models infer the market from them. The deployment’s system prompt and locale parameter were not recorded with this run, so this report makes no statement about either.

Model API name Responses In July's panel
Claude Haiku 4.5 anthropic/claude-haiku-4.5 200 Yes
Gemini 3.6 Flash gemini-3.6-flash 200 No — joined this edition
ChatGPT 5.6 Luna Pro openai/gpt-5.6-luna-pro 199 No — joined this edition
Perplexity Sonar perplexity/sonar 200 Yes
Grok 4.3 x-ai/grok-4.3 200 Yes
The five-model panel. Gemini 2.5 Flash-Lite and GPT-4o-mini, both in July's panel, were not run this edition.

2.2 Web-search configuration

Every model was asked to search the web. Whether it did is a separate question, and the table below keeps the two apart: the first column is what the deployment asked of each model; the rest is what each model returned.

Model Asked to search Answers Web-grounded Citations Answer modes
Claude Haiku 4.5 Yes 200 100.00% 1,000 searched 200
Gemini 3.6 Flash Yes 200 20.00% 388 searched 40, unknown 160
ChatGPT 5.6 Luna Pro Yes 199 97.99% 937 searched 195, unknown 5
Perplexity Sonar Yes 200 100.00% 5,930 searched 200
Grok 4.3 Yes 200 100.00% 1,051 searched 200
Answer modes: 'searched' means the model reports searching; 'unknown' means it was asked to search and returned no grounding metadata at all. ChatGPT 5.6 Luna Pro's answer modes include one empty response that was excluded from analysis.

Overall, 83.58% of the 999 responses were web-grounded. Claude Haiku 4.5, Perplexity Sonar and Grok 4.3 grounded 100.00% of their answers. ChatGPT 5.6 Luna Pro grounded 97.99%: 195 of its 199 analysed answers report a search and the other 4 carry no grounding metadata. Gemini 3.6 Flash grounded 20.00%: 40 answers report a search, and the other 160 carry no grounding metadata of any kind. An “unknown” answer is not a memory answer. The model was asked to search, and it is not known whether it did; the answer may rest on retrieval that was not reported, on the model’s own knowledge, or on both. This report never treats those answers as answers from memory.

July’s edition included a chapter on search versus memory, explaining why it did not split the two. This edition does not repeat it. Every model was asked to search, so there is no memory segment by design, and nearly all of the ungrounded answers are one model’s, of unknown provenance. Splitting the board by grounding would compare Gemini 3.6 Flash with the other four models, not search with memory.

A citation cap, not a measurement. Claude Haiku 4.5 returned exactly 5 citations on every one of its 200 answers. Its domain counts are truncated at that ceiling and do not compare with an uncapped model’s; Chapter 9 reads them that way.

2.3 Questions and cuts

All 40 questions are the ones asked in July, carried verbatim: none was added, none retired and none reworded. Each is typed by what it asks — discovery (a general “which brand?” or “where to buy?”), attribute (a specific property, such as a fabric, a plus-size fit or a dress that hides the tummy and hips) or use_case (an occasion or need, such as a henna night, a graduation or a wedding guest’s outfit) — and scoped by whether it constrains origin. A constrained question asks for Turkish brands (“iyi bir yerli abiye markası var mı” — is there a good Turkish evening-dress brand?); a neutral one does not (“en iyi abiye markaları” — the best evening-dress brands).

Question type Origin-neutral Origin-constrained Total
Discovery 7 7 14
Attribute 7 6 13
Use case 6 7 13
Question design balance: 14 discovery, 13 attribute and 13 use-case questions; 20 origin-neutral and 20 origin-constrained in all.

Every metric is computed separately within each of six cuts:

Cut Questions Models Responses Qualified Distinct entities
All questions 40 5 999 38 393
Origin-constrained questions 20 5 499 39 285
Origin-neutral questions 20 5 500 34 315
Discovery questions 14 5 350 53 291
Attribute questions 13 5 325 34 220
Use-case questions 13 5 324 32 226
The six cuts. Origin-constrained (20 questions, 499 responses) plus origin-neutral (20, 500) equals all questions (40, 999); the three question-type cuts also sum to all questions.

July called the first three cuts Main, Local and Open. The all-questions cut is referred to in this report as the general cut.

July’s questions carried no topical subcategories, and July read the three question types instead; this edition reads them in Chapter 6, each with its own leaderboard. Under this series’ rules a subcategory is published as a cut only when the study declares it before collection. This study did not, so no subcategory figure appears in this report. Chapter 7 returns to this.

2.4 Entity resolution

What counts as a mention. A second language model reads each answer and lists each brand or company it names, once, in order of first appearance, up to 25 per answer. Product categories, ingredients and cities are excluded. A name that appears only inside a web address is not listed, and a shop named only as where to buy is often left out. Position (MRR) is the mean of 1 ÷ the entity’s first-appearance position, over the answers that name it.

Every listed name is then matched against Herm.io’s brand registry. A name that matches a registered brand or one of its aliases is counted directly and never becomes a candidate. Names that do not match are reviewed as candidates: 401 this period. Of those, 359 were resolved to a brand, 18 were given a verdict by a language model, and 24 were answered by a stored verdict from earlier work. These figures count the work, not the field, and do not reconcile to the 393 entities tracked.

Twenty-four candidates were rejected as not a brand. One of them was a single item naming more than one tracked brand; it was split, and each brand in it was credited. The rules behind the rest are easiest to see in examples.

A brand written only as a web address is not counted as that brand. “Carmen.com.tr” appears as a name in 11 responses. Written that way, it is not counted as Carmen. Most of the frequent rejections this edition are names of this kind — a shop or brand given as its web address rather than by name.

Some short names mean a specific brand in this study. Two short spellings are redirected before any verdict is asked for, because in this market they mean a different brand than the global registry would assume: “nedo” is counted as Nedret Taciroğlu, and “rue” as RUE Les Créateurs. The mentions move to those brands; the entity count does not change. No name is suppressed in this edition.

Resemblance is not identity. Among entities named in at least three answers, 12 near-name pairs were flagged — Trendyolmilla and Trendyol, Vakkorama and Vakko, and Beymen Club and Beymen among them. No pair shares a website on file, and each is tracked as two entities. All 12 are listed in Chapter 11.

The result is 393 entities: 135 brands, 193 brands that are also retailers, 37 retailers, 13 marketplaces, 6 publishers, 5 sub-brands and 2 licensed brands; 2 have no entity type on file. Origin is on file as domestic for 292 and foreign for 61; 40 have no origin on file. The two untyped entities, Didem Soydan and Armut.com, are also the two provisional ones: research never resolved them, and they may duplicate a brand under another spelling. No entity has a product category on file, so scope could not be resolved for any of them, and all 393 are eligible to rank.

2.5 AI Visibility Score (0–100)

Score = 0.45 × Mention + 0.3 × Position + 0.25 × Breadth

Mention is the share of the cut’s responses that name the entity. Position is the entity’s MRR, as defined above — first is 1, second 0.5, and so on. Breadth is the number of models, of five, that name it. Each component is scaled to the highest value among qualified entities in that cut and is never clipped. An entity qualifies for ranking on a cut if it is named in at least 5% of that cut’s responses; there is no absolute mention floor.

Worked example — Trendyol, general cut. Trendyol is named in 424 of 999 responses, a mention rate of 42.44%. That is the highest mention rate among qualified entities in the cut, so its mention component is 100.00. Its MRR of 0.4766 is also the cut maximum, so its position component is 100.00. It is named by 5 of 5 models, so breadth is 100.00. The composite is 0.45 × 100.00 + 0.3 × 100.00 + 0.25 × 100.00 = 100.00. The components are computed from unrounded inputs (a mention rate of 0.424424424 and an MRR of 0.476619658, each set against itself); the four-decimal MRR printed on this page is for reading, not for the arithmetic.

Trendyol scores 100.00 because it holds the cut’s highest value on all three components, so every other score on the general board is read against a marketplace. Because the maxima come from qualified entities only, an unqualified entity can exceed 100 on a component: 88 of 1,730 entity-cut rows do (5.10%), all of them unqualified — 5.90% of the 1,500 unqualified rows. A value above 100 says the entity beats the cut leader on that one component while being named too rarely to rank.

A score belongs to one cut’s scale. The scale moves whenever the leader moves, so an AI Visibility Score is never compared across cuts or verticals. Across editions it is compared in one place only: the general board in Chapter 3 shows each entity’s rank and score change against July’s answers re-scored the way this edition scores (section 2.6). Chapter 8 compares raw mention rate. The maxima behind every table are published in the report’s data file.

2.6 What changed since July, and what is comparable

Edition 1 — July 2026 Edition 2 — October 2026
Questions 40 The same 40, carried verbatim; none added or retired
Responses analyzed 997 999
Models carried Claude Haiku 4.5, Perplexity Sonar, Grok 4.3 Claude Haiku 4.5, Perplexity Sonar, Grok 4.3
Models replaced Gemini 2.5 Flash-Lite, GPT-4o-mini Gemini 3.6 Flash, ChatGPT 5.6 Luna Pro
What counts as a mention A validated brand name appearing in the answer text A brand or company named in the answer, as listed by a second language model; not one that appears only in a web address
Origin rule Founded and headquartered in Turkey Brand home: the country the brand itself is operated from
Brand-only view removes Marketplaces and retailers Marketplaces only; retailers stay
Comparable across editions — Raw mention rate on the general cut, and the general board's rank and score change against July re-scored the same way — each with the panel change stated beside it. Not comparable: origin shares.
What changed between the two editions. The question set is identical; the panel, the counting rule, the origin rule and the brand-only view are not.

Four changes matter for any comparison with July. The panel: three models are carried and two were replaced, and nothing in this report corrects for that. The counting rule: July’s page counted a validated brand name wherever it appeared in the answer text; this edition counts a brand or company only where a second language model lists it as named in the answer, and not where it appears only in a web address. The two count differently, and the brand registry has also changed since July, so a July level in this report can differ from July’s page for any entity. So that both sides of every comparison count a brand the same way, July’s levels in this report are July’s answers, counted the way this edition counts — not the figures July’s page printed. The origin rule: see 2.7. The brand-only view: July’s removed marketplaces and retailers; this edition’s removes marketplaces only, so retailers such as Sefamerve and Beymen stay in it (section 3.3).

Change on the leaderboard. The general board in Chapter 3 shows, beside each entity, how many places it moved and how much its score changed since July. Both are measured against July’s board rebuilt the way this edition builds its own — July’s answers and models, this edition’s brand registry and scoring — not against the figures July’s page printed, which counted differently. Besides the entity itself, two things move them, and nothing adjusts for either: the two models replaced between editions, and, for the score only, the leader. A score is a share of its cut’s leader, so when the leader is named more or less often every other score moves with it; rank change does not have that problem. 15 entities on this board were not on July’s, and 8 that were on July’s are not on this one.

2.7 Origin, neutrality and self-exclusion

Origin is the brand home: the country the brand itself is operated from, which is not always its owner’s. Fendi Casa is Italian though LVMH is French; Cetaphil is American though Galderma is Swiss. Where no separate brand home can be established, the parent’s headquarters stands in. Local-market naming variants are folded into their parent before origin is read, so a brand named after its market subsidiary is not recorded as domestic. Tables show ISO country codes; shares use a domestic/foreign rollup. Entities with no origin on file stay in every table, are excluded from every origin share, and are counted — never defaulted.

July used a different rule. July’s edition counted a brand as Turkish only if it was founded and headquartered in Turkey; this edition assigns every origin by brand home. An entity can therefore be coded differently in the two editions because the rule and the research behind it changed, not because the market did. Within this edition the rule is published here beside every origin figure, which is what makes a domestic-versus-foreign comparison safe; across the change it is not, and no origin share in this report is compared with July’s. July also published a Turkish-versus-foreign split on its origin-neutral questions, under its own rule and on its own basis. It is not set beside the shares in this report.

Marketplaces and retailers rank in every leaderboard, because they are what the assistants actually name. The brand-only view in section 3.3 removes the three qualifying marketplaces — Hepsiburada, Modanisa and Trendyol — by name and changes nothing else. Retailers, Sefamerve, Beymen, Vakko, LC Waikiki and DCEY among the qualifiers, are retained everywhere, as are brands that also run their own stores.

Origin shares are read on a brand-only, qualified basis. The headline share is computed over qualified entities with origin on file, with marketplaces and publishers left out — a marketplace is where a brand is bought and a publisher is where it is written about — and own-label retailers kept in. It is always printed with its coverage: the qualifying mentions with origin on file. The same arithmetic over every tracked entity, qualified or not, is published as a sanity check; it describes the long tail, not the shelf a reader sees. This basis applies to shares only; it does not change who ranks.

