1,116 responses, 5 models, 45 Turkish questions: the baseline AI-visibility snapshot of Turkey's sportswear & activewear market

Vertical: Sportswear & activewear (spor giyim / aktif giyim), Turkey
Method: A single point-in-time study of 1,116 responses across five large language models (GPT-4o-mini, Gemini, Claude, Perplexity, Grok), each asked 45 Turkish-language questions five times
Collection window: 18–19 July 2026

Key terms: AI visibility, AI Visibility Score, sportswear brands, activewear brands, spor giyim / aktif giyim, athleisure, Turkish (yerli) sportswear 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 sportswear, activewear and athleisure in Turkey, and measures how often, in what order, and across how many models five large language models name each brand. Because the same questions can be read with or without an origin restriction, and because sportswear divides into genuinely different activity contexts, the entities are ranked in ten separate 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 sportswear and activewear 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 technical, 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 performance, fit, durability, 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: AI visibility in Turkey’s sportswear market is led by Nike, and the origin-neutral shelf is decisively foreign — but the moment a question asks for Turkish brands, the shelf reverses almost exactly, and each activity segment turns out to have a different leader.

• Nike leads; Adidas and Lescon follow. Nike tops the general leaderboard at 58.61, appearing in 34.95% of all 1,116 responses (390 of them) and named by all five models. Adidas (50.41) and Lescon (49.74) follow, with Decathlon (49.19) and XSIDE (49.07) completing the top five once the position-driven entry at #2 is set aside (see below).
• 30 entities qualified for the general ranking. Of 214 tracked entities, 30 cleared the 5% mention threshold across all 45 questions: 16 foreign and 14 Turkish. Their average score is 41.3, and 29 of the 30 are seen by all five models.
• The origin-neutral shelf is foreign-led. On the 23 questions that place no origin restriction, mentions split 81.5% foreign to 18.5% Turkish. Only 9 of the 39 qualifying entities in that cut are Turkish, and the highest-placed is XSIDE at #8.
• The domestic shelf reverses it almost exactly. On the 22 questions that explicitly ask for Turkish or yerli brands, the split becomes 81.5% Turkish, led by Lescon (57.73) with 35.5% of responses. Domestic brands are highly visible when the user asks for them and substantially less so when they do not — the central finding of this edition.
• Four segments, four leaders. Nike leads Kadın & Stüdyo (women’s & studio) and Erkek & Takım Sporları (men’s & team sports); Decathlon leads Koşu & Outdoor (running & outdoor) at 40.7%; and XSIDE leads Genel & Athleisure at 48.6%, where the top seven places are held by six Turkish names. No single brand owns the category.
• Retailers, not marketplaces, are the aggregators. Decathlon — a retailer whose own-label products dominate its shelves — ranks #5 overall and #1 in Koşu & Outdoor. Its house labels Kiprun (#15) and Domyos (#20) qualify separately. Trendyol is present at #8 but does not lead, and Hepsiburada is last of the 30 qualifiers.
• The models see the same market at very different breadths. Grok names about 9.08 distinct brands per answer and Perplexity reaches the widest vocabulary (176 distinct brands), while GPT-4o-mini (3.90) and Claude Haiku (4.44) are far more selective. Nike is the most-named entity in four of the five models; in Grok it is third, behind Trendyol and Hepsiburada.
• Search was near-universal. 97.8% of answers were web-grounded; three of five models searched on 100% of answers and the lowest was Grok at 94.1%. With almost no “memory-only” answers, this edition does not include a web-search-vs-memory comparison; see Chapter 6.
• Sources are unusually diverse. Brand-owned sites account for only 32.8% of citations, with retailers at 21.2%, editorial and media at 7.9% and forum/social at 4.8%. The most-cited domain is trendyol.com (24.0% of responses), but the second is the editorial site onedio.com (21.3%), and five of the top nine are retailer sites.

