1,500 responses, 5 models, 3 segments: the first AI-visibility snapshot of Turkey's ready-to-wear apparel market

Vertical: Ready-to-wear apparel (hazır giyim), Turkey — women’s, men’s and kids’ & baby wear
Method: A single point-in-time study of 1,500 responses across five large language models (GPT-4o-mini, Gemini, Claude, Perplexity, Grok), each asked 60 Turkish-language questions five times
Collection dates: 10–11 July 2026; 16 revised questions re-run 14 July 2026

Key terms: AI visibility, AI Visibility Score, ready-to-wear brands, hazır giyim, Turkish (yerli) apparel brands, women’s / men’s / kids’ wear, 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 clothing in Turkey, and measures how often, in what order, and across how many models five large language models name each brand. Ready-to-wear is not one market but three: the questions are split evenly across women’s, men’s and kids’ & baby wear, and each segment is ranked separately, alongside three overall cuts 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 clothing 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, cheaper, better-made or better-fitting. 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: Turkey’s ready-to-wear shelf has no single winner: Koton leads women’s wear, Kiğılı leads men’s and LC Waikiki leads kids’, while the marketplace Trendyol edges the general cut by a tenth of a point, and foreign brands hold real ground only in menswear.

• No brand wins the market. Each segment has a different leader: Koton in women’s wear (88.5), Kiğılı in men’s (91.5) and LC Waikiki in kids’ (92.7). Of the 80 entities that qualify in at least one segment, only 8 qualify in all three, and 63 qualify in exactly one. Whoever “leads Turkish apparel” depends entirely on which shopper is asking.
• The general cut is a photo finish. Across all 60 questions, the marketplace Trendyol leads on 85.8, appearing in 31.1% of all responses, just 0.1 points ahead of LC Waikiki (85.7). Koton (82.9), DeFacto (80.6) and Mavi (79.0) complete a tightly packed top five. Only 28 of the 730 tracked brands clear the 5% threshold across all 1,500 responses.
• Two entities on the headline board are marketplaces. Trendyol and Hepsiburada (plus the retailers Beymen and Boyner) are kept in the ranking because that is how shoppers meet the shelf. Remove them and the brand-only top four is LC Waikiki, Koton, DeFacto, Mavi, with Kiğılı fifth.
• Foreign brands live in menswear. They take 14 of 39 qualified places in men’s wear, against 9 of 32 in kids’ and only 5 of 34 in women’s. In the origin-neutral open cut, mentions split 70.2% Turkish / 29.8% foreign among qualified brands, and the first foreign name is Marks & Spencer at #8.
• The local shelf is completely Turkish. On the 30 questions that explicitly ask for yerli brands, all 28 qualified brands are Turkish. Not one foreign name survives the filter, which shows the assistants treat “yerli” as a real constraint rather than a stylistic hint.
• The models do not agree on a leader. Koton tops Perplexity (38.3% of its answers), Gemini (30.7%) and GPT-4o-mini (15.3%); Mavi tops Claude Haiku (18.3%); Trendyol tops Grok, which names it in 85% of its answers. Breadth differs just as sharply: Grok lists 11.74 brands per answer, GPT-4o-mini 4.42.
• Search was near-universal. 97.0% of answers were web-grounded; only Gemini fell below (88.3%). With almost no memory-only segment, this edition does not include a search-vs-memory comparison (Chapter 6).
• Sources are brand-owned first. 62.9% of cited sources are brands’ own websites, across 850 distinct domains. The single most-cited domain is trendyol.com (cited by 27.3% of responses), followed by two more marketplaces and then a run of brand sites.

Why it matters. Shoppers are shifting discovery from search engines and marketplaces to AI assistants. Whether an apparel brand appears in those answers is becoming a discovery channel in its own right. This report captures the first snapshot (baseline) of that shelf for Turkish ready-to-wear. One reminder: the numbers below describe visibility and information availability, not which brands are best.

---

2. Methodology

One-sentence takeaway: Five models were asked 60 Turkish questions five times each, producing 1,500 responses across three segments; brands were ranked with a three-component (45/30/25) score, after two matching artifacts were found and removed.

