1,098 responses, 5 models, three segments: the first AI-visibility snapshot of Turkey's mother & baby products market

Vertical: Mother & baby products (anne ve bebek ürünleri), Turkey
Method: A single point-in-time study of 1,098 responses across five large language models (GPT-4o-mini, Gemini, Claude, Perplexity, Grok), each asked 44 Turkish-language questions five times
Collection date: 26 July 2026

Key terms: AI visibility, AI Visibility Score, mother and baby brands, bebek arabası, oto koltuğu, biberon, bebek giyimi, Turkish (yerli) baby brands, LLM brand recommendations, ChatGPT/Gemini/Claude/Perplexity/Grok brand recommendation, Generative Engine Optimization (GEO).

This report reproduces the real Turkish-language questions parents ask AI assistants about baby clothing, travel gear and feeding products, and measures how often, in what order, and across how many models five large language models name each brand. Because this market divides naturally by product type and because roughly half the questions ask specifically for Turkish brands, the entities are ranked in nine cuts, defined in Chapter 2. It is a single point-in-time study and should be read within the limitations in Chapter 12.

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 mother and baby products 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, safer, 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. This matters more in this category than in most: parents buying car seats, strollers and feeding equipment are making safety-relevant decisions, and an AI assistant naming a brand first is not a safety endorsement. Product safety is governed by standards and certifications, not by visibility. 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: Chicco leads the general leaderboard on sheer volume, but no brand wins this market — clothing, travel gear and feeding are each led by a different name, and almost nothing that qualifies in one segment qualifies in another.

• Chicco leads at 85.7, on the strongest mention rate in the study. It appears in 30.7% of all 1,098 responses (337 mentions), around 1.6 times the rate of the next entity and 1.7 times the next brand, and it is the most-mentioned brand in four of the five models and third in the fifth. Philips Avent (74.6) is second and The Nest Kids (73.7) third.
• Volume and prominence are different things here. Chicco tops the mention component but is named comparatively late (MRR 0.373). Philips Avent is named in less than half as many answers (13.4%) yet holds the highest position score in the cut (MRR 0.709), which is what puts it second. Read the components together, not the composite alone.
• 30 entities qualified for the general ranking. Of the 233 entities tracked, 30 cleared the 5% mention threshold across all 44 questions: 17 Turkish and 13 foreign, an almost even split. Twenty-eight are brands or retailers and two are marketplaces. The average score of the 30 is 58.4.
• No brand wins the market. Each segment has its own leader: The Nest Kids in Clothing (91.7), Chicco in Travel & Gear (98.5), Philips Avent in Feeding (91.6). Of 69 entities that qualify somewhere across the three segments, 61 qualify in only one. Neither The Nest Kids nor Philips Avent qualifies in any segment other than its own.
• Origin splits by question, not by market. On the 23 origin-neutral (open) questions, mentions run 54.5% foreign to 45.5% Turkish and 21 of 37 qualified brands are foreign. On the 21 questions that ask for yerli brands, Kraft leads and 15 of 19 qualifiers are Turkish. The assistants clearly recognise the constraint; the two shelves are genuinely different.
• One assistant answers with shops, not brands. Grok’s three most-named entities are Trendyol (85.4% of its answers), Hepsiburada (75.3%) and ebebek (42.5%) — the brand that leads the study, Chicco, is only its fourth. Neither Trendyol nor Hepsiburada reaches any other model’s top ten.
• The one retailer that ranks is ebebek, at #17 overall on a 14.0% mention rate. It is the only retailer to qualify anywhere, and one of just four entities that qualify in all nine cuts.
• The models diverge sharply on breadth. Grok names about 9.5 brands per answer across 171 distinct entities; GPT-4o-mini names 4.0 across 129. Only two brands appear in all five models’ top tens: Chicco and Little Gusto.
• Search was near-universal, and the sources are unusual. 99.1% of answers were web-grounded. The most-cited domain is the specialist retailer e-bebek.com (29.6% of responses), ahead of trendyol.com. Brand-owned sites account for 37.4% of citations and editorial content for 20.6%.

Why it matters. Parents are shifting product research from search engines and forums to AI assistants, often at exactly the moment they are choosing a car seat or a first bottle. Whether a 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 Turkey’s mother and baby sector. One reminder: the numbers below describe visibility and information availability, not which products are best or safest.