Why it matters. Shoppers are shifting discovery from search engines and marketplaces to AI assistants. Independent survey work suggests roughly four in ten Turkish internet users were using ChatGPT monthly in mid-2025 (GWI), while TÜİK’s first measurement of the subject put self-reported generative-AI use at 19.2% of people aged 16–74 in 2025 — different definitions, both indicating a channel that now exists at scale. Whether a sportswear brand appears in those assistants’ answers is becoming a discovery channel in its own right. This report captures the first (baseline) snapshot of that shelf for Turkey’s sportswear and activewear sector. One reminder: the numbers below describe visibility and information availability, not which brands are best.

2. Methodology

One-sentence takeaway: Five large language models were asked 45 Turkish questions five times each, with no system prompt, producing 1,116 usable responses. Entities were identified by alias-based matching plus human review and ranked with a three-component (45/30/25) score, computed separately within each of ten cuts.

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: 45 questions × 5 repeats × 5 models = 1,125 responses; 9 were excluded, leaving 1,116 usable. The excluded responses returned a completed status with no error and empty answer text (5 Gemini, 4 Grok); one Grok row consumed 1,070 output tokens and returned nothing. These are treated as API artifacts rather than models declining to name brands, and are removed from all denominators — so Gemini is scored against 220 responses and Grok against 221, while the other three use the full 225. Single run, 18–19 July 2026. Final entity dictionary: 214 matched entities, of which 111 are Turkish and 103 foreign — a majority-domestic dictionary, which is worth holding in mind against the leaderboards that follow.

The ten cuts. Every metric is computed within a cut, against that cut’s own denominator. Rates from different cuts are not comparable.

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, 97.8% of answers were grounded (see Chapter 6).

2.3 Questions and segments

The 45 questions reproduce the real Turkish-language queries people bring to an assistant about sportswear: general discovery (“bir spor giyim markası söyle”), specific attributes (fabric quality, breathability, opacity, domestic production), and use cases (gym, running, hiking, football, Pilates, modest sportswear, kids). They are balanced globally across the three behavioural types — 15 discovery, 15 attribute, 15 use_case — and split 22 local against 23 open. No question rewriting was required for this edition; the seeded grid was already balanced.

The questions as seeded carried 25 distinct subcategory values across 45 questions, 17 of them appearing only once, which is too sparse to report on. Segment leaderboards are therefore built on a four-way activity rollup:

Of the 45, 22 are “local” (they explicitly ask for Turkish / yerli brands) and 23 are “open” (no origin restriction). The Turkish-vs-foreign comparison in Chapter 8 rests solely on the 23 open questions, the only fair basis for comparing Turkish and foreign brands, since the local questions cannot surface foreign brands by construction.

2.4 Entity extraction and classification

Entities were extracted by alias-based matching, word-boundary anchored, with Turkish-aware case folding — İ and ı are normalised explicitly, since naive lowercasing corrupts them. The final dictionary carries 214 matched entities across 277 alias strings, each reviewed by hand.

Origin follows the trademark, not the operator. A brand is Turkish only if the mark itself is Turkish-owned. Foreign marks operated in Turkey under Turkish licence — Slazenger, Lotto, Kappa, Umbro, Sergio Tacchini — are classified Foreign and tagged licensed-brand so they can be isolated. A product designed, manufactured and sold by a Turkish company does not make the underlying trademark Turkish.

Entity types. Five classes are tracked: brand, retailer, marketplace, sub-brand and licensed-brand. Trendyol and Hepsiburada are marketplaces. Decathlon and Intersport are retailers, retained in all cuts because they operate largely as own-label businesses — Decathlon states that around 90% of what it sells is designed in-house — but their house labels Kiprun and Domyos are tracked separately as sub-brands. Both the marketplaces and the retailers are removed in the brand-only views in §3.2.

Scope fit. 183 entities are classed Core and 32 Adjacent. Adjacent entities — modest-fashion, streetwear and kids labels with genuine sportswear ranges — are included in all cuts, because several map directly onto questions the study deliberately asked, including modest sportswear, athleisure/street style and kids. The classification is retained as a column so any table can be regenerated Core-only.