Neutrality and self-exclusion. This is a market-wide study with no focus brand. Herm.io’s own domain is excluded from the source data before analysis, so the company’s own content neither appears among the cited domains nor influences any figure.


3. Overall Visibility Leaderboard

One-sentence takeaway: Thirty-eight of 393 tracked entities qualify on the general cut; the marketplace Trendyol leads on frequency and position alike, Carmen is second at 93.77, and Mango, Zara and Oleg Cassini are the three foreign qualifiers.

The list below ranks the 38 entities named in at least 5% of all 999 responses, by AI Visibility Score. As a reminder, it ranks visibility, not quality.

# Brand AI Score Δ
1
Trendyol
100.00
+6.20
2
Carmen
93.77
+18.62
3
The ZKS
79.74
+29.72
4
ADL
73.23
+8.18
5
Abiyefon
69.13
+12.22
6
AlfaBeta
68.22
+4.50
7
Koton
67.38
+10.14
8
Sefamerve
66.64
+1.15
9
Tarık Ediz
65.14
+13.91
10
Beymen
61.80
New
11
Modanisa
60.45
+0.55
12
GIZIA
56.61
+4.62
13
Hepsiburada
54.42
+1.07
14
Mango
53.01
New
15
Vakko
52.86
New
16
Roman
51.60
−2.03
17
Kenzel
51.40
−1.75
18
ModaSelvim
50.82
−3.59
19
Setre
50.10
+11.10
20
İpekyol
49.74
New
21
EKOL
47.93
+0.75
22
Zara
47.58
+3.67
23
Lafaba
47.00
−3.18
24
Cengiz Aktürk
46.27
+1.44
25
Nocturne
46.13
New
26
Rengin
44.04
−2.69
27
Armine
44.01
New
28
Oleg Cassini
42.29
New
29
LC Waikiki
42.10
−1.78
30
Twist
41.98
New
31
Setrms
40.36
New
32
By Saygı
40.35
New
33
Sagaza
37.81
New
34
Iconas Line
37.31
−14.52
35
Machka
35.10
New
36
Kayra
34.92
New
37
Berneva
28.26
New
38
DCEY
20.92
New
Δ: change in AI Score since the July 2026 edition. ▲/▼: places moved. New: not on July's board. Both are measured against July's answers re-scored the way this edition scores, not against July's published page. Two of the five models changed between editions, and a score also moves when the leader's does, so read Δ as direction more than exact size (section 2.6).

Average score of the 38 qualified entities: 52.64. Thirty-five are domestic and three are foreign.

AI Visibility Score — qualified entities, all 40 questions

3.1 Tier narrative

The top two (100.00, 93.77). Trendyol is named in 42.44% of responses (424 of 999), by all five models, and earliest, with an MRR of 0.4766: it holds the cut’s highest value on every component. Carmen is named in 38.44% (384 of 999), also by all five models, with an MRR of 0.4452.

The ZKS to AlfaBeta (79.74–68.22). Position separates these four more than frequency does. ADL is named in 26.03% of responses (260 of 999) and The ZKS in 25.83% (258 of 999), but The ZKS ranks third on an MRR of 0.4347 against ADL’s 0.3279. AlfaBeta is named more often than Abiyefon, in 24.22% (242) against 23.32% (233), and ranks one place lower, with an MRR of 0.2786 against 0.3083.

Koton to Beymen (67.38–61.80). Koton, Sefamerve and Tarık Ediz follow. Tarık Ediz ranks ninth on 15.62% (156 of 999), carried by an MRR of 0.3747. Beymen, a retailer, ranks tenth on 7.61% (76 of 999) — by far the lowest mention rate in the top ten — because its MRR of 0.4565 is second only to Trendyol’s: when it is named, it is named early.

The middle of the board (60.45–46.13). Two marketplaces open this band. Modanisa (#11) and Hepsiburada (#13) are named often — 18.82% (188 of 999) and 17.82% (178 of 999) — but late, with MRRs of 0.2461 and 0.2467, and Hepsiburada by 4 of 5 models. Mango, Spanish, is 14th at 53.01, the highest-placed foreign entity, on 6.51% (65 of 999) with an MRR of 0.3354. Zara, also Spanish, is 22nd at 47.58 on 8.31% (83 of 999). Scores in this band sit close together, and differences of a point or less carry no meaning.

The tail of the board (44.04–20.92). Rengin to DCEY. Oleg Cassini, American, is 28th at 42.29 on 7.21% (72 of 999), named by 3 of 5 models. Breadth thins toward the bottom: Sagaza (33rd, 6.61%) is named by 2 models, Berneva (37th, 5.01%) by 1, and DCEY (38th, 5.31%) by 2, with the board’s lowest MRR, 0.0841. By Saygı and Berneva each qualify on 50 of 999 responses (5.01%), the minimum on this board. Chapter 5 shows which of these ranks rest mainly on one model. The remaining tracked entities sit below the threshold, and the appendix lists them all.

3.2 The same board by mention rate

Scores reward prominence as well as frequency, so the ranking by raw mention rate is a different list, and the difference is informative.

# Entity Mention rate Mentions Score Rank by score Δ (+ = higher on score)
1 Trendyol 42.44% 424 100.00 1 +0
2 Carmen 38.44% 384 93.77 2 +0
3 ADL 26.03% 260 73.23 4 -1
4 The ZKS 25.83% 258 79.74 3 +1
5 AlfaBeta 24.22% 242 68.22 6 -1
6 Abiyefon 23.32% 233 69.13 5 +1
7 Sefamerve 21.32% 213 66.64 8 -1
8 Koton 19.62% 196 67.38 7 +1
9 Modanisa 18.82% 188 60.45 11 -2
10 Hepsiburada 17.82% 178 54.42 13 -3
11 Tarık Ediz 15.62% 156 65.14 9 +2
12 GIZIA 13.91% 139 56.61 12 +0
Top 12 of the general cut by raw mention rate (denominator 999 responses), with each entity's rank on the score board.

Trendyol and Carmen are first and second either way. Below them, three pairs swap places — ADL and The ZKS, AlfaBeta and Abiyefon, Sefamerve and Koton — each time in favor of the entity named earlier. The marketplaces lose most to position: Modanisa is 9th by volume and 11th by score, Hepsiburada 10th and 13th. Tarık Ediz goes the other way, 11th by volume and 9th by score. GIZIA is 12th on both.

3.3 Brand-only view

Removing the three qualifying marketplaces — Hepsiburada, Modanisa and Trendyol, named explicitly, with nothing else changed — leaves 35 entities. Scores are unchanged, because the scale is still the general cut’s, which Trendyol set; entities ranked below a removed marketplace move up, and Carmen heads this board. Sefamerve, Beymen, Vakko, LC Waikiki and DCEY stay: they are retailers, not marketplaces. July’s brand-only view also removed retailers, so the two views are not the same list (section 2.6).

# Entity Origin Type Score Mention rate
1 Carmen TR Brand & retailer 93.77 38.44%
2 The ZKS TR Brand 79.74 25.83%
3 ADL TR Brand 73.23 26.03%
4 Abiyefon TR Brand & retailer 69.13 23.32%
5 AlfaBeta TR Brand & retailer 68.22 24.22%
6 Koton TR Brand 67.38 19.62%
7 Sefamerve TR Retailer 66.64 21.32%
8 Tarık Ediz TR Brand & retailer 65.14 15.62%
9 Beymen TR Retailer 61.80 7.61%
10 GIZIA TR Brand 56.61 13.91%
11 Mango ES Brand 53.01 6.51%
12 Vakko TR Retailer 52.86 10.71%
13 Roman TR Brand 51.60 10.41%
14 Kenzel TR Brand 51.40 12.01%
15 ModaSelvim TR Brand & retailer 50.82 6.31%
16 Setre TR Brand & retailer 50.10 9.61%
17 İpekyol TR Brand 49.74 10.91%
18 EKOL TR Brand & retailer 47.93 6.51%
19 Zara ES Brand 47.58 8.31%
20 Lafaba TR Brand 47.00 7.71%
21 Cengiz Aktürk TR Brand & retailer 46.27 11.11%
22 Nocturne TR Brand 46.13 10.91%
23 Rengin TR Brand 44.04 9.41%
24 Armine TR Brand 44.01 7.31%
25 Oleg Cassini US Brand 42.29 7.21%
26 LC Waikiki TR Retailer 42.10 6.71%
27 Twist TR Brand 41.98 6.11%
28 Setrms TR Brand & retailer 40.36 7.41%
29 By Saygı TR Brand & retailer 40.35 5.01%
30 Sagaza TR Brand & retailer 37.81 6.61%
31 Iconas Line TR Brand 37.31 6.11%
32 Machka TR Brand 35.10 7.21%
33 Kayra TR Brand 34.92 7.41%
34 Berneva TR Brand 28.26 5.01%
35 DCEY TR Retailer 20.92 5.31%
Brand-only board, general cut: entity type Marketplace removed, nothing else changed. Scores remain on the general cut's scale. Denominator 999 responses.

4. Two Shelves: Origin-Constrained and Origin-Neutral Questions

One-sentence takeaway: On the 20 origin-constrained questions domestic entities take 100.00% of qualifying mentions and foreign entities 0.00%, and on the 20 origin-neutral questions they take 83.21% and 16.79% — with origin on file for all 2,485 and all 1,769 qualifying mentions respectively.

The 40 questions divide into 20 that ask for Turkish brands and 20 that do not. Each set has its own leaderboard, its own denominator and its own scale. Neither is the market on its own, so the two figures are always quoted together, each with the number of questions behind it.

4.1 Origin share on both cuts

Origin-constrained Origin-neutral All questions (blended — not an origin cut)
Questions 20 20 40
Responses 499 500 999
Qualified entities in the share (domestic / foreign / unknown) 36 (36 / 0 / 0) 30 (25 / 5 / 0) 35 (32 / 3 / 0)
Qualified entities left out (marketplaces and publishers), with mentions Trendyol (176), Modanisa (93), Hepsiburada (85) Trendyol (248), Modanisa (95), Hepsiburada (93), Akakçe (25) Trendyol (424), Modanisa (188), Hepsiburada (178)
Domestic share of qualifying mentions 100.00% 83.21% 94.68%
Foreign share of qualifying mentions 0.00% 16.79% 5.32%
Qualifying mentions with origin on file 2,485 of 2,485 1,769 of 1,769 4,133 of 4,133
Sanity check — domestic share, all tracked entities 99.64% 83.28% 92.59%
Sanity check — foreign share, all tracked entities 0.36% 16.72% 7.41%
All-entity mentions with origin on file 3,842 of 3,973 2,906 of 2,983 6,748 of 6,956
Headline basis: qualified entities with origin on file; marketplaces and publishers left out, own-label retailers kept in. Domestic and foreign are each computed, not derived from the other. The all-questions column blends both question scopes and is shown for reconciliation only.

Origin-constrained (20 questions, 499 responses): domestic entities take 100.00% of qualifying mentions and foreign entities 0.00%, with origin on file for 2,485 of 2,485 qualifying mentions. All 36 qualified entities in the share are domestic.

Origin-neutral (20 questions, 500 responses): domestic entities take 83.21% and foreign entities 16.79%, with origin on file for 1,769 of 1,769. Of the 30 qualified entities in the share, 25 are domestic and 5 foreign.

The assistants read “yerli” as a constraint and answer it with domestic names only. Without it, foreign entities enter the shelf — Mango, H&M, Zara, Oleg Cassini and Pronovias — but domestic entities still take most of the qualifying mentions, because the four most-named entities on the neutral shelf that are not marketplaces — Carmen, Abiyefon, ADL and The ZKS — are domestic.

The sanity check points the same way. Across every tracked entity, qualified or not, domestic entities take 99.64% and foreign entities 0.36% of mentions on the constrained questions (origin on file for 3,842 of 3,973), and 83.28% and 16.72% on the neutral questions (2,906 of 2,983). On the constrained questions the long tail carries a sliver of foreign names that the shelf does not; on the neutral questions the two bases are close.

Left out of every share, and still ranked. On the constrained cut, three qualifying marketplaces are left out of the share: Trendyol (176 mentions), Modanisa (93) and Hepsiburada (85). On the neutral cut, four entities are: the same three marketplaces, Trendyol (248), Modanisa (95) and Hepsiburada (93), and the publisher Akakçe (25). Across all tracked entities, 14 marketplaces and publishers with 416 mentions are left out on the constrained cut, and 17 with 559 on the neutral cut. Every qualifying entity left out is domestic, so including them would raise the domestic share for a reason that has nothing to do with brands. They still rank in every leaderboard.