2.1 Scope and models

The study queried five large language models directly via API with identical questions. All models received only the user question. No system prompt was used, temperature was not set manually (provider default), and no region or locale parameter was defined. Because the questions are in Turkish, the models infer the Turkish market from language alone.

Scale: 60 questions × 5 repeats × 5 models = 1,500 responses, all completed, exactly 5 per model-question pair. No responses were excluded. Final brand dictionary: 730 entities (479 Turkish, 217 foreign, 34 unknown origin).

2.2 Web-search configuration

Web search was offered as an enabled tool to all five models; each decided for itself whether to use it (Perplexity’s Sonar is search-grounded by design). Whether a response drew on web search or the model’s own knowledge was inferred from whether the provider returned any citation or source: a citation means “web-search,” none means “own knowledge.” This is a reasonable proxy, not direct proof of the model’s internal process. In this run, 97.0% of answers were grounded (see Chapter 6).

2.3 Questions, segments and cuts

The 60 questions reproduce the real Turkish-language queries people bring to an assistant about clothing: open discovery (“kadın giyim markaları nelerdir?”), specific attributes (a sweater that does not pill, a linen shirt that breathes, a print that does not crack), and use cases (a first office job, a wedding suit, a child starting nursery, a bayram outfit).

They are split into three segments of 20 questions each, and within every segment into 10 local questions (which explicitly ask for Turkish / yerli brands) and 10 open questions (no origin restriction). Every question names its audience explicitly, so no answer has to guess who is being dressed.

Brands are then ranked in six cuts of the same 1,500 responses: one per segment, and three overall.

The Turkish-versus-foreign comparison in Chapter 8 rests solely on the 30 open questions, the only fair basis for the comparison, since the local questions cannot surface foreign brands by construction.

2.4 What we corrected before ranking

Two artifacts were found by audit and removed before any score was computed. Both are documented here because either would have produced confidently wrong rankings.

(a) The audience fix. In the first pass, 16 of the 60 prompts never named an audience. “Tüylenmeyen kazak yapan yerli marka var mı?” (“is there a domestic brand that makes a non-pilling sweater?”) was filed under women’s wear, but the text says nothing about women. The models answered for an unspecified adult shopper and frequently defaulted to menswear, which placed the menswear brand Altınyıldız Classics at #12 on the women’s leaderboard. Those 16 questions were reworded to state the audience explicitly and re-run on 14 July 2026 (400 fresh responses); the other 44 questions and their 1,100 responses are unchanged.

The fix is verifiable in the text of the answers themselves. In the linen-shirt question, the original answers used the word erkek (man) 43 times and kadın (woman) 5; the reworded answers use erkek 0 times and kadın 50. Altınyıldız Classics fell from 7.0% to 1.8% mentions in the women’s cut, where it no longer qualifies, and now sits at #4 in men’s, which is where it belongs.

(b) A style term wearing a brand’s name. Brand matching in fashion has a specific hazard: style labels and brand names overlap. Old money is a widely used aesthetic term in Turkish fashion writing, and it is also part of several real brand names. Question 27 asks for “yerli old money tarzı erkek giyim” — so every answer echoes the phrase back, usually in its opening heading. A bare “Old Money” brand entity matched all 26 of these, and none of them was a genuine brand reference: 96% sat at rank 1 because they appeared in a title, giving the entity an MRR of 0.98 against 0.55–0.58 for real brands, the tell-tale signature of a phrase being matched rather than a recommendation.

It was removed. The three genuine brands — Old Money Türkiye, Atelier Old Money and Prime Old Money, all Turkish — are unaffected, because longest-match-first matching stops the bare phrase absorbing them. The artifact was also quietly distorting every other brand in the men’s cut: since each component is rescaled to the maximum among qualified brands, its inflated MRR was the denominator for everyone else’s position score. Removing it raised legitimate scores, with Hatemoğlu going from 76.0 to 81.5.

The audit that found these is repeatable, and we now run it on every vertical: check whether any brand name appears in a question’s own text; check whether any qualified brand has an MRR above roughly 0.80, since a real brand is rarely always-first; and check whether a brand’s mentions all come from a single question.