Definitions.

• Mention: a validated entity (or alias) appears in the answer text, counted once per response (presence, not frequency).
• Position (MRR): mean reciprocal of the entity’s first-appearance rank within each answer (first = 1.0; second = 0.5; …), averaged only over the responses where it appears.
• Breadth: number of the 5 models (0–5) that mention the entity within that cut.

2.5 AI Visibility Score (0–100)

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

The components are used at their absolute values rather than rescaled so that each cut’s leader reaches 100. A score of 100 would require an entity to be named in every single response, always first, by all five models; nothing approaches it, which is why the general leader sits at 58.61 rather than near the top of the scale. Scores are therefore comparable across cuts in a way rescaled scores are not — but rates still are not, because denominators differ.

Qualification (≥5% rule): only entities mentioned in at least 5% of that cut’s responses are ranked. This limits the risk that a rare brand which happens to appear first in a few answers inflates the position component. It does not eliminate it — see §3.1 and Chapter 11. Unfiltered metrics for all 214 entities appear in Chapter 12.

2.6 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. Overall Visibility Leaderboards

One-sentence takeaway: Nike leads the general and open cuts, Lescon leads the local cut, and the gap between those two pictures is the story of this edition.

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

3.1 General leaderboard: 30 qualified entities

Average score of the 30 qualified entities: 41.3. Origin split: 16 foreign, 14 Turkish.

Reading. Nike tops the table at 58.61, named in 34.95% of answers and by all five models, with the strongest position component of any high-volume entity (MRR 0.596). Adidas is third on comparable volume (30.20%) but is named noticeably later in answers (MRR 0.394). Lescon (#4) is the highest-placed Turkish entity at 22.31%, and XSIDE (#6) clears 49 on a much thinner 16.49% because it is named early when it appears. 29 of the 30 qualified entities are seen by all five models; the exception is Hepsiburada, at #30 on three.

The entry at #2 requires explanation. Ayma Active ranks second on a 6.54% mention rate — 73 responses, against Adidas’s 337, a brand mentioned 4.6 times more often. Its MRR of 0.762 means that on the rare occasions it is named, it is almost always named first. The 30% position weight is enough to carry it past brands with five times the volume. Ayma Active is a real and independently documented brand — a small family-run activewear label producing in small batches in Ankara — but its rank here reflects the score’s design, not its share of the conversation. The same mechanism places FD Sports at #5 in the local cut on 5.89% and Shimano at #10 in Koşu & Outdoor on 5.45%. The next table makes the effect visible rather than asking readers to take it on trust.

3.2 The same cut, ranked by mention rate

Identical data, ranked by volume alone rather than by the weighted score:

The reordering is instructive. Ayma Active leaves the top ten entirely. Trendyol climbs from #8 to #3 and Decathlon from #5 to #4, because both are named often but late. XSIDE falls from #6 to #8 and Kinetix from #12 to #9. Neither ranking is more correct than the other: the first asks how prominently is this brand named, the second asks how often. The AI Visibility Score deliberately blends the two, and readers should hold both pictures in mind for the rest of the report.

3.3 Brand-only view (marketplaces and retailers removed)

Three of the 30 qualified entities are not brands in the ordinary sense: the retailer Decathlon (#5) and the marketplaces Trendyol (#8) and Hepsiburada (#30). 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 top of the market is Nike, Ayma Active, Adidas, Lescon and XSIDE. Note that this removes retailers as well as marketplaces, a stricter rule than the underlying data files apply: Decathlon is a larger presence in this vertical than either marketplace, and leaving it in a “brand” ranking would be misleading. Decathlon’s house labels Kiprun and Domyos remain, as sub-brands of a retailer rather than the retailer itself.

3.4 Local leaderboard: 24 qualified entities (Turkish-only questions)

When the 22 questions that explicitly ask for Turkish/yerli brands are isolated, the shelf inverts:

Average score of the 24 qualified local entities: 42.3. Origin split: 17 Turkish, 7 foreign.

