This summary reflects the September 2026 edition (edition 2). Earlier editions are listed with the current one and keep their own figures.
More and more shoppers are skipping the search bar and the marketplace filter and simply asking an AI assistant what to buy. So we asked five of them, 1,125 times, about underwear, socks and swimwear in Turkey. Here’s what they named, what changes the answer, and why “most recommended” is not the same as “best.”
Type “hangi marka iç giyim önerirsin” into ChatGPT, and you’ll get a confident answer in seconds. Ask Gemini for a good Turkish sock brand, and you’ll get another tidy list. Ask Grok, and its answers name 11.48 brands, retailers or marketplaces on average; Claude’s name 4.19. Discovery is moving from the search engine and the marketplace grid, where you scroll and decide for yourself, to the AI assistant, where a single answer arrives pre-sorted. When the assistant names three brands, those three have effectively been placed on a shelf at eye level, and everything else is in the stockroom.
Why this is worth paying attention to
This is a large, export-heavy corner of Turkish manufacturing. Globally, the lingerie market alone was worth roughly $98.7 billion in 2025 and is forecast to keep growing at about 7% a year (Grand View Research). Hosiery is where Turkey’s weight is clearest: the country is the world’s second-largest sock exporter, shipping about $864 million of socks in the first nine months of 2024, with Germany alone taking 22.4% ($193 million), while the domestic Turkish socks market is valued at roughly $223 million in 2025, or about 3.3 pairs per person (Kohan Textile Journal, citing Statista and export data). Underwear, socks and swimwear together sit on top of one of the largest textile and ready-to-wear bases in the region.
Discovery for all of it is increasingly AI-mediated. Adobe, tracking more than a trillion visits to U.S. retail sites, found generative-AI-driven traffic up roughly 4,700% year on year by July 2025, with 38% of surveyed shoppers saying they had used generative AI to shop and 52% planning to (Adobe). Over the 2025 holiday season, AI referrals rose nearly seven-fold year on year and converted about 31% better than other traffic, a signal that assistants are becoming discovery and comparison interfaces, not just a novelty (Adobe). Personal-care and apparel categories are squarely in that shift.
All of which makes one question pressing: as shoppers move to AI assistants, who do those assistants actually name when someone asks for underwear, socks or a swimsuit in Turkey? So we measure that shelf every quarter. On 21 September 2026 we put the same 45 Turkish-language questions we asked in June to five leading AI assistants — Claude, Gemini, ChatGPT, Perplexity and Grok — asking each question five times: 1,125 answers across innerwear, hosiery and beachwear.
How Turkish the shelf looks depends on how you ask
The strongest finding in this edition is about the question, not any brand.
Twenty-three of our 45 questions ask for Turkish brands explicitly, the way a shopper would: “iyi bir yerli iç giyim markası var mı.” The other twenty-two ask about the product and say nothing about origin: “dikişsiz belli olmayan külot önerisi.”
When the question asks for Turkish brands, domestic brands take 97.38% of the qualifying mentions and foreign brands 2.62%, with origin known for every one of them. Only one foreign brand qualifies at all: MOEVA, a London swimwear brand, named in 55 of those 575 answers. When the question says nothing about origin, domestic brands take 55.31% and foreign brands 44.69% — again with origin known for every qualifying mention — and 16 of the 28 brands in that count are foreign. Penti still leads, but the French retailer Decathlon is second and Marks & Spencer seventh, with Intimissimi and Calzedonia close behind.
June’s edition reported its own split for the origin-neutral questions. It decided a brand’s origin by a different rule — founded and headquartered in Turkey — while this edition uses the country the brand is run from. Several brands changed sides under the new rule, so the two splits measure different things and are not compared.
One brand leads on every cut
Penti appeared in 50.49% of all 1,125 answers, was named by all five assistants, and tops the overall ranking with a score of 97.27. Dagi, Suwen, Ayyıldız and Dündar Çorap complete the top five. Penti also leads when the question asks for Turkish brands, when it doesn’t, and on every type of question we asked.
Below the leaders the list thins out fast. Of 487 brands, retailers and marketplaces we tracked, 26 were named often enough to rank across all 45 questions — 19 domestic and 7 foreign.
June’s summary also named a different leader for underwear, socks and swimwear. This edition publishes no figures by product type, because the study did not set those groupings up as separate cuts before the answers were collected.
The current ranking
Below are the 26 brands, retailers and marketplaces that cleared the 5% mention threshold across all 45 questions, ranked by AI Visibility Score. This is a visibility ranking, not a quality ranking. The last column shows how much more or less often each one was named than in June.