2.5 Brand extraction and matching

Brands were extracted in two stages. (1) Automated extraction: emphasised brand names and citation domains were collected, and spelling and format variants merged into a single canonical name. (2) Human validation: the candidate list was reviewed by hand before any metric was computed, with real brands confirmed, non-brands (category words, fabrics, places, style terms) filtered out, aliases merged, and each brand labelled Turkish, foreign or unknown. Final dictionary: 730 entities.

Matching is binary per response (presence, not frequency), word-boundary aware, diacritic-folded and longest-match-first, which stops a short name being absorbed into a longer one (“Gusto” inside “Little Gusto”, “Beymen” inside “Beymen Club”). Around 25 brands are ordinary Turkish or English words — Only, Next, GAP, Roman, Civil, Network, Mudo, Tween — and are matched case-sensitively, so the Turkish noun roman (novel) is never counted as the brand Roman.

Origin rule. A brand is Turkish if founded and headquartered in Turkey, and foreign otherwise, even when it is widely sold here (Zara, H&M, Mango). A .com.tr domain is not evidence of Turkish origin.

Entity scope. Marketplaces (Trendyol, Hepsiburada) and retailers (Beymen, Boyner) are kept as entities so the shelf is described as shoppers actually meet it; a brand-only view is given in §3.2. Vertically integrated chains with their own labels and stores (LC Waikiki, Koton, DeFacto, Mavi) are treated as brands.

Definitions.

• Mention: a validated brand appears in the answer text, counted once per response.
• Position (MRR): mean reciprocal of the brand’s first-appearance rank within each answer (first = 1.0; second = 0.5; …), averaged only over the responses where the brand appears.
• Breadth: number of the 5 models (0–5) that mention the brand within that cut.

2.6 AI Visibility Score (0–100)

Score = 0.45 × Mention + 0.30 × Position + 0.25 × Breadth

Each component is scaled so the leading qualified brand in each cut = 100, then combined with the weights. Qualification (≥5% rule): only brands mentioned in at least 5% of that cut’s responses are ranked. One caveat travels with that rule in this study: a segment has 500 responses, so a single question (25 responses) is exactly 5%. A niche brand named in every answer to one question therefore qualifies on the strength of that question alone. Three brands sit in this position — Zülays, SIRRI and Supreme — and are flagged where they appear. Unfiltered metrics for all 730 brands are in Chapter 12.

2.7 Neutrality and self-exclusion

This is a market-wide, neutral study with no focus brand. To prevent any conflict of interest, all citations to Herm.io’s own domains were excluded from the source data before analysis, so the company’s own content neither appears among the most-cited domains nor influences any figure.

---

3. The General Leaderboard

One-sentence takeaway: Across all 60 questions the marketplace Trendyol leads by a tenth of a point over LC Waikiki, on a shelf so flat that the entire qualified field spans 45 points.

As a reminder, these rank visibility, not product quality, fit, price or reliability.

3.1 General cut: 28 qualified brands (all 1,500 responses)

Average score of the 28 qualified brands: 57.7. 24 are Turkish, 4 foreign.

AI Visibility Score: qualified brands (general cut)

Reading. The top of this table is a photo finish: Trendyol (85.8, named in 31.1% of all answers) leads LC Waikiki (85.7) by a tenth of a point, with Koton (82.9), DeFacto (80.6) and Mavi (79.0) close behind. These five are the general-purpose names the assistants reach for whatever the question. The largest single gap in the table falls immediately after them — 11.6 points between Mavi and Kiğılı (67.4) — which is where the general shelf stops being general and starts being specialist: Kiğılı and Hatemoğlu (#11) are classic menswear, Little Gusto (#7) and The Nest Kids (#10) are children’s wear, Vakko (#8) is premium.

Two things stand out. First, breadth is nearly universal here: 27 of the 28 qualified brands are named by all five models, so at this level of the market the models differ on ordering, not on who exists. Second, only four foreign brands qualify at all — Zara (#17), H&M (#19), Nike (#20) and GAP (#27) — and none of them reaches the top 15.