Reading. Lescon leads at 57.73 on 35.54% of responses, ahead of Ayma Active (55.22, again on position), XSIDE (53.78) and DeFacto (51.47, the highest mention rate in the cut after Lescon at 32.78%). The composition tells the real story: 17 of 24 qualifiers are Turkish, and the top seven are Turkish without exception. Nike, which leads every other overall cut, falls to #11 here.

Seven foreign brands nonetheless survive a question that asks for domestic ones — Hummel (#8), Tchibo (#9), Nike (#11) and others. This is not necessarily model error. Several are long-established in the Turkish market through local manufacturing, licensing or retail partnerships, and the assistants may be reflecting that familiarity rather than misreading the constraint. It is worth noting that the constraint is otherwise respected: the domestic share rises from 18.5% of open-cut mentions to 81.5% of local-cut mentions, which is a large and deliberate shift.

4. Segment Leaderboards

One-sentence takeaway: There is no single sportswear shelf. Four activity segments produce four different leaders, and the domestic-versus-foreign balance changes sharply between them.

The four segments partition the same 45 questions by activity context. Because question types are not balanced within segments (Genel & Athleisure is discovery-heavy, Koşu & Outdoor use_case-heavy), counts of distinct brands should not be compared across segments — see Chapter 11. Rankings within each segment are unaffected.

4.1 Kadın & Stüdyo — women’s & studio (15 questions, 371 responses)

The largest segment, and the most internationally contested: 21 of its 33 qualifiers are foreign. Nike leads at 57.41 (36.93%), with Lululemon (#4) and Under Armour (#5) close behind — Lululemon having entered Turkey physically only in May 2025. The two Turkish entries near the top are both position-driven rather than high-volume: Ayma Active (#2, 14.56%) and FD Sports (#3, 8.63%, MRR 0.901), the latter a modest-activewear label. Their placement is a useful signal even after discounting for the score’s position weight, because it indicates the assistants have specific, retrievable information attached to a narrow specialism.

4.2 Koşu & Outdoor — running & outdoor (11 questions, 275 responses)

The only segment led by a retailer. Decathlon tops it at 56.80 on 40.73% of responses — the highest mention rate of any entity in any segment — and its house label Kiprun ranks separately at #6 on 24.36%. Between them they occupy a large share of the running conversation. The North Face (#2) and Nike (#3) follow, with Lescon (#4) the leading Turkish name at 17.09%. This segment also produces the widest brand vocabulary of any cut, at 43 qualifiers, and contains the clearest question-design artifact in the study: Shimano ranks #10 on a 5.45% mention rate with an MRR of 0.967, resting almost entirely on the single cycling question, where it is effectively the only cycling brand available to name.

4.3 Genel & Athleisure — general & athleisure (10 questions, 247 responses)

The one segment where Turkish brands lead outright. XSIDE tops it at 66.41 — the highest score recorded anywhere in this edition — on 48.58% of responses, and six of the top seven places are Turkish: XSIDE, DeFacto, Lescon, Kinetix, Exuma and LC Waikiki, with only Nike (#2) interrupting. DeFacto actually has the higher mention rate (52.63%) but is named much later in answers (MRR 0.298 against XSIDE’s 0.652).

The pattern is coherent with how these brands are structured. XSIDE is LC Waikiki’s own label and Kinetix belongs to the FLO Group; both reach consumers through large domestic retail and e-commerce networks rather than as standalone technical sportswear companies. Where the question is about everyday sportswear rather than performance, the assistants reach for exactly those names. This segment is discovery-heavy by design (7 of 10 questions), which inflates the number of distinct brands named per answer but does not affect the ordering.

4.4 Erkek & Takım Sporları — men’s & team sports (9 questions, 223 responses)

The thinnest segment — 9 questions, 223 responses — and should be read as indicative rather than settled. Nike leads at 64.43, with Lescon (#2) the strongest domestic showing in any segment outside Genel & Athleisure at 31.39%, and Adidas third on a marginally higher mention rate than Nike (42.60% against 42.15%) but a much weaker position. Hummel (#4) is notably stronger here than in any other cut, consistent with its long association with team kit.