4.2 The origin-constrained shelf

# Brand AI Score
1
Carmen
93.20
2
The ZKS
75.96
3
Trendyol
74.85
4
AlfaBeta
72.93
5
ADL
70.34
6
Sefamerve
68.23
7
Abiyefon
61.49
8
Tarık Ediz
61.31
9
Koton
56.61
10
GIZIA
55.97
11
Modanisa
53.36
12
Roman
49.07
13
Kenzel
47.10
14
Cengiz Aktürk
46.69
15
Eyyo Atelier
46.29
16
İpekyol
46.13
17
Raisa Vanessa
45.38
18
ModaSelvim
45.31
19
Rengin
44.26
20
Eser Giyim
44.18
21
Hepsiburada
44.08
22
Armine
43.62
23
EKOL
43.56
24
Lafaba
43.15
25
Setrms
41.31
26
Nocturne
38.57
27
By Saygı
38.42
28
Trendyolmilla
38.03
29
Sagaza
36.86
30
Vakko
36.43
31
Lacrima
35.84
32
Setre
35.07
33
Kayra
30.97
34
Nihan
30.07
35
Alvina
29.58
36
Berneva
27.41
37
Machka
26.94
38
Iconas Line
26.33
39
Exquise
19.72

39 entities qualify on the 20 origin-constrained questions (499 responses). Thirty-six enter the origin share, all domestic, and the three marketplaces are left out of it.

Carmen leads at 93.20, named in 46.89% of these responses (234 of 499), the highest mention rate on the cut. It scores below 100 because the cut’s highest MRR is not Carmen’s but Raisa Vanessa’s, 0.6311. The ZKS is second at 75.96 on 31.06% (155 of 499). Trendyol is third at 74.85: named more often than The ZKS, in 35.27% (176 of 499), but later, with an MRR of 0.3366 against 0.4449. AlfaBeta (72.93, 34.27%) and ADL (70.34, 30.06%) complete the top five.

This shelf brings smaller domestic names forward on position. Raisa Vanessa ranks 17th on 5.61% (28 of 499), named by 2 of 5 models, with the cut’s highest MRR. Eyyo Atelier ranks 15th on 6.81% (34 of 499) with an MRR of 0.3102, named by all five models, and Eser Giyim 20th on 7.01% (35 of 499). Kenzel and Cengiz Aktürk are named equally often, in 16.63% (83 of 499) each, and rank 13th and 14th. Near the foot of the board, Berneva, Machka, Iconas Line and Exquise each rest on one or two models (Chapter 11). Ranks built mainly on position, or on a single model, should be read cautiously.

No foreign entity qualifies on this cut. Foreign entities are named on the constrained questions — they take 0.36% of all-entity mentions with origin on file — but none often enough to reach the 5% threshold.

4.3 The origin-neutral shelf

# Entity Origin Type Score Mention rate Mentions / responses Breadth
1 Trendyol TR Marketplace 100.00 49.60% 248 / 500 5 of 5
2 Carmen TR Brand & retailer 71.91 30.00% 150 / 500 5 of 5
3 The ZKS TR Brand 65.53 20.60% 103 / 500 5 of 5
4 Beymen TR Retailer 63.32 13.60% 68 / 500 5 of 5
5 Abiyefon TR Brand & retailer 63.25 22.00% 110 / 500 5 of 5
6 ADL TR Brand 60.67 22.00% 110 / 500 5 of 5
7 Koton TR Brand 59.03 19.60% 98 / 500 5 of 5
8 Modanisa TR Marketplace 56.37 19.00% 95 / 500 5 of 5
9 Tarık Ediz TR Brand & retailer 55.07 12.40% 62 / 500 5 of 5
10 Mango ES Brand 54.26 13.00% 65 / 500 5 of 5
11 Hepsiburada TR Marketplace 53.72 18.60% 93 / 500 4 of 5
12 Vakko TR Retailer 53.18 14.00% 70 / 500 5 of 5
13 H&M SE Brand 53.10 8.80% 44 / 500 5 of 5
14 Zara ES Brand 51.31 16.40% 82 / 500 5 of 5
15 Sefamerve TR Retailer 51.06 13.40% 67 / 500 5 of 5
16 AlfaBeta TR Brand & retailer 47.63 14.20% 71 / 500 5 of 5
17 GIZIA TR Brand 47.31 8.60% 43 / 500 5 of 5
18 Roman TR Brand 44.61 9.40% 47 / 500 5 of 5
19 Oleg Cassini US Brand 44.17 13.60% 68 / 500 3 of 5
20 Pronovias ES Brand & retailer 43.38 7.60% 38 / 500 5 of 5
21 Lafaba TR Brand 43.09 7.80% 39 / 500 5 of 5
22 EKOL TR Brand & retailer 41.97 5.00% 25 / 500 4 of 5
23 Setre TR Brand & retailer 41.57 9.40% 47 / 500 4 of 5
24 Kenzel TR Brand 41.00 7.40% 37 / 500 5 of 5
25 Twist TR Brand 39.89 7.80% 39 / 500 5 of 5
26 Nocturne TR Brand 39.54 9.80% 49 / 500 4 of 5
27 LC Waikiki TR Retailer 39.14 9.00% 45 / 500 4 of 5
28 İpekyol TR Brand 38.18 8.80% 44 / 500 3 of 5
29 Cengiz Aktürk TR Brand & retailer 37.77 5.60% 28 / 500 4 of 5
30 Rengin TR Brand 33.69 5.40% 27 / 500 4 of 5
31 Machka TR Brand 31.20 6.80% 34 / 500 3 of 5
32 Akakçe TR Publisher 29.13 5.00% 25 / 500 3 of 5
33 Network TR Brand 27.68 5.00% 25 / 500 3 of 5
34 DCEY TR Retailer 21.18 6.80% 34 / 500 2 of 5
Origin-neutral cut: all 34 qualifiers. Denominator 500 responses across 20 questions.

Trendyol leads this shelf at 100.00, named in 49.60% of these responses (248 of 500) with an MRR of 0.5760. Carmen (71.91, 30.00%) and The ZKS (65.53, 20.60%) follow. Beymen is fourth at 63.32 on 13.60% (68 of 500), with an MRR of 0.4989; Abiyefon and ADL, fifth and sixth, are each named in 22.00% (110 of 500) and rank below it.

Five foreign entities qualify. Mango is 10th at 54.26 on 13.00% (65 of 500). H&M, Swedish, is 13th at 53.10 on 8.80% (44 of 500), with an MRR of 0.3861. Zara is 14th at 51.31, named more often than either, in 16.40% (82 of 500), but later (MRR 0.2195). Oleg Cassini is 19th at 44.17 on 13.60% (68 of 500), named by 3 of 5 models, and Pronovias, Spanish, 20th at 43.38 on 7.60% (38 of 500). Oleg Cassini’s visibility leans on one model: on the general cut, 57 of its 72 mentions (79.17%) come from Gemini 3.6 Flash, which joined the panel this edition (Chapter 5).

Modanisa (8th, 19.00%) and Hepsiburada (11th, 18.60%) are the other marketplaces on this shelf. Akakçe, a publisher, qualifies here too (32nd); like the marketplaces, it ranks and is left out of the share.

4.4 What this chapter does not compare

No origin share here is compared with July’s, because July assigned origin by a different rule (section 2.7) and published its origin-neutral split on its own basis. The constrained figure is never read on its own: 100.00% domestic and 0.00% foreign on 20 questions that ask for Turkish brands describes the question as much as the market, and it stands beside 83.21% domestic and 16.79% foreign on the 20 that do not. The all-questions column in the table above blends the two scopes and is not an origin measurement. And the question-type cuts in Chapter 6 each blend constrained and neutral questions, so their counts of domestic qualifiers are not origin measurements either.


5. Differences Between Models

One-sentence takeaway: Carmen is in every model’s top three, but the models differ sharply in length — Gemini 3.6 Flash names 12.38 entities per answer and Claude Haiku 4.5 4.44 — and 11 qualifying ranks draw at least 60% of their general-cut mentions from a single model.

5.1 Per-model behavior

Model Responses Web-grounded Citations Distinct entities Entities per answer Most-named entity Its rate In July's panel
Claude Haiku 4.5 200 100.00% 1,000 148 4.44 The ZKS (TR) 27.00% Yes
Gemini 3.6 Flash 200 20.00% 388 170 12.38 Carmen (TR) 47.50% No
ChatGPT 5.6 Luna Pro 199 97.99% 937 159 4.91 Trendyol (TR) 35.68% No
Perplexity Sonar 200 100.00% 5,930 160 6.54 Carmen (TR) 37.50% Yes
Grok 4.3 200 100.00% 1,051 171 11.40 Trendyol (TR) 87.50% Yes
Four models answered all 40 questions five times, 200 responses each; ChatGPT 5.6 Luna Pro returned 199. 'Its rate' is the share of that model's responses naming its most-named entity.

The leaders are shared; their order is not. Trendyol is the most-named entity for Grok 4.3 (87.50% of its 200 answers) and ChatGPT 5.6 Luna Pro (35.68% of its 199). Carmen is the most-named for Gemini 3.6 Flash (47.50%) and Perplexity Sonar (37.50%), and The ZKS for Claude Haiku 4.5 (27.00%). Carmen is in the top three of all five models.

Answer length is the largest difference. Gemini 3.6 Flash names 12.38 entities per answer and Grok 4.3 11.40, and they reach the widest repertoires, 170 and 171 distinct entities. Perplexity Sonar names 6.54 per answer (160 distinct). ChatGPT 5.6 Luna Pro (4.91, 159 distinct) and Claude Haiku 4.5 (4.44, 148 distinct) give short, selective answers. The two models that joined this edition sit at opposite ends: Gemini 3.6 Flash writes the longest lists on the panel, ChatGPT 5.6 Luna Pro some of the shortest. For an entity at the edge of visibility, being named at all depends heavily on which assistant answers.

Entities named per answer, by model

5.2 Each model’s most-named entities

# Claude Haiku 4.5 Gemini 3.6 Flash ChatGPT 5.6 Luna Pro Perplexity Sonar Grok 4.3
1 The ZKS 27.00% Carmen 47.50% Trendyol 35.68% Carmen 37.50% Trendyol 87.50%
2 Carmen 19.00% İpekyol 42.50% The ZKS 31.16% Trendyol 35.50% Carmen 59.50%
3 Abiyefon 14.50% Trendyol 40.00% Carmen 28.64% ADL 28.50% Hepsiburada 55.50%
4 Trendyol 13.50% AlfaBeta 38.00% Berneva 25.13% Koton 27.50% The ZKS 52.50%
5 ADL 11.50% Tarık Ediz 36.50% ADL 15.58% Abiyefon 23.00% Sefamerve 51.00%
6 AlfaBeta 11.50% Modanisa 34.00% Tuba Ergin 12.56% AlfaBeta 20.50% Abiyefon 42.00%
7 Hepsiburada 9.00% ADL 33.00% Modanisa 12.06% LC Waikiki 19.50% ADL 41.50%
8 Kenzel 8.50% Nocturne 32.50% Coupe Official 10.55% Sefamerve 19.50% AlfaBeta 40.50%
9 Koton 7.50% Machka 31.00% Koton 10.55% The ZKS 16.00% Modanisa 38.50%
10 GIZIA 7.50% Setrms 30.00% AlfaBeta 10.55% Hepsiburada 15.00% Koton 33.50%
Top ten entities per model. Rate = share of that model's responses naming the entity: 200 per model, 199 for ChatGPT 5.6 Luna Pro.

The models agree on a core: Trendyol, Carmen, ADL and AlfaBeta are in every model’s top ten. They differ beneath it. Grok 4.3 has three marketplaces in its top ten — Trendyol first (87.50%), Hepsiburada third (55.50%) and Modanisa ninth (38.50%) — and the retailer Sefamerve fifth (51.00%). Gemini 3.6 Flash has İpekyol second (42.50%), Tarık Ediz fifth (36.50%), and Nocturne, Machka and Setrms eighth to tenth (32.50%, 31.00% and 30.00%); none of the five is in any other model’s top ten, and The ZKS is not in Gemini 3.6 Flash’s. ChatGPT 5.6 Luna Pro has Berneva fourth (25.13%), Tuba Ergin sixth (12.56%) and Coupe Official eighth (10.55%), none of them in another model’s top ten. Claude Haiku 4.5 is the most compact: its eighth entity, Kenzel, is named in 8.50% of its answers and its tenth, GIZIA, in 7.50%. Perplexity Sonar is the only model with LC Waikiki in its top ten (seventh, 19.50%).