3.2 Brand-only view (marketplaces and retailers removed)

Four entities in the table above are not brands: the marketplaces Trendyol and Hepsiburada, and the retailers Beymen and Boyner. They are kept in the headline ranking because that is how shoppers meet the shelf, but removing them gives a pure brand-versus-brand picture:

With the aggregators set aside, the market’s four general-purpose Turkish chains — LC Waikiki, Koton, DeFacto and Mavi — take the top four places, all within 6.7 points of one another, followed by Kiğılı, Little Gusto and Vakko. Unlike verticals where no standalone brand can outrank the marketplaces, here the top brand is effectively level with Trendyol.

3.3 The local shelf: 28 qualified brands, all Turkish

When the 30 questions that explicitly ask for Turkish / yerli brands are isolated (750 responses), the shelf is entirely domestic: all 28 qualified brands are Turkish. Not one foreign name survives the filter.

Average score of the 28 qualified local brands: 60.3. All 28 are Turkish.

Reading. The local cut reorders the market. Kiğılı (84.1) leads, not Trendyol (83.0), and Hatemoğlu (#6), Vakko (#7) and The Nest Kids (#8) all climb: when a shopper asks for a domestic brand, the assistants reach past the general-purpose chains toward specialists.

The clean absence of foreign names is worth pausing on. In the June furniture edition, IKEA still qualified at #17 in a cut whose questions explicitly asked for Turkish brands — evidence that models do not always honour an origin instruction. Here there is no such leakage at all: in apparel, the assistants treat yerli as a hard constraint rather than a stylistic hint. Note also that breadth thins in this cut: only 19 of the 28 qualified brands are seen by all five models, against 27 of 28 in the general cut, so domestic specialist visibility is more model-dependent than the headline names.

---

4. The Three Segments

One-sentence takeaway: Ready-to-wear is not one market: Koton leads women’s, Kiğılı men’s and LC Waikiki kids’, and only 8 of the 80 qualifying entities appear in all three.

Each segment is 20 questions and 500 responses, ranked independently. This is the level at which the market actually behaves, and the three leaderboards look very different from one another.

4.1 Women’s Apparel: 34 qualified brands

Average score of the 34 qualified brands: 55.1. 29 are Turkish, 5 foreign.

Reading. Koton leads on 88.5, named in 35.4% of women’s-wear answers, with Trendyol (86.3) second. Third place goes to Vakko (83.9), and it gets there differently: it is named in 23.4% of answers, well behind Koton, but very early when it is named (MRR 0.5606, against 0.33–0.38 for the brands around it). That is the signature of a brand the models treat as a definitive answer rather than one item on a list. The same pattern, more sharply, produces adL at #8 (7.4% mentions, MRR 0.5776) and Oxxo at #11.

This is the least foreign-contested segment in the study: just 5 of 34 qualified brands are foreign, and the highest-placed, Levi’s, sits at #9 on a thin 5.8% mention rate. Zara does not appear until #24.

Read with care: Zülays (#16) draws 25 of its 26 mentions from the single tesettür (modest wear) question. It is a real brand genuinely dominating a niche, but its rank rests on one question and should not be read as broad visibility across women’s wear (see §2.6 and Chapter 11).

4.2 Men’s Apparel: 39 qualified brands

Average score of the 39 qualified brands: 52.4. 25 are Turkish, 14 foreign.

Reading. Men’s wear is the most crowded segment (39 qualified brands) and the only one with a decisive leader: Kiğılı (91.5), named in 37.4% of answers and named early (MRR 0.5182), a full 10 points clear of Hatemoğlu (81.5). Behind them sits an unusually coherent cluster — Altınyıldız Classics (#4), Sarar (#9), D’S Damat (#11), Süvari (#18) — all classic tailoring and shirting. When the question is about menswear, the assistants think first about suits.

It is also the most foreign-contested segment: 14 of 39 qualified brands are foreign, against 5 of 34 in women’s. This is where Nike, Uniqlo, Adidas, GAP, Carhartt WIP, Levi’s and Jack & Jones actually land in the Turkish market.

Read with care: Supreme (#10) carries the highest position score in the entire study (MRR 0.7237) on a 5.2% mention rate, essentially all of it from the single streetwear question. When Supreme is named it is named first — but it is named in one context only, and #10 overstates its presence in menswear as a whole.