4.5 Segment comparison

5. Differences Between Models

One-sentence takeaway: The models broadly agree on Nike, but they differ enormously in breadth — Grok names more than twice as many brands per answer as GPT-4o-mini — and Grok alone puts marketplaces at the top.

5.1 Per-model behaviour summary

Web search was near-universal: three of the five models searched on every answer, Gemini on 95.0% and Grok on 94.1%. The clearest difference is breadth. Grok lists about 9.08 brands per answer and Perplexity reaches the longest tail (176 distinct brands), while GPT-4o-mini (3.90 per answer, 116 distinct) and Claude Haiku (4.44, 133) name a much more selective set. Gemini sits between them at 7.64 and 163.

This matters for interpretation. A brand in the long tail has a materially different chance of being named depending on which assistant the shopper uses. In Grok, an answer that lists nine brands has room for specialists; in GPT-4o-mini, an answer listing under four rarely goes beyond the best-known names.

5.2 Each model’s most-named brands

The tables below give each model’s five most-mentioned entities (rate = % of that model’s answers naming the brand).

Nike is the most-named entity in four of the five models. Grok is the exception, and instructively so: it names Trendyol in 82.8% of its answers and Hepsiburada in 63.8%, putting both ahead of Nike. Grok answers Turkish sportswear questions substantially by pointing at where to buy rather than what to buy — a behaviour no other model in the set shares at anything like that rate. Because Grok also has the widest brand vocabulary, its marketplace habit sits on top of, rather than instead of, a long brand list.

Two domestic names are consistent across every model: Lescon and DeFacto appear in all five top tens. XSIDE reaches four of five (all but Grok) and Kinetix three (Claude Haiku, GPT-4o-mini and Perplexity — the three most selective models). Claude Haiku and GPT-4o-mini both place Lescon second, ahead of Adidas, which is a stronger domestic showing than either the general or the open cut suggests.

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. About 97.8% of the 1,116 answers returned at least one citation or search result, and the “own-knowledge” segment is a small residue, far too small to support a comparison. Three of the five models (GPT-4o-mini, Claude Haiku, Perplexity) searched on 100% of answers, with Gemini at 95.0% and Grok at 94.1%.

Because almost all queries used web search, a separate web-search-vs-memory chapter would compare roughly 1,091 grounded answers against about 25 ungrounded ones, which is not a meaningful contrast, so it is left out this edition. The takeaway is that, for Turkish sportswear questions, these assistants reach for live web content almost every time. This report therefore largely measures what the models retrieve and select, not what they know unprompted — and retrieval is sensitive to the live web, which changes independently of the models. 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: Discovery questions surface Turkish brands, use-case questions surface Decathlon, and the number of qualifying brands halves as questions get more specific.

The 45 questions are read through three behavioural types: discovery (general “best / recommended”), attribute (specific qualities such as fabric, breathability, opacity or domestic production) and use_case (activity-driven, such as gym, running, hiking, football or Pilates). Each type is a cut in its own right, with its own denominator: 371 discovery, 372 attribute, 373 use_case responses.

Reading. The three types behave differently in three ways.

Who leads. Nike leads discovery and attribute; Decathlon leads use_case at 37.3%, ahead of Nike at 34.3%. When the question describes an activity — what to wear for running, hiking, the gym — the assistants reach for the retailer that sells across all of them.

How Turkish the answer is. Discovery is the most domestic type by a distance: XSIDE (#2), Lescon (#3), DeFacto (#4) and Kinetix (#6) all place in the top six, and DeFacto has the single highest mention rate of the cut at 44.5%. Open “recommend me a sportswear brand” questions in Turkish surface Turkish brands. Attribute and use-case questions surface fewer.