5.3 One model’s habit

Eleven qualifying entities on the general cut draw at least three in five of their mentions from a single model:

Entity General rank Mentions, all models From one model That model's share That model's rate Models naming
Berneva 37 50 ChatGPT 5.6 Luna Pro (50) 100.00% 25.13% of 199 1
DCEY 38 53 Gemini 3.6 Flash (50) 94.34% 25.00% of 200 2
Sagaza 33 66 Gemini 3.6 Flash (59) 89.39% 29.50% of 200 2
Iconas Line 34 61 Grok 4.3 (54) 88.52% 27.00% of 200 4
Machka 35 72 Gemini 3.6 Flash (62) 86.11% 31.00% of 200 3
Setrms 31 74 Gemini 3.6 Flash (60) 81.08% 30.00% of 200 4
Kayra 36 74 Gemini 3.6 Flash (59) 79.73% 29.50% of 200 3
Oleg Cassini 28 72 Gemini 3.6 Flash (57) 79.17% 28.50% of 200 3
İpekyol 20 109 Gemini 3.6 Flash (85) 77.98% 42.50% of 200 4
Setre 19 96 Grok 4.3 (60) 62.50% 30.00% of 200 5
Hepsiburada 13 178 Grok 4.3 (111) 62.36% 55.50% of 200 4
Qualified entities whose general-cut mentions come 60% or more from one model.

Berneva is ChatGPT 5.6 Luna Pro’s

Berneva qualifies on the general cut — 37th, on 5.01% of responses (50 of 999) — on mentions from one model only. All 50 come from ChatGPT 5.6 Luna Pro, which names it in 25.13% of its 199 answers and ranks it fourth; no other model names it at all. ChatGPT 5.6 Luna Pro joined the panel this edition. Berneva’s rank describes one assistant’s answers, not a consensus across the panel.

DCEY, Sagaza, Machka, Setrms, Kayra, Oleg Cassini and İpekyol lean on Gemini 3.6 Flash

Gemini 3.6 Flash supplies 50 of DCEY’s 53 general-cut mentions (94.34%), 59 of Sagaza’s 66 (89.39%), 62 of Machka’s 72 (86.11%), 60 of Setrms’s 74 (81.08%), 59 of Kayra’s 74 (79.73%), 57 of Oleg Cassini’s 72 (79.17%) and 85 of İpekyol’s 109 (77.98%). Gemini 3.6 Flash also joined this edition, and it writes the longest lists on the panel, 12.38 entities per answer; a long answer gives more names a chance to be counted at all. All of them sit in the lower half of the general board, from İpekyol (20th) to DCEY (38th). Oleg Cassini is one of the three foreign qualifiers on the general cut and one of the five on the origin-neutral cut (Chapter 4), so part of the foreign presence on the shelf is this one model’s.

Iconas Line, Setre and Hepsiburada lean on Grok 4.3

Grok 4.3 supplies 54 of Iconas Line’s 61 general-cut mentions (88.52%), 60 of Setre’s 96 (62.50%) and 111 of Hepsiburada’s 178 (62.36%), naming them in 27.00%, 30.00% and 55.50% of its 200 answers. Grok 4.3 answered in both editions. Hepsiburada, 13th, is a marketplace: its rank describes one assistant’s habit of pointing to where to buy as much as a consensus across the panel.

5.4 What only one model names

Every model names entities no other model names at all: 57 for Gemini 3.6 Flash, 42 for Grok 4.3, 33 for ChatGPT 5.6 Luna Pro, 25 for Claude Haiku 4.5 and 24 for Perplexity Sonar.

Model Entities named by this model alone Examples
ChatGPT 5.6 Luna Pro 33 Berneva, Coupe Official, So Lady, Farika, Miss Diamond, Pionto
Claude Haiku 4.5 25 Didem Soydan, GILLA'S QUEEN, Desigual, Just Cavalli, Karl Lagerfeld, United Colors of Benetton
Gemini 3.6 Flash 57 Tuvanam, Ju Istanbul, Touche Privé, Exquise, Tuğba, Gigii's
Grok 4.3 42 Mayabella, Beyza Giyim, Beyza Online, Dressland, Davetten Davete, Milagron
Perplexity Sonar 24 Ceylan Otantik, Civil, Laranor, Jeyan Gedik, İmajButik, HESNA
Entities named by a single model. They carry a breadth component of one model in five and rarely qualify; Berneva, in ChatGPT 5.6 Luna Pro's list, is the general-cut exception. Claude Haiku 4.5's examples include Didem Soydan, one of the two provisional entities (section 2.4).

6. Question-Type Ownership

One-sentence takeaway: Carmen leads the discovery questions and Trendyol leads the attribute and use-case questions; discovery qualifies 53 entities against 34 for attribute and 32 for use-case questions, and the specialists below the leaders change with the type.

Question type Questions Responses Distinct entities Qualified Domestic qualifiers (blended) Average score Leader Highest mention rate Highest MRR
Discovery 14 350 291 53 48 43.80 Carmen (88.84) 48.29% (Carmen) 0.6988 (Raisa Vanessa)
Attribute 13 325 220 34 33 48.40 Trendyol (90.85) 43.38% (Trendyol) 0.6133 (Faik Sönmez)
Use case 13 324 226 32 31 49.13 Trendyol (100.00) 38.58% (Trendyol) 0.5269 (Trendyol)
Discovery rests on 14 questions and 350 responses, attribute on 13 and 325, use case on 13 and 324. Each mixes origin-constrained and origin-neutral questions (section 2.3), so 'Domestic qualifiers' counts domestic entities over a blended question set and is not an origin measurement.
# Entity Score Rate MRR
1 Carmen 88.84 48.29% 0.4388
2 Trendyol 87.76 45.14% 0.4819
3 ADL 79.67 43.43% 0.3307
4 Tarık Ediz 70.17 30.29% 0.3948
5 AlfaBeta 69.43 37.14% 0.2287
Discovery — top 5 of 53 (350 responses).
# Entity Score Rate MRR
1 Trendyol 90.85 43.38% 0.4262
2 Carmen 85.94 39.69% 0.4041
3 The ZKS 84.65 30.77% 0.5669
4 Abiyefon 77.80 30.77% 0.4268
5 Faik Sönmez 61.38 6.15% 0.6133
Attribute — top 5 of 34 (325 responses).
# Entity Score Rate MRR
1 Trendyol 100.00 38.58% 0.5269
2 Carmen 85.52 26.54% 0.5192
3 Koton 77.85 22.53% 0.4667
4 The ZKS 77.53 22.84% 0.4546
5 AlfaBeta 67.43 14.20% 0.4544
Use case — top 5 of 32 (324 responses).

Discovery — “which brand?” or “where to buy?” — spreads visibility widest: 291 distinct entities, 53 qualifiers and the lowest average score of the three, 43.80. Carmen leads at 88.84 and is named most, in 48.29% of these responses (169 of 350). Trendyol is second at 87.76 on 45.14% (158 of 350), and ADL third at 79.67 on 43.43% (152 of 350). The cut’s highest MRR is Raisa Vanessa’s, 0.6988, on mentions from two models (27 mentions). Berneva, Tuvanam, Iconas Line, Ju Istanbul and Touche Privé each qualify on this cut on mentions from a single model.

Attribute — a specific property — draws 220 distinct entities and 34 qualifiers, with an average score of 48.40. Trendyol leads at 90.85 and is named most, in 43.38% of these responses (141 of 325); Carmen is second at 85.94 on 39.69%. The ZKS and Abiyefon are named equally often, in 30.77% (100 of 325) each, and The ZKS ranks third on an MRR of 0.5669 against Abiyefon’s 0.4268. Faik Sönmez is fifth at 61.38 on a 6.15% mention rate (20 of 325), carried by the cut’s highest MRR, 0.6133, and named by all five models. Oleg Cassini, Kayra, Deniz Butik, Setrms and Alvina each qualify here on two models.

Use case — an occasion or need — draws 226 distinct entities and 32 qualifiers, with an average score of 49.13. Trendyol leads at 100.00, named in 38.58% of these responses (125 of 324) and holding the cut’s highest MRR, 0.5269; Carmen is second at 85.52 on 26.54%. Koton is third at 77.85 on 22.53% (73 of 324), just ahead of The ZKS (77.53), which is named slightly more often, in 22.84%, but later: MRR 0.4546 against Koton’s 0.4667. Nocturne, Sagaza, Tarık Ediz and Iconas Line each qualify here on two models.

The three types rest on 350, 325 and 324 responses. The differences between them are not origin findings: discovery splits 7 origin-neutral and 7 origin-constrained questions, attribute 7 and 6, and use case 6 and 7, and Chapter 4 is where origin is measured.


7. Attribute Axes

One-sentence takeaway: No attribute axis can be published this edition: no entity has a product category or audience on file, and no subcategory was declared before collection, which this series requires before publishing one as a cut.

An attribute chapter ranks entities by something they are — the product categories they make, or the audience they serve — rather than by the questions asked. This edition’s data carries no such attribute: none of the 393 tracked entities has a product category on file, and no audience attribute is recorded. A product-shelf chapter cannot be written from this export, and this report does not substitute an estimate for one.

No question in this study records a subcategory either, and under this series’ rules a subcategory is published as a cut only when the study declares it before collection. This study did not. July’s questions carried no topical subcategories, and July read the three question types instead; that reading continues, with full leaderboards, in Chapter 6.


8. What Changed Since July

One-sentence takeaway: On the same 40 questions, against July’s answers counted the way this edition counts, 74 of the 198 entities named in both editions rose and 111 fell — The ZKS gained 19.01 points and Iconas Line lost 8.54 — with two of the five models replaced between editions, which no figure here corrects for.

8.1 How this comparison is built

This chapter compares raw mention rate on the general cut; the general board’s rank and score change since July are shown in Chapter 3. All 40 questions were asked in both editions, so the change on the shared questions is the plain difference between the two levels: July’s 997 responses against October’s 999.

July’s levels are recounts. Every July figure in this chapter is July’s answers, counted the way this edition counts — re-read with this edition’s brand registry so both sides count a brand the same way. They are not the figures July’s page printed. July’s page counted names under a different rule (section 2.6), and the registry has changed since July, so any entity’s July level here can differ from July’s page. No figure from July’s page is quoted in this chapter.

The panel is the confound, and it is declared, not corrected. Claude Haiku 4.5, Perplexity Sonar and Grok 4.3 answered in both editions. Gemini 2.5 Flash-Lite and GPT-4o-mini answered in July only; Gemini 3.6 Flash and ChatGPT 5.6 Luna Pro answered in October only. Rebasing adjusts for the question set, which did not change; nothing adjusts for the panel. Every number below carries that confound: the direction of a large move is better supported than its exact size.

Origin in this chapter is this edition’s coding, applied to both editions. July’s own origin rule is not re-read, so the domestic and foreign counts below are one rule applied twice, not an origin share compared across the rule change.

Summary
Questions 40 shared; 0 added; 0 retired
Responses July 997 (recounted); October 999
Models in both editions Claude Haiku 4.5, Perplexity Sonar, Grok 4.3
Joined / left Gemini 3.6 Flash, ChatGPT 5.6 Luna Pro / Gemini 2.5 Flash-Lite, GPT-4o-mini
Entities named in both editions 198: rose 74, fell 111, unchanged 13
— domestic: held / rose 157 / 67
— foreign: held / rose 33 / 6
— no origin on file: held 8
Named only in October / only in July 195 / 96
Largest gain The ZKS (+19.01 points)
Largest fall Iconas Line (-8.54 points)
Raw mention rate, general cut. July = July's answers counted the way this edition counts. Two of five models differ between editions. Largest gain and fall are among entities named in both editions.

Of the 198 entities named in both editions, 74 rose, 111 fell and 13 were unchanged. Domestic entities account for 157 of the 198, of which 67 rose; foreign entities for 33, of which 6 rose; 8 have no origin on file and are neither. A further 195 entities were named only in October’s answers and 96 only in July’s. Those entities are not in the tables below, which rank only entities named in both editions.