4.3 Kids & Baby: 32 qualified brands

Average score of the 32 qualified brands: 54.0. 23 are Turkish, 9 foreign.

Reading. LC Waikiki dominates (92.7), named in 39.0% of answers — the strongest mention rate anywhere in the study. But the two brands behind it are the interesting ones: Little Gusto (80.0) and The Nest Kids (78.6) are specialists, not chains, and both rank on position as much as volume (MRR 0.638 and 0.5722). Kids’ wear is the segment where the assistants most readily leave the big chains for a dedicated label.

A second cluster follows the same logic: Organickid (#12), Kitikate (#14), Biorganic (#18) and TinyLamb (#17) are all organic-cotton specialists, and they surface because the questions ask about sensitive skin, organic fabric and newborns. Foreign presence is mid-range (9 of 32) and mostly babywear-specific: Carter’s (#13), Chicco (#21), U.S. Polo Assn (#20).

Read with care: SIRRI (#10) takes 24 of its 25 mentions from the boys’ wedding-suit question. It also carries the study’s one unresolved origin conflict: SIRRI validated as foreign while “Sirri Giyim” validated as Turkish, and the merged entity carries Foreign (Chapter 11).

4.4 What crosses segments, and what does not

Eighty distinct entities qualify in at least one segment. Only 8 qualify in all three; 63 qualify in exactly one.

The eight are the four general-purpose Turkish chains (LC Waikiki, Koton, DeFacto, Mavi), the two marketplaces (Trendyol, Hepsiburada), and the only two foreign brands broad enough to cross every segment (Zara, H&M) — both of which qualify near the bottom of each table. Nothing else spans the market.

The practical reading is that AI visibility in apparel is earned per audience. Kiğılı is #1 in men’s and absent from the women’s and kids’ boards entirely. Vakko is #3 in women’s and #15 in men’s. Little Gusto is #2 in kids’ and nowhere else. A brand that is invisible in one segment is not necessarily invisible in the market; it is answering a different question.

---

5. Differences Between Models

One-sentence takeaway: The five models do not even agree on who leads — Koton tops three of them, Mavi tops Claude and Trendyol tops Grok — and they differ by a factor of nearly three in how many brands they name per answer.

5.1 Per-model behaviour summary

Web search was near-universal: four of the five models grounded 96.7–100% of their answers, and only Gemini fell below, at 88.3%. The differences that matter are elsewhere. Grok names 11.74 brands per answer, GPT-4o-mini 4.42 — an answer from one model is a shortlist, an answer from the other is a catalogue. Gemini has the widest vocabulary overall (462 distinct brands across its 300 answers), while GPT-4o-mini uses the narrowest (254).

Distinct brands named per answer, by model

5.2 Each model’s most-named brands

Unlike previous Herm.io editions, where one entity topped every model, there is no consensus leader here. Koton is the most-named brand for Perplexity (38.3%), Gemini (30.7%) and GPT-4o-mini (15.3%). Claude Haiku puts Mavi first (18.3%). Grok puts Trendyol first and names it in 85% of its answers — while Claude Haiku names Trendyol in only 8.7%.

That single contrast is the most important model finding in the report. A brand’s measured visibility depends heavily on which assistant the shopper happens to use, and the marketplace-versus-brand balance of an answer is close to a house style: Grok reaches for Trendyol and Hepsiburada (179 mentions) reflexively, while Claude and GPT-4o-mini prefer to name the label directly.

---

6. Search vs. Memory: Why This Edition Omits the Split

One-sentence takeaway: Almost every answer in this run used web search, so there is no meaningful “memory-only” segment to compare against, which is itself a finding.

Some earlier Herm.io editions split answers into web-grounded and memory-only segments that produced different leaders. That split does not exist in this run. 97.0% of the 1,500 answers returned at least one citation or search result. Four of the five models grounded 96.7–100% of the time, and only Gemini fell below, at 88.3%.

A separate chapter would therefore compare roughly 1,455 grounded answers against about 45 ungrounded ones, which is not a meaningful contrast, so it is left out this edition. The takeaway is that for Turkish apparel questions these assistants reach for live web content almost every time. That makes a current, well-structured web presence especially closely tied to visibility here (see Chapter 9). We will keep measuring the grounding rate each edition; if a memory-only segment grows, the comparison will return.