How many brands qualify. 45 entities clear the 5% threshold on discovery questions, 31 on attribute and only 22 on use_case. The more specific the question, the narrower the set of brands the assistants consider — and the more the answer concentrates on a handful of large names.

Two caveats. Ayma Active places #2 on attribute and #3 on use_case on mention rates of 10.2% and 6.7%, the position effect described in §3.1 appearing again. And because question types are not balanced within segments, the discovery cut draws disproportionately from Genel & Athleisure, which is itself the most domestic segment; part of the “discovery is Turkish” pattern is design, not market.

8. Open Market: Turkish vs. Foreign

One-sentence takeaway: Across origin-neutral questions, 81.5% of mentions go to foreign brands and 18.5% to Turkish — and the local cut reverses that almost exactly, showing that domestic visibility here depends on being asked for.

This chapter relies only on the 23 open questions with no origin restriction (573 responses), the only fair basis for comparing Turkish and foreign brands. A blended figure across all 45 questions would be misleading, because 22 of them constrain the answer to domestic brands by construction.

8.1 Origin share

Origin share can be measured two ways, and both are reported here because they answer different questions:

The gap between the two measures is itself informative. Foreign brands hold 76.9% of the positions on the shelf but 81.5% of the volume, because the largest entities are foreign: Nike alone accounts for 315 open-cut mentions, more than half the combined total of all nine qualifying Turkish entities.

The local cut is the mirror image. On the 22 questions asking for Turkish brands, Turkish entities take 81.5% of mentions and 70.8% of qualifying entities — a near-perfect inversion of the open-cut mention share. The two cuts use the same models, the same window and the same scoring; the only thing that changes is whether the question asks for domestic brands.

8.2 Open-market top 15 entities (any origin)

8.3 The pattern: domestic visibility is conditional

The individual ranking shows what the aggregate hides. The open-market top seven are entirely foreign — Nike, Adidas, Decathlon, Under Armour, Lululemon, The North Face, Patagonia — and the first Turkish name is XSIDE at #8, on an 8.20% mention rate. Trendyol at #11 is a marketplace rather than a brand. Below that the Turkish entries are thin and scattered.

Set that against §3.4, where Lescon reaches 35.54% and the top seven are Turkish without exception. The same assistants, asked the same kind of question with one clause changed, produce two different markets. The reasonable interpretation is that the models hold retrievable information about Turkish sportswear brands — otherwise the local cut could not look as it does — but do not surface it by default when the question leaves origin open. Domestic visibility in this vertical is conditional on being asked for.

This is a description of model behaviour, not of the market. Turkey is a net exporter under the customs headings used for sportswear, with official 2025 exports of USD 394.7m against USD 111m of imports, so a foreign-led AI shelf does not imply a foreign-led industry. What it does suggest is a discovery gap: the domestic brands that exist are less findable through this channel than their market position would predict. Whether that gap 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: Unlike other verticals in this series, AI’s picture of Turkish sportswear rests less on brands’ own sites than on retailers and editorial content — a substantially more contested source base.

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, giving a map of where these systems read about sportswear brands. Across the run the models cited 623 distinct domains.

9.1 The 9 most-cited domains

The most-cited domain is the marketplace trendyol.com, at 24.0% of responses — but its lead is narrow, and what follows is unusual. Second is onedio.com, an editorial and listicle site, at 21.3%. Then come four retailer domains in a row: sporthink.com.tr, decathlon.com.tr, intersport.com.tr and qlu.com.tr. The first brand-owned domain in the list is nike.com, at #9 on 9.9%.

Five of the top nine are retailers. In a vertical where the most-cited sources are the places that sell many brands and the sites that write listicles about them, a brand’s visibility depends heavily on how it is represented in third-party assortments and roundups, not only on its own site.

9.2 Source-type mix

9.3 The mixed-ecosystem finding

Brand-owned sites are the largest single class at 32.8%, but they are not dominant. Retailers (21.2%) and marketplaces (6.1%) together contribute 27.3%, and the commercial classes combined — brand, retailer and marketplace — reach 60.1%. The remainder is genuinely varied: a long tail classed as “other” spanning 424 domains (23.5%), editorial and media (7.9% across 25 domains), forum and social (4.8% across 10) and B2B or supplier sources (3.7%).