8.2 The movers

Entity Origin July (recounted) October Δ points
The ZKS TR 6.82% (68 / 997) 25.83% (258 / 999) +19.01
Nocturne ⚠ TR 0.20% (2 / 997) 10.91% (109 / 999) +10.71
İpekyol TR 1.10% (11 / 997) 10.91% (109 / 999) +9.81
Vakko TR 1.40% (14 / 997) 10.71% (107 / 999) +9.31
Tarık Ediz TR 7.22% (72 / 997) 15.62% (156 / 999) +8.39
Sefamerve TR 12.94% (129 / 997) 21.32% (213 / 999) +8.38
Kayra ⚠ TR 0.30% (3 / 997) 7.41% (74 / 999) +7.11
Abiyefon TR 16.25% (162 / 997) 23.32% (233 / 999) +7.07
Armine ⚠ TR 0.50% (5 / 997) 7.31% (73 / 999) +6.81
Modanisa TR 12.14% (121 / 997) 18.82% (188 / 999) +6.68
Ten largest gains in raw mention rate among entities named in both editions, general cut. July = July's answers counted the way this edition counts; two of five models differ between editions. ⚠ Nocturne rests on 2 mentions in July's recounted answers (2 of 997), Kayra on 3 (3 of 997) and Armine on 5 (5 of 997); read their gains as indicative only.
Entity Origin July (recounted) October Δ points
Iconas Line TR 14.64% (146 / 997) 6.11% (61 / 999) -8.54
ModaSelvim TR 11.63% (116 / 997) 6.31% (63 / 999) -5.33
Pronovias ES 9.03% (90 / 997) 3.90% (39 / 999) -5.12
VznModa TR 6.62% (66 / 997) 1.60% (16 / 999) -5.02
Hülya Keser TR 7.52% (75 / 997) 2.50% (25 / 999) -5.02
Lacrima TR 8.02% (80 / 997) 3.20% (32 / 999) -4.82
EKOL TR 10.13% (101 / 997) 6.51% (65 / 999) -3.62
Yeşim Gelinlik ⚠ TR 4.21% (42 / 997) 0.90% (9 / 999) -3.31
Boyner TR 5.62% (56 / 997) 2.30% (23 / 999) -3.31
Modazehrada ⚠ TR 3.81% (38 / 997) 0.70% (7 / 999) -3.11
Ten largest falls among entities named in both editions, same basis. ⚠ Yeşim Gelinlik rests on 9 mentions this edition (9 of 999) and Modazehrada on 7 (7 of 999); read their falls as indicative only.

Gains. The ZKS moved from 6.82% of July’s recounted answers (68 of 997) to 25.83% (258 of 999), +19.01 points, and is third on the general board this edition. İpekyol moved from 1.10% (11 of 997) to 10.91% (109 of 999), +9.81, and Vakko from 1.40% (14 of 997) to 10.71% (107 of 999), +9.31. Tarık Ediz (+8.39, from 7.22% to 15.62%), Sefamerve (+8.38, from 12.94% to 21.32%), Abiyefon (+7.07, from 16.25% to 23.32%) and the marketplace Modanisa (+6.68, from 12.14% to 18.82%) rose too. Three of the ten largest gains rest on too few July mentions to read as more than direction: Nocturne ⚠ (+10.71, from 2 of 997 to 109 of 999), Kayra ⚠ (+7.11, from 3 to 74) and Armine ⚠ (+6.81, from 5 to 73). All ten gainers are domestic.

Falls. Iconas Line fell 8.54 points, from 14.64% (146 of 997) to 6.11% (61 of 999), and still qualifies, 34th on the general board. ModaSelvim (-5.33, from 11.63% to 6.31%) and EKOL (-3.62, from 10.13% to 6.51%) also fell and still qualify. Pronovias, Spanish, fell 5.12 points, from 9.03% to 3.90%, and is below the general-cut threshold this edition; it does qualify on the origin-neutral cut, 20th (Chapter 4). VznModa (-5.02), Hülya Keser (-5.02), Lacrima (-4.82) and the retailer Boyner (-3.31, from 5.62% to 2.30%) also sit below the threshold this edition. Yeşim Gelinlik ⚠ (-3.31, to 9 of 999) and Modazehrada ⚠ (-3.11, to 7 of 999) rest on too few mentions this edition to read as more than direction.

8.3 Where the gains sit

This report publishes no shared-model comparison and carries no per-model July figures, so it cannot say how much of any gain the panel change accounts for. What it can show is where each mover is named now.

The ZKS’s current visibility does not sit only in the models that joined. It is Claude Haiku 4.5’s most-named entity (27.00%), fourth for Grok 4.3 (52.50%) and ninth for Perplexity Sonar (16.00%) — all three answered in both editions — as well as second for ChatGPT 5.6 Luna Pro (31.16%), which joined. It is not in Gemini 3.6 Flash’s top ten. Abiyefon and Sefamerve have similar profiles: Abiyefon is third for Claude Haiku 4.5 (14.50%), fifth for Perplexity Sonar (23.00%) and sixth for Grok 4.3 (42.00%), and Sefamerve fifth for Grok 4.3 (51.00%) and eighth for Perplexity Sonar (19.50%); neither is in the top ten of a joining model. Whether any of them rose within the carried models cannot be read from this report.

Nocturne ⚠, İpekyol, Kayra ⚠ and Tarık Ediz lean on Gemini 3.6 Flash. Nocturne is eighth in Gemini 3.6 Flash’s answers (32.50%) and in no other model’s top ten. Gemini 3.6 Flash supplies 85 of İpekyol’s 109 general-cut mentions and 59 of Kayra’s 74 (Chapter 5). Tarık Ediz is fifth for Gemini 3.6 Flash (36.50%) and in no other model’s top ten. Gemini 3.6 Flash joined the panel this edition, so for these entities the panel change cannot be ruled out as a large part of the gain.

Modanisa is spread across both. It is in the top ten of one carried model, Grok 4.3 (38.50%), and of both joining models, Gemini 3.6 Flash (34.00%) and ChatGPT 5.6 Luna Pro (12.06%). Vakko and Armine ⚠ are in no model’s top ten, so this report cannot say where their gains sit.

Iconas Line’s remaining mentions are mostly Grok 4.3’s — 54 of its 61 (88.52%) — and Grok 4.3 answered in both editions. This report cannot say whether the fall happened within Grok 4.3 or reflects the models that left the panel.


9. The Discovery Ecosystem: Where AI Learns About Brands

One-sentence takeaway: The most widely cited of 213 domains is trendyol.com, in 84 of 200 question-model cells (42.00%), followed by carmen.com.tr and thezks.com; one model’s citations are capped and another’s grounding is partial, so the map is uneven by model.

When an assistant answers with web search, it draws on whatever is available to it at that moment. The cited addresses were reduced to domains and counted by question-model cell — one cell is one question put to one model, holding that question’s repeats — never by response. Herm.io’s own domain was removed before any of these counts. Three units appear in this chapter, and they are not interchangeable:

  • Citations recorded — 9,306. The citation entries the models returned.
  • Link rows — 2,907. Attribution rows, one per model, entity, domain and attribution rule; one mention can be attributed to several domains.
  • Mentions attributed — 7,738. Those rows’ mention counts, summed. Because one mention can sit on several rows, this is not a count of distinct mentions and does not compare with any mention rate’s numerator.

Across all answers, 213 domains were cited, 70 of them in a single cell only.

9.1 The most-cited domains

# Domain Cells citing Share of cells Mentions attributed Owner
1 trendyol.com 84 42.00% 888 Trendyol
2 carmen.com.tr 69 34.50% 432 Carmen
3 thezks.com 59 29.50% 537 The ZKS
4 abiyefon.com 51 25.50% 218 Abiyefon
5 alfa-beta.com.tr 43 21.50% 218 AlfaBeta
6 adl.com.tr 41 20.50% 239 ADL
7 sefamerve.com 30 15.00% 178 Sefamerve
8 modanisa.com 28 14.00% 163 Modanisa
9 hepsiburada.com 26 13.00% 126 Hepsiburada
10 koton.com 25 12.50% 236 Koton
11 kenzel.com.tr 22 11.00% 80 Kenzel
12 gizia.com 19 9.50% 122 GIZIA
13 cengizakturk.com 19 9.50% 63 Cengiz Aktürk
14 roman.com.tr 18 9.00% 71 Roman
15 lcw.com 17 8.50% 96 LC Waikiki
Top 15 of the 30 domains in the report's data file, by question-model cells citing them (of 200). Mentions attributed are attribution-row units, not distinct mentions.

trendyol.com is cited in 84 of the 200 cells (42.00%), followed by carmen.com.tr (69 cells, 34.50%), thezks.com (59, 29.50%), abiyefon.com (51, 25.50%), alfa-beta.com.tr (43, 21.50%) and adl.com.tr (41, 20.50%). These six are the domains of the general board’s top six entities. Every domain in the table has an owner among the tracked entities: three are marketplaces — trendyol.com, modanisa.com and hepsiburada.com — and the rest are the own sites of brands and retailers. thezks.com carries more attributed mentions than carmen.com.tr, 537 against 432, from fewer cells — a reminder that attribution rows count every entity a cited page is credited with.

A cited address is a source, not a mention. carmen.com.tr counts here as a domain the models read; the same address written as a name inside an answer is not counted as Carmen (section 2.4).

9.2 Entities’ own sites

184 of the 213 domains are the own site of some tracked entity. Tracked is not ranked: marketplaces, publishers and entities that qualify in no cut have sites too, so this count does not reconcile with the leaderboard, and it should not be read as brands citing themselves. Rows on those domains account for 4,645 of the 7,738 attributed mentions — attribution-row units, not distinct mentions.

That figure is understated. Seven cited domains look like a tracked entity’s own site but are not attributed to it, because the registry holds a different website: aytimurtekstil.com (Aytimur Tekstil), erenabiye.com (Eren Abiye; registry erenabiye.com.tr), esergiyim.com.tr (Eser Giyim), matikabiyem.com (Matik Abiyem), nihan.com.tr (Nihan; registry nihangiyim.com), tubaergin.com.tr (Tuba Ergin; registry tubaergin.com) and vogue.com.tr (Vogue; registry vogue.com). They are therefore not counted among own sites.

In 61 rows, the rule that assigns a mention to the entity’s own domain and the rule that assigns it by position in the answer disagreed about which citation the mention belongs to. Those conflicts are surfaced in the data, not resolved.

9.3 Why the map is uneven

The models do not contribute equally. Claude Haiku 4.5 returned exactly 5 citations on every one of its 200 answers — 1,000 in all — so its domain counts stop at a ceiling and do not compare with an uncapped model’s. Perplexity Sonar alone returned 5,930 of the 9,306 citations. Gemini 3.6 Flash returned 388 and grounded 20.00% of its answers. A domain’s place in this table therefore reflects which models cite at length as much as what any one model reads.

This edition carries no source-type classification of domains and no per-response citation counts, so July’s source-type mix has no counterpart here, and July’s counts of responses citing a domain are a different unit from the cells counted above. Nothing in this chapter tests whether being cited causes visibility or merely accompanies it.


10. What the Patterns Suggest

One-sentence takeaway: Six neutral readings, not recommendations: a marketplace sets the top of the board, the shelf stays mostly domestic on both question scopes, frequency and position are separate routes onto the board, the two joining models carry several ranks near the threshold, answer format decides who is seen in the long tail, and the top entities’ own domains are the most cited.

This chapter describes patterns in the data. It is not advice, and visibility remains separate from quality.

A marketplace sets the top of the board. Trendyol leads the general cut (100.00), the origin-neutral cut (100.00) and the attribute (90.85) and use-case (100.00) questions; Carmen leads the origin-constrained cut (93.20) and the discovery questions (88.84). Because Trendyol holds every maximum on the general cut, every general-cut score — including those on the brand-only board — is read against a marketplace.

The shelf stays mostly domestic on both question scopes. On the 20 origin-constrained questions domestic entities take 100.00% of qualifying mentions and foreign entities 0.00%; on the 20 origin-neutral questions, 83.21% and 16.79% — with origin on file for every qualifying mention on both. The assistants read “yerli” as a constraint; without it, five foreign entities qualify, the highest-placed of them Mango in 10th.

Frequency and position are separate routes onto the board. ADL is named in 26.03% of all responses and ranks fourth; The ZKS, in 25.83%, ranks third because it is named earlier. Beymen ranks tenth on 7.61% with an MRR of 0.4565. On narrower cuts the effect is sharper: Raisa Vanessa ranks 17th on the origin-constrained questions on 5.61% with an MRR of 0.6311, and Faik Sönmez fifth on the attribute questions on 6.15% with an MRR of 0.6133.