---

7. Question-Type Ownership

One-sentence takeaway: Who owns a segment depends on what is being asked: discovery questions surface specialists, use-case questions surface marketplaces, and attribute questions sit in between.

Each segment’s 20 questions are split across three behavioural types: discovery (open “which brands / recommend one”), attribute (a specific property: non-pilling, non-creasing, organic cotton, a fit that holds), and use_case (an occasion or need: a wedding, a first office job, nursery, a growth spurt). The tables give the five most visible brands within each type; rate is relative to that type’s responses.

Reading. The pattern repeats across all three segments and is the clearest structural finding in the report.

Discovery questions surface specialists. Asked simply “which women’s brands should I look at?”, the assistants lead with İpekyol (54.3%) and Vakko, not with the chains. In men’s, discovery is owned by Hatemoğlu (60.6%) and Kiğılı. In kids’, by LC Waikiki, Civil and Panço. Trendyol appears in none of the three discovery top fives.

Use-case questions surface marketplaces. The moment a question describes a situation — a wedding, a nursery, a first office job — Trendyol tops all three segments (women’s 42.0%, men’s 40.6%, kids’ 50.3%), with Hepsiburada joining it in kids’ (37.1%). When the assistant has to solve a problem rather than name a brand, it sends the shopper somewhere with everything in stock.

Attribute questions are the contested middle. Here the brands with a technical story hold on: Mavi and Altınyıldız in men’s, Koton and Mavi in women’s, and in kids’ the organic specialists Little Gusto (37.7%) and Organickid (29.7%) lead outright, ahead of any chain or marketplace. If a brand has a fabric or fit claim the models can attach to a question, this is where it earns its visibility.

---

8. Open Market: Turkish vs. Foreign

One-sentence takeaway: On origin-neutral questions, 70.2% of qualified-brand mentions go to Turkish brands and 29.8% to foreign, with foreign visibility concentrated in menswear rather than spread across the market.

This chapter relies only on the 30 open questions with no origin restriction (750 responses), the only fair basis for comparing Turkish and foreign brands.

8.1 Origin share

Among the 35 brands that qualify in the open cut, Turkish names take 2,316 mentions to foreign brands’ 985 — a 70.2 / 29.8 split. Turkish brands lead clearly, but this is a more contested shelf than evening wear, the one prior edition measured on the same basis, where the split was 82.4 / 17.6. Counting every brand that surfaced at least once, ranked or not, foreign presence is larger still (61.6 / 38.4, across 292 Turkish and 203 foreign distinct names), which tells you the foreign long tail here is deep even where it is thin at the top.

8.2 Open-market top 15 brands (any origin)

8.3 The pattern: foreign brands hold the menswear shelf, not the market

The aggregate hides where the contest actually is. In the open cut the top seven places are Turkish (LC Waikiki, Koton, Trendyol, DeFacto, Mavi, Hepsiburada, Vakko), and the first foreign name is Marks & Spencer at #8 — named in only 8.5% of open answers but named early (MRR 0.5081), the classic profile of a brand the models treat as a specific recommendation rather than a default. Zara (#10) and H&M (#12) follow on the opposite profile: named often, named late.

The segment tables in Chapter 4 explain the split. Foreign brands take 14 of 39 qualified places in men’s wear but only 5 of 34 in women’s. Turkish women’s-wear brands — Koton, İpekyol, Vakko, Machka, Mudo, Network — occupy the space that Zara, Mango and H&M might otherwise hold, whereas in menswear the assistants readily reach for Nike, Uniqlo, Adidas, GAP, Levi’s and Carhartt WIP alongside Kiğılı and Hatemoğlu. Whether that asymmetry narrows is exactly the kind of movement a quarterly baseline is built to track.

---

9. The Discovery Ecosystem: Where AI Learns About Brands

One-sentence takeaway: AI’s picture of this market rests first on brands’ own websites (62.9% of cited sources), with marketplaces supplying the single most-cited domain and editorial content playing a minor role.