Two things follow. First, editorial and social content matter measurably more here than in verticals where brand-owned sites approach two-thirds of citations: roughly one citation in eight comes from editorial or forum/social sources, and a single editorial domain is the second most-cited in the study. Second, the concentration is low — 424 distinct domains sit in the “other” bucket alone — which means no small set of sources controls the picture. Read as a description rather than advice: in this market, visibility appears associated with being present across retailer assortments and third-party roundups, not only with maintaining a good own-site.

10. What the Patterns Suggest

One-sentence takeaway: The most visible names share broad model coverage and a clear activity context; the widest gap in the data is between Turkish brands’ domestic-cut visibility and their near-absence when origin is left open.

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) Breadth is a baseline condition. 29 of the 30 qualified general entities appear in all five models. The single exception, Hepsiburada, sits last despite a 14.96% mention rate, precisely because only three models name it. Broad model coverage is what the visible entities share; a brand strong in one assistant and absent from three will not rank.

2) Mention and position are different signals. Some entities are named often but late — Trendyol (23.92%, MRR 0.303), DeFacto (21.68%, MRR 0.349), Puma (14.61%, MRR 0.237). Others are named rarely but first — Ayma Active (6.54%, MRR 0.762), FD Sports, Scorp. Both patterns clear the 5% bar and both are real, but they describe different positions in the market: one is a name that comes up in passing, the other a name that comes up as an answer.

3) Context determines the leader. No entity leads all four segments. Nike leads two, Decathlon and XSIDE one each, and XSIDE’s 66.41 in Genel & Athleisure is the highest score anywhere in the edition. A brand’s visibility is not a single number; it is a number per context, and the useful question is which segment and question type a brand owns rather than where it sits overall.

4) Retail intermediation is unusually strong here. Decathlon ranks #5 overall and #1 in one segment; its house labels Kiprun and Domyos qualify independently; five of the nine most-cited domains are retailers; and Grok answers largely by naming marketplaces. Sub-brands also under-count structurally — Kiprun (6.99%) and Domyos (5.56%) are named far less often than Decathlon (23.57%) even where the house label is the actual product being recommended. Where shoppers meet this category through retailers, the assistants describe it that way too.

5) Domestic visibility is conditional, and that is the gap. Turkish brands take 81.5% of mentions when asked for and 18.5% when not. The models clearly hold the information; the local cut proves it. The entity dictionary sharpens the point: 111 of the 214 tracked entities are Turkish, a domestic majority, yet only 14 of 30 qualify in the general cut and only 9 of 39 in the open cut. Turkish brands are not missing from these models’ vocabulary — they are missing from the default answer. What is missing is the default association between an unconstrained Turkish-language sportswear question and a domestic brand. Given that 97.8% of answers were web-grounded and that retailer and editorial sources carry unusual weight in this vertical, the mechanism is likely retrieval rather than knowledge — which is a different kind of gap from simply not being known.

How a brand can locate itself in the data. Using the dataset (available on request), a brand can read five things in order: whether it appears in all five models; how often it is mentioned within its cut; how early it is named when mentioned; which segment and question type it is strongest and weakest in; and whether it appears at all in the open cut as well as the local one. 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.