The two joining models carry several ranks near the threshold. Gemini 3.6 Flash supplies most of the general-cut mentions of DCEY, Sagaza, Machka, Setrms, Kayra, Oleg Cassini and İpekyol; ChatGPT 5.6 Luna Pro supplies all of Berneva’s. All of them sit between 20th and 38th on the general board. A change of panel moved the edge of this board more visibly than its top.

Answer format decides who is seen in the long tail. Gemini 3.6 Flash names 12.38 entities per answer, reaches 170 distinct entities and names 57 that no other model does; Claude Haiku 4.5 names 4.44 per answer, 148 distinct and 25 alone. For an entity at the edge of the board, whether it is seen depends on the assistant’s format as much as on the entity.

The top entities’ own domains are the most cited. trendyol.com, carmen.com.tr, thezks.com, abiyefon.com, alfa-beta.com.tr and adl.com.tr are the six most widely cited domains, in 84, 69, 59, 51, 43 and 41 of 200 cells, and they belong to the general board’s top six entities. This is an observed association between being cited and being named, not evidence that one causes the other.


11. Limitations and Notes

One-sentence takeaway: This is one market on one collection day, measured on a panel, a counting rule and an origin rule that all changed since July, with scope unresolved for every entity; read every figure with the notes below.

  • The panel changed. Gemini 2.5 Flash-Lite and GPT-4o-mini were replaced by Gemini 3.6 Flash and ChatGPT 5.6 Luna Pro. Chapter 8 declares this confound beside every figure; it does not correct for it, and some movers — Nocturne ⚠, İpekyol and Kayra ⚠ among them — are named now mainly by Gemini 3.6 Flash, which joined.
  • The counting rule changed. July’s page counted a validated brand name wherever it appeared in the answer text; this edition counts a brand or company only where a second language model lists it as named in the answer, not where it appears only in a web address. July’s levels in this report are therefore recounts of July’s answers, not the figures July printed.
  • The registry changed. Brands have been renamed and registrations reconciled since July. The recount of July’s answers uses this edition’s registry.
  • The origin rule changed. July required a brand to be founded and headquartered in Turkey; this edition uses brand home. No origin share is compared across the change.
  • The brand-only view changed. July’s removed marketplaces and retailers; this edition’s removes marketplaces only, so the two views are different lists.
  • Origin classification involves judgment. Brand home is a researched decision, and brands run from one country under an owner in another are genuinely arguable cases. 40 entities have no origin on file; they appear in every table and in no share.
  • Scope was not resolved. No entity has a product category on file, so the study could not separate core evening-wear entities from adjacent ones, and all 393 are eligible to rank. An entity whose main business lies outside evening wear can rank if the assistants name it often enough.
  • Untyped and provisional entities. Two entities, Didem Soydan and Armut.com, are both untyped and provisional; neither qualifies in any cut. Armut.com is a likely duplicate of the marketplace Armut, which is tracked separately.
  • Market-suffix names. Two names carry a market suffix whose parent brand is not tracked in this edition, so nothing could be folded into a parent; each keeps its own origin.
  • Brands written as web addresses. A brand written in an answer only as its web address is not counted as that brand: “Carmen.com.tr”, in 11 responses, is not counted as Carmen (section 2.4).
  • One model’s answers have unknown grounding. 160 of Gemini 3.6 Flash’s 200 answers, and 4 of ChatGPT 5.6 Luna Pro’s 199 analysed answers, returned no grounding metadata. They are not memory answers, and nothing in this report treats them as such.
  • One model’s citations are capped. Claude Haiku 4.5 returned exactly 5 citations per answer, so its source counts are truncated (Chapter 9).
  • Own-site figures are understated. Seven cited domains — aytimurtekstil.com, erenabiye.com, esergiyim.com.tr, matikabiyem.com, nihan.com.tr, tubaergin.com.tr and vogue.com.tr — look like a tracked entity’s own site but are not attributed to it, because the registry holds a different website.
  • Qualifying positions on fewer than three models. Thirty qualifying entity-cut positions rest on one or two models. On the general cut: Sagaza (2 models, 66 mentions), Berneva (1 model, 50) and DCEY (2 models, 53). On the origin-constrained cut: Raisa Vanessa, Sagaza, Kayra, Berneva, Machka, Iconas Line and Exquise. On the origin-neutral cut: DCEY. The question-type cases are named in Chapter 6. Their visibility is model-specific, not panel-wide.
  • Rank built on position. Beymen on the general cut, Raisa Vanessa on the origin-constrained and discovery cuts, and Faik Sönmez on the attribute cut rank high mainly because they are named early when named at all. That is the score working as designed, but it is a different kind of visibility from being named in a large share of all answers.
  • Small score gaps are not meaningful. In the packed middle and lower part of each board — By Saygı (40.35) and Setrms (40.36), or Machka (35.10) and Kayra (34.92) — differences of a point or less should not be over-interpreted.
  • Components can exceed 100. Because each component is scaled to the qualified leader and never clipped, 88 of 1,730 entity-cut rows (5.10%) exceed 100 on a component — all of them unqualified.
  • Near-name pairs. Twelve pairs of similar names among entities named in at least three answers were flagged: Armut / Armut.com, Beymen Club / Beymen, DCEY / Dcey Bridal, Dekolte / EKOL, Farika / Fabrika, Kayra / Kayıra, Nihan / Nihan Peker, OGGUSTO / Gusto, Pazarama / Zara, Trendyolmilla / Trendyol, Tuğba / Bituğba Store and Vakkorama / Vakko. None shares a website on file, and each pair is tracked as two entities. The outcome is in the data; the reasoning behind each merge or non-merge is not.
  • Trend rows on thin counts. Five rows in the two mover tables in Chapter 8 rest on fewer than 10 mentions in one edition — Nocturne ⚠, Kayra ⚠ and Armine ⚠ in July’s recounted answers, Yeşim Gelinlik ⚠ and Modazehrada ⚠ in October’s — and are marked ⚠ wherever they appear.
  • System prompt and locale. Neither was recorded with this run.
  • Single, point-in-time measurement. All responses were collected on 2 October 2026. Repeats within each question-model cell capture variation within a model, not drift from day to day.
  • Models are probabilistic. The same question can produce different answers; repeats reduce this variability but do not remove it.
  • Web search depends on momentary conditions. Grounded answers depend on what was indexed and served at the time of collection.
  • Precision over recall. A brand or company is counted only where it is named in the answer, not where it appears only in a web address, and a shop named only as the place to buy is often not counted. Brands referred to indirectly are not counted either.
  • Visibility is not quality. This report measures mention, position and breadth only. It does not assess whether a brand was described positively, whether a recommendation was accurate, or whether a dress is well made. A score is not an endorsement.
  • Model versions date quickly. These findings are specific to the model versions in section 2.1, as of October 2026.

12. Appendix & Data

One-sentence takeaway: All 40 questions and all 393 tracked entities behind this report are listed below, and the underlying data is available on request.

12.1 All 40 questions

# Type Scope Question (Turkish)
1 Attribute Origin-neutral yazın terletmeyen abiye elbise
2 Use case Origin-neutral nikah için sade beyaz elbise nereden alınır
3 Attribute Origin-neutral göbek ve basen kapatan abiye nereden alınır
4 Attribute Origin-constrained türk malı büyük beden abiye önerisi
5 Use case Origin-constrained yerli markadan anne abiyesi arıyorum
6 Discovery Origin-constrained yerli mezuniyet elbisesi önerisi
7 Attribute Origin-constrained yerli tesettür abiye zayıf gösteren modeller
8 Attribute Origin-neutral saten abiye hangi marka iyi
9 Use case Origin-neutral hamile abiye nereden bulunur
10 Use case Origin-constrained türk malı kına elbisesi önerir misin
11 Use case Origin-constrained yerli markalardan mezuniyet elbisesi bakıyorum
12 Attribute Origin-constrained yerli saten abiye arıyorum
13 Use case Origin-neutral salon düğününde giymelik uzun abiye
14 Attribute Origin-neutral büyük beden abiye güzel duran markalar
15 Discovery Origin-neutral davet elbisesi nereden alınır
16 Use case Origin-neutral kına gecesi için kırmızı elbise önerisi
17 Use case Origin-constrained tesettürlü biri için yerli düğün abiyesi
18 Discovery Origin-constrained türk malı davet elbisesi önerir misin
19 Discovery Origin-constrained iyi bir yerli abiye markası var mı
20 Attribute Origin-neutral kısa boylular için abiye önerisi
21 Attribute Origin-neutral zayıf gösteren abiye elbise önerisi
22 Discovery Origin-neutral şık gece elbisesi hangi marka iyi
23 Discovery Origin-constrained davet elbisesi için yerli marka arıyorum
24 Discovery Origin-constrained Türk markalarından şık gece elbisesi
25 Use case Origin-neutral oğlumun mezuniyetinde giyeceğim şık elbise
26 Discovery Origin-neutral en iyi abiye markaları
27 Discovery Origin-constrained yerli tesettür abiye nereden alınır
28 Use case Origin-constrained yerli marka kısa mezuniyet elbisesi
29 Discovery Origin-constrained kaliteli yerli abiye nereden bakayım
30 Discovery Origin-neutral düğün misafiri için elbise nereden bakılır
31 Use case Origin-constrained Türk markalarından nikah sonrası yemek elbisesi
32 Use case Origin-constrained evde söz için sade yerli elbise
33 Attribute Origin-constrained iç göstermeyen astarlı yerli abiye nereden alınır
34 Attribute Origin-constrained kaliteli kumaşlı yerli davet elbisesi
35 Discovery Origin-neutral abiye elbise önerisi
36 Discovery Origin-neutral mezuniyet elbisesi markaları
37 Use case Origin-neutral arkadaşımın kır düğünü için elbise önerisi
38 Attribute Origin-constrained uygun fiyatlı yerli abiye markası
39 Discovery Origin-neutral tesettür abiye hangi siteden alınır
40 Attribute Origin-neutral uygun fiyatlı kaliteli abiye
All 40 questions, each carried verbatim from July 2026. No question records a subcategory (section 2.3).

12.2 All tracked entities (393)

The table lists all 393 entities tracked in this study, ordered by general-cut mentions, and marks the 38 that qualify on the general cut. 65 entities qualify in at least one cut. Entities with no origin on file are shown as Unknown. Brand pages are not linked in this edition.