When an assistant answers with web search, it draws on the content available to it at that moment. We collected the source addresses returned by the models, reduced them to domains, and ranked them by how many distinct answers cited each. Across the run the models cited 850 distinct domains.

9.1 The 12 most-cited domains

The most-cited domain is the marketplace trendyol.com (cited by 27.3% of responses), followed by two more marketplaces (ciceksepeti.com, hepsiburada.com) and then a long run of brand-owned sites: lcw.com, koton.com, defacto.com.tr, hatemoglu.com, thenestkids.com.tr, kigili.com. Two editorial domains break in (bifikir.com at #10, oggusto.com at #11), but they are the exception.

There is a direct line between this table and the leaderboards. The brand sites cited most often — LC Waikiki, Koton, DeFacto, Hatemoğlu, The Nest Kids, Kiğılı — are, with almost no exceptions, the brands that top the cuts they compete in.

9.2 Source-type mix

Citation share by source type (%)

9.3 The brand-owned finding

The mix is dominated by official sites: 62.9% of citations point to brands’ own domains, from 385 distinct brand sites, with marketplaces adding another 10.5%. Editorial and media (5.4%), forums (1.9%) and social (1.8%) are marginal, and 17.4% sits in a long tail of 381 miscellaneous domains. Read as a description rather than advice: in this market, a current, machine-readable brand site — one the models can actually read a product range off — appears far more closely associated with AI visibility than press coverage or reviews.

---

10. What the Patterns Suggest

One-sentence takeaway: Visibility here is earned per audience and per question type, on the back of a readable own-site presence and coverage across all five models.

This chapter is a neutral reading of the data. It describes patterns associated with visibility; it is not advice, a service, or a recommendation, and visibility remains separate from quality.

1) There is no “Turkish apparel” shelf — there are three. Only 8 of 80 qualifying entities appear in all three segments, and 63 appear in exactly one. A brand’s rank in the general cut mostly reflects how many audiences it serves, not how strongly it serves any of them. Kiğılı is #1 in men’s and #6 overall; that is not a demotion, it is what a specialist looks like when you average it across shoppers it does not dress.

2) Mention and position are different signals. Some brands are named often but late (Trendyol, Hepsiburada, Network: high mentions, low MRR); others are named rarely but first (Supreme 0.72, Marks & Spencer 0.67, Cemi Kids 0.66, adL 0.58). A brand can clear the bar on position and breadth alone. The strongest position scores in the study belong to brands the models treat as the answer to a narrow question — which is a real form of visibility, but a narrow one.

3) Question type reorders the market more than segment does. The same segment produces different winners depending on whether the shopper is browsing (discovery → specialists), solving a problem (use-case → marketplaces) or checking a property (attribute → brands with a fabric story). A brand that wants to be found should know which of the three it is currently winning.

4) Breadth is a baseline condition at the top, and thins below it. 27 of the 28 general-cut brands are named by all five models, but in the local cut only 19 of 28 are. Being seen by every assistant is table stakes for the leaders and a real differentiator further down: three- or four-model brands (Farren, Mehry Mu, Piece of White, Lufian) sit low largely for this reason.

5) The web presence is the product description. With 97.0% of answers grounded and 62.9% of citations pointing at brand-owned domains, the assistants are, in effect, reading brands’ own sites aloud. The brands with the most-cited sites are the brands that lead the tables.

How a brand can locate itself in the data. Using the dataset (available on request), a brand can read four things in order: whether it appears in all five models; how often it is mentioned in its own segment rather than the general cut; how early it is named when it is named; and which question type it is weakest in. These readings show where a visibility gap sits, without saying anything about whether its product is good.

---

11. Limitations and Notes

One-sentence takeaway: This is a single, point-in-time snapshot of one Turkish vertical; read the figures within the limitations below.