• Single collection window. All data comes from a roughly four-hour window on 18–19 July 2026. Model outputs drift; these figures are a snapshot, not a stable property of the models. This is the first (baseline) edition; quarterly editions will enable trend analysis.
• The score rewards position as well as frequency. This is by design, but it produces counterintuitive placings. Ayma Active ranks #2 in the general cut on a 6.54% mention rate because its MRR is 0.762. Fifteen entities across the ten cuts sit in a top ten on under 10% mentions, including FD Sports, Tchibo, Patagonia, Tutku, Scorp, Champion and Lacoste. §3.2 gives the mention-rate view of the general cut for comparison; read thin-mention ranks cautiously everywhere.
• Shimano is a question-design artifact. Its MRR of 0.967 in Koşu & Outdoor rests on the single cycling question, where it is effectively the only cycling brand available to name. This is not a market signal.
• Question types are not balanced within segments. Genel & Athleisure is discovery-heavy (7 of 10), Koşu & Outdoor use_case-heavy (6 of 11). Discovery prompts reliably surface more distinct brands per answer, so cross-segment comparisons of distinct-brand counts partly reflect the design, not the market. Within-segment rankings are unaffected.
• Erkek & Takım Sporları is the thinnest segment — 9 questions, 223 responses — and should be read as indicative.
• Sub-brands under-count structurally. Kiprun and Domyos are named far less often than “Decathlon” even where the house label is the actual product being recommended.
• Origin follows the trademark, which has consequences. Foreign marks operated under Turkish licence are classified Foreign. Notably, no licensed brand qualifies in any cut: Slazenger, Lotto, Kappa, Umbro and Sergio Tacchini are all commercially present in Turkey and all sit below the 5% threshold everywhere. They therefore contribute nothing to either origin share. Readers who consider licensed operations part of the domestic industry should treat the origin figures accordingly.
• Adjacent entities are included. 32 of the 214 entities are modest-fashion, streetwear or kids labels with genuine sportswear ranges, retained because the study deliberately asked questions they answer. All tables can be regenerated Core-only from the underlying data.
• Web-grounded, not recall. 97.8% of responses carried at least one citation, so this largely measures what models retrieve and select, not what they know unprompted. Retrieval is sensitive to the live web, which changes independently of the models.
• Turkish-language prompts only. Results may not transfer to the same questions asked in English about the Turkish market.
• Nine responses were excluded. These returned a completed status with no error and empty answer text (5 Gemini, 4 Grok). They are treated as API artifacts, not as models declining to name brands, and are removed from all denominators.
• Matching corrections were applied before the figures were finalised. Markdown link targets and domain-style anchors were stripped before matching, so that being cited as a source is not counted as being recommended as a brand; this affected 28 entities. The alias “mavi” was removed, because all 44 text hits were the Turkish colour word in product names rather than the brand. And 172 generic Turkish phrases such as kumaş kalitesi and nefes alabilirlik were reviewer-dropped, removing 707 spurious mentions.
• Gemini’s sources are read indirectly. All Gemini reference URIs are redirect wrappers; domains are taken from the search-result title field instead, since extracting from the URI would have ranked a Google redirect as the study’s most-cited source.
• Small score gaps are not meaningful. In the packed middle of each leaderboard, differences of a point or two should not be over-interpreted; only larger gaps are robust.
• 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.
• Source-type mix is heuristic. Domain types were auto-classified; treat the mix as indicative. Source coverage also varies by model.
• Precision over recall. Entity matching was cautious to avoid false positives, so brands mentioned indirectly may be undercounted. There is no sentiment layer; the study is strictly quantitative.
• 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 (45) and their categories

Distribution: 15 discovery, 15 attribute, 15 use_case; 22 local and 23 open; across four activity segments (15 / 11 / 10 / 9). No question was rewritten for this edition.

12.2 All tracked brands (214)

12.3 Data availability

The study tracked 214 entities; of these, 30 cleared the 5% threshold in the general cut, 24 in the local cut, 39 in the open cut, and 80 in at least one of the ten cuts. To support scrutiny and reproduction, the underlying data is available on request: response-level brand mentions, all ten qualified leaderboards and their brand-only variants, full unfiltered entity metrics, per-model and per-question-type breakdowns, the open-market breakdown, the full 623-domain source list, the entity dictionary with its decisions log, and the methodology and data dictionary.

All figures reflect a single point-in-time run (18–19 July 2026) and are subject to the limitations in Chapter 11. This is the first (baseline) measurement; the study will be repeated quarterly.

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.