393 brands · 38 ranked
Brands that qualified for ranking (≥5% mentions) are marked with a green tag.
Brand Origin Mentions Rate Models Status
Trendyol TR 424 42.44% 5 Ranked
Carmen TR 384 38.44% 5 Ranked
ADL TR 260 26.03% 5 Ranked
The ZKS TR 258 25.83% 5 Ranked
AlfaBeta TR 242 24.22% 5 Ranked
Abiyefon TR 233 23.32% 5 Ranked
Sefamerve TR 213 21.32% 5 Ranked
Koton TR 196 19.62% 5 Ranked
Modanisa TR 188 18.82% 5 Ranked
Hepsiburada TR 178 17.82% 4 Ranked
Tarık Ediz TR 156 15.62% 5 Ranked
GIZIA TR 139 13.91% 5 Ranked
Kenzel TR 120 12.01% 5 Ranked
Cengiz Aktürk TR 111 11.11% 4 Ranked
Nocturne TR 109 10.91% 4 Ranked
İpekyol TR 109 10.91% 4 Ranked
Vakko TR 107 10.71% 5 Ranked
Roman TR 104 10.41% 5 Ranked
Setre TR 96 9.61% 5 Ranked
Rengin TR 94 9.41% 5 Ranked
Zara Foreign 83 8.31% 5 Ranked
Lafaba TR 77 7.71% 5 Ranked
Beymen TR 76 7.61% 5 Ranked
Kayra TR 74 7.41% 3 Ranked
Setrms TR 74 7.41% 4 Ranked
Armine TR 73 7.31% 4 Ranked
Machka TR 72 7.21% 3 Ranked
Oleg Cassini Foreign 72 7.21% 3 Ranked
LC Waikiki TR 67 6.71% 4 Ranked
Sagaza TR 66 6.61% 2 Ranked
EKOL TR 65 6.51% 4 Ranked
Mango Foreign 65 6.51% 5 Ranked
ModaSelvim TR 63 6.31% 5 Ranked
Iconas Line TR 61 6.11% 4 Ranked
Twist TR 61 6.11% 5 Ranked
DCEY TR 53 5.31% 2 Ranked
Berneva TR 50 5.01% 1 Ranked
By Saygı TR 50 5.01% 4 Ranked
Alvina TR 48 4.80% 3 Tracked
Eser Giyim TR 48 4.80% 5 Tracked
Trendyolmilla TR 46 4.60% 4 Tracked
H&M Foreign 44 4.40% 5 Tracked
Nihan TR 41 4.10% 3 Tracked
Network TR 39 3.90% 4 Tracked
Pronovias Foreign 39 3.90% 5 Tracked
BSL TR 38 3.80% 5 Tracked
Eyyo Atelier TR 37 3.70% 5 Tracked
Raisa Vanessa TR 37 3.70% 2 Tracked
Deniz Butik TR 36 3.60% 3 Tracked
Tuvanam TR 36 3.60% 1 Tracked
Akakçe TR 34 3.40% 3 Tracked
Lacrima TR 32 3.20% 4 Tracked
Ju Istanbul TR 31 3.10% 1 Tracked
Manuka TR 31 3.10% 5 Tracked
Touche Privé TR 31 3.10% 1 Tracked
Aker TR 29 2.90% 4 Tracked
Arya Moda TR 29 2.90% 5 Tracked
Exquise TR 28 2.80% 1 Tracked
Kıyafet Sepeti TR 27 2.70% 4 Tracked
Tuba Ergin TR 27 2.70% 2 Tracked
Tuğba TR 27 2.70% 1 Tracked
N11 TR 26 2.60% 2 Tracked
Faik Sönmez TR 25 2.50% 5 Tracked
Hülya Keser TR 25 2.50% 3 Tracked
Gigii's TR 24 2.40% 1 Tracked
Nurmelek TR 24 2.40% 3 Tracked
Özgür Masur TR 24 2.40% 2 Tracked
Boyner TR 23 2.30% 5 Tracked
LİLAS XXL TR 23 2.30% 4 Tracked
DeFacto TR 22 2.20% 5 Tracked
Minel Aşk TR 22 2.20% 1 Tracked
Şule Giyim TR 22 2.20% 5 Tracked
Amine Hüma TR 21 2.10% 3 Tracked
Coupe Official TR 21 2.10% 1 Tracked
Ola Abiye TR 21 2.10% 2 Tracked
PULLIMM Unknown 21 2.10% 2 Tracked
Tozlu TR 21 2.10% 3 Tracked
Nedret Taciroğlu TR 20 2.00% 1 Tracked
Forever New Foreign 19 1.90% 4 Tracked
Hüseyin Baştürk TR 19 1.90% 3 Tracked
Museum of Fine Clothing TR 19 1.90% 1 Tracked
Glorvia TR 18 1.80% 3 Tracked
My Best Friends TR 18 1.80% 2 Tracked
Seçil TR 18 1.80% 4 Tracked
So Lady TR 18 1.80% 1 Tracked
Ömür Inn TR 18 1.80% 4 Tracked
Ardanewline TR 17 1.70% 3 Tracked
Tuvid XXL TR 17 1.70% 3 Tracked
Dolap TR 16 1.60% 2 Tracked
Doridorca TR 16 1.60% 4 Tracked
House of Ogan TR 16 1.60% 1 Tracked
Pınar Şems TR 16 1.60% 2 Tracked
Sorbe TR 16 1.60% 1 Tracked
SpringStore TR 16 1.60% 3 Tracked
VznModa TR 16 1.60% 3 Tracked
Çiçeksepeti TR 16 1.60% 3 Tracked
Abiye Merkezi TR 15 1.50% 3 Tracked
Butik Nilyam TR 15 1.50% 3 Tracked
Farika TR 15 1.50% 1 Tracked
Gardrops TR 15 1.50% 2 Tracked
Burak Baran Fashion TR 14 1.40% 3 Tracked
Düğün.com TR 14 1.40% 3 Tracked
Patırtı TR 14 1.40% 2 Tracked
RUE Les Créateurs TR 14 1.40% 2 Tracked
Saygın TR 14 1.40% 2 Tracked
Öznur Unknown 14 1.40% 2 Tracked
Betty & Sam TR 13 1.30% 4 Tracked
Elben Moda Unknown 13 1.30% 3 Tracked
HOX'S TR 13 1.30% 3 Tracked
Miss Diamond TR 13 1.30% 1 Tracked
Şeref Vural TR 13 1.30% 3 Tracked
Accort TR 12 1.20% 3 Tracked
Cercis TR 12 1.20% 3 Tracked
Cimri TR 12 1.20% 2 Tracked
Kevser Sarıoğlu TR 12 1.20% 1 Tracked
Nazzan TR 12 1.20% 2 Tracked
Nurtuba TR 12 1.20% 3 Tracked
Pazarium TR 12 1.20% 2 Tracked
Pierre Cardin Foreign 12 1.20% 2 Tracked
Sarar TR 12 1.20% 2 Tracked
Beyyoglu TR 11 1.10% 3 Tracked
Dans TR 11 1.10% 2 Tracked
Gusto TR 11 1.10% 5 Tracked
Nihan Peker TR 11 1.10% 2 Tracked
Olcay TR 11 1.10% 4 Tracked
ESSEDRESS Unknown 10 1.00% 2 Tracked
Elbise Kirala TR 10 1.00% 4 Tracked
Gamze Özkul TR 10 1.00% 1 Tracked
Mayabella Unknown 10 1.00% 1 Tracked
Pionto TR 10 1.00% 1 Tracked
Serica Moda TR 10 1.00% 3 Tracked
Tofisa TR 10 1.00% 2 Tracked
Zühre TR 10 1.00% 3 Tracked
Al-Tatari TR 9 0.90% 4 Tracked
Alia TR 9 0.90% 2 Tracked
Aysa Tekin TR 9 0.90% 3 Tracked
Aysira TR 9 0.90% 2 Tracked
Disentis Modest TR 9 0.90% 1 Tracked
Elie Saab Foreign 9 0.90% 3 Tracked
Ilmio TR 9 0.90% 2 Tracked
Neva Style TR 9 0.90% 2 Tracked
Perspective TR 9 0.90% 1 Tracked
PodyumPlus TR 9 0.90% 3 Tracked
Sheeny Unknown 9 0.90% 3 Tracked
Tarz Kadın Unknown 9 0.90% 3 Tracked
Venn Unknown 9 0.90% 1 Tracked
Yeşim Gelinlik TR 9 0.90% 3 Tracked
Zagrep Unknown 9 0.90% 4 Tracked
Beymen Club TR 8 0.80% 2 Tracked
Beyza Giyim TR 8 0.80% 1 Tracked
Beyza Online TR 8 0.80% 1 Tracked
Cengiz Abazoğlu Unknown 8 0.80% 1 Tracked
ESRAHELVACI TR 8 0.80% 2 Tracked
Gökçe Moda Evi TR 8 0.80% 2 Tracked
Hürrem Aldinç TR 8 0.80% 2 Tracked
Jovani Foreign 8 0.80% 2 Tracked
Leyuze Butik TR 8 0.80% 2 Tracked
MK Moda Tasarım TR 8 0.80% 2 Tracked
Nefeli Store TR 8 0.80% 2 Tracked
Odrica TR 8 0.80% 2 Tracked
Rmg TR 8 0.80% 3 Tracked
Stradivarius Foreign 8 0.80% 3 Tracked
Didem Soydan Unknown 7 0.70% 1 Tracked
DreamON TR 7 0.70% 2 Tracked
Dressland TR 7 0.70% 1 Tracked
E-Tesettür Unknown 7 0.70% 1 Tracked
GILLA'S QUEEN Unknown 7 0.70% 1 Tracked
Love My Body TR 7 0.70% 2 Tracked
Modacadiri TR 7 0.70% 2 Tracked
Modazehrada TR 7 0.70% 3 Tracked
Nika Fashion TR 7 0.70% 2 Tracked
Qupra TR 7 0.70% 1 Tracked
Raşit Bağzıbağlı TR 7 0.70% 1 Tracked
Rentony TR 7 0.70% 2 Tracked
Tesettür Island TR 7 0.70% 3 Tracked
Valentino Foreign 7 0.70% 3 Tracked
Yusem Tesettür TR 7 0.70% 2 Tracked
Zeynep Tosun TR 7 0.70% 1 Tracked
Aytimur Tekstil Unknown 6 0.60% 3 Tracked
Ceylan Otantik TR 6 0.60% 1 Tracked
Davet Çok Elbisem Yok TR 6 0.60% 2 Tracked
Davetten Davete TR 6 0.60% 1 Tracked
Dekolte TR 6 0.60% 2 Tracked
Dilvin TR 6 0.60% 1 Tracked
Fabrika TR 6 0.60% 2 Tracked
Giyze TR 6 0.60% 2 Tracked
Merci MS Fancy TR 6 0.60% 2 Tracked
Milagron TR 6 0.60% 1 Tracked
Mina Lidya TR 6 0.60% 2 Tracked
Modavina TR 6 0.60% 2 Tracked
OXXO TR 6 0.60% 1 Tracked
Tiara TR 6 0.60% 2 Tracked
Wom Boutique TR 6 0.60% 1 Tracked
Şans TR 6 0.60% 2 Tracked
A46 Unknown 5 0.50% 1 Tracked
ALFUDO TR 5 0.50% 2 Tracked
Abiyenial TR 5 0.50% 2 Tracked
Alisse Nuera TR 5 0.50% 3 Tracked
Babil Butik TR 5 0.50% 1 Tracked
Berissima Bride TR 5 0.50% 2 Tracked
Bituğba Store TR 5 0.50% 2 Tracked
Busa TR 5 0.50% 2 Tracked
CML Collection Unknown 5 0.50% 2 Tracked
Civil TR 5 0.50% 1 Tracked
Desigual Foreign 5 0.50% 1 Tracked
Dilek Hanif TR 5 0.50% 1 Tracked
Doucée TR 5 0.50% 1 Tracked
GiyzaMila TR 5 0.50% 1 Tracked
Gizem Butik TR 5 0.50% 1 Tracked
Gönül Kolat TR 5 0.50% 1 Tracked
Just Cavalli Foreign 5 0.50% 1 Tracked
Karl Lagerfeld Foreign 5 0.50% 1 Tracked
LACASE Unknown 5 0.50% 2 Tracked
Laranor TR 5 0.50% 1 Tracked
Massimo Dutti Foreign 5 0.50% 2 Tracked
Melek Araz TR 5 0.50% 1 Tracked
Meliha Tarhanacı TR 5 0.50% 3 Tracked
Nur Karaata TR 5 0.50% 2 Tracked
OGGUSTO TR 5 0.50% 2 Tracked
Puane TR 5 0.50% 2 Tracked
Seamoda TR 5 0.50% 3 Tracked
Self-Portrait Foreign 5 0.50% 1 Tracked
Sherri Hill Foreign 5 0.50% 1 Tracked
Societta Abiye TR 5 0.50% 2 Tracked
Spazio TR 5 0.50% 2 Tracked
United Colors of Benetton Foreign 5 0.50% 1 Tracked
UnoMoi TR 5 0.50% 1 Tracked
Zuhair Murad Foreign 5 0.50% 1 Tracked
AYŞEYAMAÇLI Unknown 4 0.40% 1 Tracked
Abiyecim TR 4 0.40% 3 Tracked
Akay Gelinlik TR 4 0.40% 1 Tracked
Ankyra TR 4 0.40% 2 Tracked
Ayşegül Kalelio TR 4 0.40% 2 Tracked
Azra Design TR 4 0.40% 1 Tracked
Brandroom TR 4 0.40% 1 Tracked
Ceylen TR 4 0.40% 1 Tracked
Chanel Foreign 4 0.40% 2 Tracked
Efelif TR 4 0.40% 2 Tracked
Eren Abiye TR 4 0.40% 1 Tracked
InStyle Unknown 4 0.40% 2 Tracked
Jeyan Gedik TR 4 0.40% 1 Tracked
Kaftan Collection TR 4 0.40% 1 Tracked
Kayıra Unknown 4 0.40% 1 Tracked
Mac Duggal Foreign 4 0.40% 1 Tracked
Modarana TR 4 0.40% 2 Tracked
Muse for All TR 4 0.40% 1 Tracked
Needle & Thread Foreign 4 0.40% 1 Tracked
Nursena Moda TR 4 0.40% 2 Tracked
Ottomanwear TR 4 0.40% 1 Tracked
Reformation Foreign 4 0.40% 3 Tracked
Rent Code Unknown 4 0.40% 1 Tracked
Rent The Look Unknown 4 0.40% 1 Tracked
Rivus TR 4 0.40% 1 Tracked
Seninmoda TR 4 0.40% 1 Tracked
Sima Couture Foreign 4 0.40% 1 Tracked
Solace London Foreign 4 0.40% 1 Tracked
Sunfere Foreign 4 0.40% 2 Tracked
Terani Foreign 4 0.40% 1 Tracked
Tesettür Home TR 4 0.40% 1 Tracked
Vizyon La Mira TR 4 0.40% 1 Tracked
Vogue Foreign 4 0.40% 1 Tracked
Zimmermann Foreign 4 0.40% 1 Tracked
Zishan TR 4 0.40% 2 Tracked
eHamileGiyim TR 4 0.40% 1 Tracked
İmajButik TR 4 0.40% 1 Tracked
İpek Kıramer TR 4 0.40% 2 Tracked
Şirin Butik TR 4 0.40% 3 Tracked
ASOS Foreign 3 0.30% 2 Tracked
Afife Unknown 3 0.30% 2 Tracked
Angelino TR 3 0.30% 1 Tracked
Armani Exchange Foreign 3 0.30% 1 Tracked
Armut TR 3 0.30% 2 Tracked
Armut.com Unknown 3 0.30% 1 Tracked
Arıkan TR 3 0.30% 1 Tracked
Azazie Foreign 3 0.30% 2 Tracked
Aşkın Göksu TR 3 0.30% 2 Tracked
BDM Exclusive TR 3 0.30% 1 Tracked
Burcum By B.C.C. Unknown 3 0.30% 2 Tracked
Bürün Unknown 3 0.30% 1 Tracked
Büyük Bedeniz TR 3 0.30% 1 Tracked
Chilia Design TR 3 0.30% 1 Tracked
Dcey Bridal TR 3 0.30% 1 Tracked
Devr-i Tesettür TR 3 0.30% 2 Tracked
Divina Wedding TR 3 0.30% 1 Tracked
Emsale TR 3 0.30% 2 Tracked
Esma Karadağ TR 3 0.30% 3 Tracked
Gamze Polat Unknown 3 0.30% 1 Tracked
Gebe TR 3 0.30% 1 Tracked
Glr & Gym Unknown 3 0.30% 2 Tracked
HESNA TR 3 0.30% 1 Tracked
Işşıl TR 3 0.30% 1 Tracked
Jacquemus Foreign 3 0.30% 1 Tracked
Kameya TR 3 0.30% 1 Tracked
Keikei TR 3 0.30% 2 Tracked
Kırmızı Butik Unknown 3 0.30% 1 Tracked
Lamia Giyim TR 3 0.30% 3 Tracked
Lirenza Abiye TR 3 0.30% 1 Tracked
Love on Friday TR 3 0.30% 1 Tracked
Lulus Foreign 3 0.30% 2 Tracked
Matik Abiyem Unknown 3 0.30% 1 Tracked
MegaButik TR 3 0.30% 2 Tracked
Merhaba Boutique TR 3 0.30% 2 Tracked
Mevra TR 3 0.30% 2 Tracked
Moda Berray TR 3 0.30% 1 Tracked
Modafeminen Unknown 3 0.30% 2 Tracked
Modapinhan TR 3 0.30% 1 Tracked
Modatalika TR 3 0.30% 1 Tracked
Mossta TR 3 0.30% 1 Tracked
Nefise TR 3 0.30% 1 Tracked
Next Foreign 3 0.30% 2 Tracked
Patrizia Pepe Foreign 3 0.30% 1 Tracked
Pazarama TR 3 0.30% 1 Tracked
Pinko Foreign 3 0.30% 1 Tracked
Refka Abiye TR 3 0.30% 2 Tracked
SUBU TASARIM Unknown 3 0.30% 1 Tracked
Sedef Unknown 3 0.30% 1 Tracked
Semala Butik TR 3 0.30% 1 Tracked
Setr-i Nur TR 3 0.30% 1 Tracked
Tommy Hilfiger Foreign 3 0.30% 1 Tracked
Trend Tesettür TR 3 0.30% 2 Tracked
Vakkorama TR 3 0.30% 2 Tracked
Vipiri Unknown 3 0.30% 1 Tracked
Zoezoe Unknown 3 0.30% 1 Tracked
Züleyha Kuru TR 3 0.30% 1 Tracked
lovebox TR 3 0.30% 1 Tracked
vuvutasarım Unknown 3 0.30% 1 Tracked
ALLDAY TR 2 0.20% 1 Tracked
Anthropologie Foreign 2 0.20% 2 Tracked
Arzu Kaprol TR 2 0.20% 2 Tracked
BAQA TR 2 0.20% 1 Tracked
Büyük Moda TR 2 0.20% 2 Tracked
David's Bridal Foreign 2 0.20% 1 Tracked
Denokids TR 2 0.20% 1 Tracked
DressIstanbul TR 2 0.20% 1 Tracked
Forever 21 Foreign 2 0.20% 1 Tracked
Givenchy SA Foreign 2 0.20% 1 Tracked
Gül Hürgel TR 2 0.20% 2 Tracked
HİRAMİR TR 2 0.20% 1 Tracked
LOEWE Foreign 2 0.20% 1 Tracked
Lenta Moda TR 2 0.20% 2 Tracked
Mavi TR 2 0.20% 1 Tracked
Miraggio TR 2 0.20% 1 Tracked
Modailgi TR 2 0.20% 2 Tracked
Morhipo TR 2 0.20% 1 Tracked
SIRRI TR 2 0.20% 1 Tracked
Sateen TR 2 0.20% 1 Tracked
Suud Collection TR 2 0.20% 1 Tracked
Tuay Karaca TR 2 0.20% 2 Tracked
Abercrombie & Fitch Foreign 1 0.10% 1 Tracked
AbiyeElbisecim Unknown 1 0.10% 1 Tracked
Alexander McQueen Foreign 1 0.10% 1 Tracked
Alexandre Vauthier Foreign 1 0.10% 1 Tracked
Balmain Türkiye Foreign 1 0.10% 1 Tracked
Bershka Türkiye Foreign 1 0.10% 1 Tracked
Beyliss TR 1 0.10% 1 Tracked
Davet Elegance TR 1 0.10% 1 Tracked
Dior Foreign 1 0.10% 1 Tracked
Dolce&Gabbana Foreign 1 0.10% 1 Tracked
ESOTTE TR 1 0.10% 1 Tracked
ETRO Foreign 1 0.10% 1 Tracked
Enderun Unknown 1 0.10% 1 Tracked
Etsy Foreign 1 0.10% 1 Tracked
GUESS Foreign 1 0.10% 1 Tracked
Ghisa TR 1 0.10% 1 Tracked
Gülhan Moda TR 1 0.10% 1 Tracked
Hanezza TR 1 0.10% 1 Tracked
Haza Moda TR 1 0.10% 1 Tracked
House of CB Foreign 1 0.10% 1 Tracked
Işıltı Abiye TR 1 0.10% 1 Tracked
Jeep Foreign 1 0.10% 1 Tracked
KIKIRIKI TR 1 0.10% 1 Tracked
Kadriye Baştürk TR 1 0.10% 1 Tracked
Lady Special TR 1 0.10% 1 Tracked
Le Mabelle TR 1 0.10% 1 Tracked
Lee Cooper Foreign 1 0.10% 1 Tracked
Los Banditos TR 1 0.10% 1 Tracked
Macy's Foreign 1 0.10% 1 Tracked
Moda Babil TR 1 0.10% 1 Tracked
Modasena TR 1 0.10% 1 Tracked
Mudo TR 1 0.10% 1 Tracked
My Rental Dress TR 1 0.10% 1 Tracked
NAZRA Unknown 1 0.10% 1 Tracked
Nordstrom Foreign 1 0.10% 1 Tracked
Panço TR 1 0.10% 1 Tracked
SAMARA Unknown 1 0.10% 1 Tracked
Siren Ertan TR 1 0.10% 1 Tracked
Styling Park TR 1 0.10% 1 Tracked
TOM FORD Foreign 1 0.10% 1 Tracked
Versace Foreign 1 0.10% 1 Tracked
Victoria Beckham Foreign 1 0.10% 1 Tracked
Zülays TR 1 0.10% 1 Tracked
klodsy TR 1 0.10% 1 Tracked
letgo TR 1 0.10% 1 Tracked
All 393 tracked entities, general cut (999 responses, 2 October 2026).