| # | Entity | AI Score | Mention rate | Δ mention rate since June (points) |
|---|---|---|---|---|
| 1 | Penti | 97.27 | 50.49% | +9.49 |
| 2 | Dagi | 70.16 | 30.49% | +5.69 |
| 3 | Suwen | 66.34 | 22.58% | +4.14 |
| 4 | Ayyıldız | 64.55 | 19.56% | -10.88 |
| 5 | Dündar Çorap | 63.32 | 9.33% | -3.20 |
| 6 | Bolero | 60.97 | 9.60% | +0.20 |
| 7 | Kom | 59.94 | 19.02% | +13.56 |
| 8 | Decathlon | 57.69 | 11.82% | +6.27 |
| 9 | Mısırlı | 56.65 | 8.36% | -2.30 |
| 10 | Intimissimi | 55.70 | 5.16% | -1.47 |
| 11 | Calzedonia | 55.53 | 8.71% | +0.56 |
| 12 | Marks & Spencer | 53.97 | 10.04% | -6.07 |
| 13 | Tutku | 53.94 | 11.38% | +5.92 |
| 14 | Blackspade | 53.34 | 14.13% | +8.85 |
| 15 | Koton | 53.19 | 8.62% | +4.95 |
| 16 | Aytuğ | 52.37 | 7.91% | +4.42 |
| 17 | MOEVA | 49.43 | 6.84% | +2.91 |
| 18 | Anıl | 49.39 | 5.78% | +4.26 |
| 19 | DeFacto | 49.35 | 10.40% | +1.81 |
| 20 | Cottonhill | 48.89 | 5.33% | -2.10 |
| 21 | Trendyol | 47.57 | 15.29% | -6.20 |
| 22 | Nike | 46.41 | 5.16% | +1.75 |
| 23 | NBB | 43.91 | 5.96% | +3.63 |
| 24 | LC Waikiki | 43.61 | 6.31% | -0.85 |
| 25 | Oysho | 42.11 | 5.60% | +2.74 |
| 26 | Hepsiburada | 35.82 | 11.02% | -1.06 |
Two of the entries above, Trendyol and Hepsiburada, are marketplaces rather than brands; the full report includes a brand-only view, the separate leaderboards for questions that do and don’t ask for Turkish brands, and the full Turkish-versus-foreign breakdown.
What changed since June
This is the second edition, so for the first time we can look at movement — with one caution stated up front: two of the five assistants changed. Gemini 2.5 Flash-Lite and GPT-4o-mini answered in June; Gemini 3.6 Flash and ChatGPT 5.6 Luna Pro answered in September. The 45 questions were identical.
On those questions, Kom gained the most, from 5.46% of June’s answers to 19.02% of September’s (+13.56 points), followed by Penti (+9.49) and Blackspade (+8.85). Ayyıldız fell the most (-10.88), though it is still fourth overall, followed by Trendyol (-6.20) and Marks & Spencer (-6.07). Of the 326 names that appeared in both editions, 174 were named more often and 130 less often.
Kom’s gain sits mostly in the two assistants that joined this edition, which is exactly why we read these changes as direction rather than exact size. June’s figures here are June’s answers counted the way this edition counts, so for a few names — above all the two marketplaces — they differ from what June’s page showed.
The assistants don’t all list the same number of brands
At the very top the five assistants agree: every one of them names Penti more often than anything else. Below that they differ sharply in breadth. Grok was the most generous, naming 11.48 brands, retailers or marketplaces in a typical answer, with Gemini at 9.51; Claude (4.19) and ChatGPT (4.21) were the most selective. The more names an assistant lists, the more likely a smaller label is to make the cut.
Grok also behaves differently in kind: it names the marketplaces far more than the others do. Of Hepsiburada’s 124 mentions across all five assistants, 108 came from Grok; of Trendyol’s 172, 134 did. Both marketplaces’ places in the ranking largely reflect one assistant’s habit of pointing to where to buy.
Most answers were built with a web search
Every assistant was asked to search the web, and 87.11% of answers show that they did. Four of the five grounded every answer. Gemini grounded 35.56%; the rest of its answers came back with no information about whether it searched, which is not the same as answering from memory. The site cited most often was Penti’s own, penti.com, in 80 of the 225 question-and-assistant combinations (35.56%), followed by ayyildiz.com.tr (49) and the marketplace trendyol.com (46).
The most important finding isn’t a ranking
Here’s what we’d most like readers to take away: being recommended by AI does not mean a brand is good. When an assistant names a brand often, it usually means one thing: there is a lot of information available about that brand. It does not mean the brand’s products fit better, last longer, cost less, or are more comfortable. AI visibility measures how easily a brand is found, not how good it is. Shoppers should not read an AI recommendation, or this report, as a verdict on quality.
Read the full report
This page is a summary. The complete report contains the full ranking and the brand-only view, the separate leaderboards for questions that do and don’t ask for Turkish brands, the Turkish-versus-foreign breakdown, the breakdown for each AI assistant, the winners by question type, the comparison with June 2026, the source analysis, and the full methodology. This is the second edition of a study we repeat every quarter; you’ll find the current and past editions listed on the right.
Frequently asked questions
Across all 45 questions, Penti leads with a 97.27 score, appearing in 50.49% of 1,125 answers and named by all five assistants. Dagi, Suwen, Ayyıldız and Dündar Çorap complete the top five.
It depends on how the question is phrased. On the 23 questions that ask for Turkish brands, domestic brands take 97.38% of qualifying mentions and foreign brands 2.62%. On the 22 questions that say nothing about origin, they take 55.31% and 44.69%. Origin is known for every qualifying mention on both.
No. A high AI visibility score means these systems have a lot of information available about a brand and surface it readily. It does not measure quality, durability, price or reliability. Visibility is about how easily a brand is found, not how good it is.
On the leader, yes: every one names Penti more often than anything else. On length, no: Grok names 11.48 brands, retailers or marketplaces per answer and Claude 4.19. Grok also accounts for most of the marketplace mentions — 108 of Hepsiburada's 124 and 134 of Trendyol's 172.
Claude Haiku 4.5, Gemini 3.6 Flash, ChatGPT 5.6 Luna Pro, Perplexity Sonar and Grok 4.3 were each asked the same 45 Turkish-language questions five times on 21 September 2026, producing 1,125 answers. 487 brands, retailers and marketplaces were tracked in total.
The 45 questions were the same, but two of the five assistants were replaced, so changes show direction rather than exact size. Measured by how often each brand was named, Kom gained the most (+13.56 points) and Ayyıldız lost the most (-10.88). The Turkish-versus-foreign split is not compared with June's, because the rule for deciding a brand's origin changed.
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
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