• The 5% floor equals exactly one question. A segment has 500 responses, so one question (25 responses) is precisely the 5% qualification threshold. A brand named in every answer to a single question therefore qualifies on that question alone. Three brands sit in this position: Zülays (25 of 26 mentions from the tesettür question), SIRRI (24 of 25 from the boys’ wedding-suit question) and Supreme (streetwear). They are real brands genuinely dominating a niche — not artifacts — but their ranks rest on one question and are not broad market visibility.
• Mixed collection window. 1,100 responses were collected on 10–11 July 2026 and 400 on 14 July 2026 (the re-run of the 16 revised questions). Model versions and web indexes may have shifted slightly between the two dates.
• Model verbosity inflates mention counts. Grok names 11.74 brands per answer against GPT-4o-mini’s 4.42. The breadth component partly offsets this, but a brand favoured by a verbose model retains a structural advantage.
• 19 new brands are unvalidated. The re-run surfaced small labels (Gençwear, LOUX, Wheraver, Jihangir, Zeyder, MegaButik and others) whose origin is marked Unknown. All sit far below the qualification floor, so no ranking is affected, but they are counted in the 730 tracked brands.
• One unresolved origin conflict. SIRRI validated as Foreign and “Sirri Giyim” as Turkish; the merged entity carries Foreign, which affects the foreign count in the kids’ segment by one.
• Small score gaps are not meaningful. This is a flat market — Trendyol leads LC Waikiki by 0.1 points — and differences of a point or two in the packed middle of any leaderboard should not be over-interpreted.
• Single, point-in-time measurement. A snapshot, not a trend. This is the first (baseline) edition; quarterly editions will enable trend analysis.
• Visibility is not quality. This report measures only mention and position. It does not assess whether a brand was described positively, whether a recommendation was accurate, or whether a product is good. A score is not an endorsement.
• Models are probabilistic. The same question can produce different answers; five repeats reduce but do not eliminate this.
• Web search depends on momentary conditions, meaning the content indexed and served at measurement time. “Web search vs. own knowledge” is a proxy based on whether a citation accompanied the answer.
• Precision over recall. Brand matching was cautious to avoid false positives (see §2.4 and §2.5), so brands mentioned indirectly may be undercounted. There is no sentiment layer; the study is strictly quantitative.
• Source-type mix is heuristic. Domain types were auto-classified; treat the mix as indicative. Source coverage also varies by model.
• Marketplaces and retailers are included as entities per study scope (Trendyol, Hepsiburada, Beymen, Boyner); they compete differently from direct brands, so §3.2 gives a brand-only view.
• Output-length limits and model versions differ and date quickly; findings are specific to the versions in Chapter 2 as of July 2026.

---

Frequently Asked Questions

---

12. Appendix & Data

12.1 All questions (60), their segments and categories

12.2 All tracked brands (730)

12.3 Data availability

The study tracked 730 brands (479 Turkish, 217 foreign, 34 unknown origin); 28 cleared the 5% threshold in the general cut, with 34 qualifying in women’s, 39 in men’s, 32 in kids’, 28 in the local cut and 35 in the open cut. To support scrutiny and reproduction, the underlying data is available on request: response-level brand mentions, all six qualified leaderboards, full unfiltered brand metrics, per-model and per-question-type breakdowns, the open-market breakdown, the top source domains, the question-revision log, and the methodology and data dictionary.

All figures reflect a point-in-time run (10–11 July 2026, with the 16 revised questions re-run on 14 July 2026) and are subject to the limitations in Chapter 11. This is the first (baseline) measurement; the study will be repeated quarterly.

Want the deeper cut?

The full domain-level source breakdown, per-model leaderboards for each segment, and the question-by-question brand visibility are not reproduced in the public report. To request them, or to be notified when the next quarterly edition is published, book a call with the Herm.io team. There is no fee and nothing to buy; it is part of how we share what we learn.

12.4 About Herm.io & disclosure

Herm.io is a consumer-behavior and marketing-data company. We study how people discover and choose brands so that brands can reach the right customers. This report is part of our public research and is a recurring quarterly study.

Disclosure & neutrality. Herm.io does not sell SEO or GEO (search/AI-ranking) services, and this report does not recommend any. No brand paid to be included, ranked, or described, and a brand’s score is not an endorsement or a judgment of its quality; it is a measure of visibility only. To keep the analysis objective, all citations to Herm.io’s own domains were excluded from the source data. Brands that want to understand their position in the data are welcome to book a call for a neutral walkthrough; this is advisory and free.