12.3 Data availability

The data behind this report is available on request: per-entity metrics across all six cuts, the qualified leaderboards and their brand-only variants, the entity registry with origins, the per-model breakdowns, the origin-share tables on both bases, the source attribution tables, the raw-layer trend file, and the published scale registry.

All figures reflect a single collection day (2 October 2026) and are subject to the limitations in Chapter 11. This is the second edition; the study is repeated quarterly.

Methodology note
Five models (Claude Haiku 4.5, Gemini 3.6 Flash, ChatGPT 5.6 Luna Pro, Perplexity Sonar, Grok 4.3), each asked to search the web, answered 40 Turkish questions up to five times each: 999 responses, collected on 2 October 2026. A mention is a brand or company that a second language model lists as named in the answer; a name that appears only inside a web address is not counted. AI Visibility Score = 0.45 × Mention + 0.3 × Position + 0.25 × Breadth, each component scaled to the highest value among qualified entities in the cut; an entity qualifies at a 5% mention rate within the cut. Origin is the brand home. Origin shares are computed over qualified entities with origin on file, with marketplaces and publishers left out, and are always printed with their coverage. The comparison with July uses raw mention rate on the general cut, with July's answers counted the way this edition counts; two of five models differ between editions. Herm.io's own domain is excluded from the source data before analysis.

Frequently asked questions

Across all 40 questions, the marketplace Trendyol leads with a 100.00 score, named in 42.44% of 999 responses and by all five assistants. Carmen (93.77) is second, named in 38.44% of responses, and The ZKS, ADL and Abiyefon complete the top five. Carmen leads the 20 questions that ask for Turkish brands.

Mostly Turkish. On the 20 questions that ask for Turkish brands, domestic entities take 100.00% of qualifying mentions and foreign entities 0.00%. On the 20 origin-neutral questions they take 83.21% and 16.79%, and five foreign brands qualify: Mango, H&M, Zara, Oleg Cassini and Pronovias. Origin is on file for every qualifying mention on both.

393 entities were tracked, including retailers and marketplaces. 38 cleared the 5% mention threshold on the full question set — 35 domestic and 3 foreign — with an average score of 52.64, and 65 qualify in at least one cut.

It combines how often an entity is named (45%), how early it appears in the answer (30%), and how many of the five models name it (25%). Each part is scaled against the leading qualified entity within that cut, so scores are never compared across cuts; across editions, only the general board's change since July is shown, against July re-scored the same way. Only entities named in at least 5% of a cut's responses are ranked.

On the leaders, largely: Carmen is in every model's top three, and Trendyol, Carmen, ADL and AlfaBeta are in every model's top ten. On length, no: Gemini 3.6 Flash names 12.38 entities per answer and Claude Haiku 4.5 names 4.44. Some ranks lean on one model: ChatGPT 5.6 Luna Pro supplies all 50 of Berneva's mentions, and Grok 4.3 supplies 111 of Hepsiburada's 178.

The 40 questions are the same, but two of the five models were replaced, so changes show direction more reliably than size. Measured by raw mention rate against July's answers counted the way this edition counts, The ZKS gained 19.01 points and Iconas Line lost 8.54. Origin shares are not compared with July's, because the origin rule changed.

No. The score measures only how often and how prominently a brand is named across the assistants. It does not assess quality, fit, price, accuracy or sentiment, so a score is a measure of visibility, not an endorsement.


About Herm.io & disclosure

Herm.io is a consumer behaviour and marketing data company. We study how people discover and choose brands to help businesses reach the right customers. This report forms part of our public research and is conducted quarterly.

Disclosure & neutrality

The rankings and metrics in this report are entirely impartial and are based on the methodology described within the report. No brand can pay or provide sponsorship to be included in the report, improve its ranking or influence how it is described in the content.

A brand’s score in the report is not an endorsement or assessment of quality; it is solely a measure of its current visibility within AI models. No service or product offered by Herm.io is sold for the purpose of directly changing the results of this report or guaranteeing a ranking. To preserve objectivity, all references to Herm.io’s own domain are removed from the source data before analysis. This ensures that the company does not appear in, measure itself through or benefit from its own study.

Brands seeking to understand their position in the data can schedule a consultation for an impartial assessment of the findings. The consultation is advisory and free of charge.


Mert Can Elkaya
Written by

Mert Can Elkaya

Contributor

I'm a product builder working at the intersection of product, fintech, and growth. From martech and venture capital to leading product at a proptech platform and co-founding a fintech startup, I help teams—and shoppers—make smarter, more confident decisions.

More from Mert