2. Methodology

One-sentence takeaway: Five large language models were asked 44 Turkish questions five times each, with no system prompt, producing 1,098 usable responses. Entities were identified by automated extraction plus human review and ranked with a three-component (45/30/25) score, computed separately within nine 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: 44 questions × 5 repeats × 5 models = 1,100 responses; 2 were excluded, leaving 1,098 usable. Both exclusions returned a completed status with an empty answer body — one from Gemini, one from Grok. Denominators are computed per (question, model) cell rather than as 44 × 5 × 5, so those two missing cells reduce the relevant denominators rather than being quietly backfilled. Single run, 26 July 2026. Final dictionary: 233 entities — 226 brands and retailers, 6 marketplaces, and 1 entity resolved to an adjacent industry and marked out of scope.

The nine cuts. This vertical is read three ways at once: by origin constraint, by product segment, and by question type. Each cut is scored independently.

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, 1,088 of 1,098 answers were grounded — 99.1% (see Chapter 6).

2.3 Questions

The 44 questions reproduce the real Turkish-language queries parents bring to an assistant: general discovery (“bebek arabasında hangi markalar gerçekten kullanışlı?”), specific attributes (organic fabric, i-Size safety, anti-colic), and use cases (hospital discharge sets, travel systems, baby-led weaning). They are written the way people actually type them — short, colloquial, often ungrammatical, and not disambiguated. That is deliberate: the study measures what an assistant tells a real person asking a real question, not what it tells a carefully-worded prompt. Chapter 8 covers what happens when a question is genuinely ambiguous.

Of the 44, 21 are “local” (they explicitly ask for Turkish / yerli / Türk malı brands) and 23 are “open” (no origin restriction). The Turkish-vs-foreign comparison in Chapter 9 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 validation

Entities were extracted in two stages. (1) Automated extraction: candidate names were collected from all 1,098 analysed responses at a floor of three responses, then filtered for Turkish common words and generic descriptors. Matching is alias-based and word-boundary anchored, with Turkish case folding handled through an explicit İ/ı translation table rather than a naive lowercase, which corrupts those characters. (2) Human review: the candidate list was reviewed by hand, with each entity assigned a keep/drop decision, trademark and operator country, entity type, scope fit, product category, verified website and a one-line rationale.

Two matching rules are worth stating because they change the numbers. Link targets are stripped before matching: markdown link destinations and bare domains are removed and only anchor text is kept, so a brand that owns a cited domain is not credited with a recommendation it did not receive. Sourcing is measured separately, in Chapter 10. And every alias of six characters or fewer was inspected in context — 139 of them — which caught three false merges that would otherwise have inflated three brands: “Emma” (a Finn + Emma reference, not a Lionelo car seat), “Asya” (the continent, in sentences about export markets) and “Pegasus” (the airline, in answers about cabin-baggage stroller dimensions).

Origin rule. Origin follows the trademark, not the operating company. Foreign marks sold in Turkey under licence or through a local distributor are recorded as foreign and typed as licensed brands: Kanz, Pierre Cardin, Carter’s, OshKosh, Bambino. This is consequential — Kanz alone ranks #13 in the general cut and #4 in the local cut, and an operator-based rule would have counted it as Turkish and changed the origin balance of the whole Clothing segment. Eight entities whose trademark country could not be established carry an unknown origin; they remain in the leaderboards and are excluded from origin shares.

Definitions.

• Mention: a validated entity (or alias) appears in the answer text, counted once per response (presence, not frequency). Where an answer names both a brand and one of its products — “Chicco” and “Chicco Goody Plus” — that is a single mention.
• 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

Each component is scaled so the leading qualified brand in each cut = 100, then combined with the weights. Qualification (≥5% rule): only entities mentioned in at least 5% of that cut’s responses are ranked. At these denominators the threshold requires at least 18 responses in the smallest cut, so no qualifier rests on a thin base.

A caution specific to this edition. Because the composite rewards position as well as frequency, a brand named in a third of answers but always mid-list can score below a brand named in a tenth of answers but always first. That trade-off is unusually visible here. Chicco has by far the highest mention rate in the study but a middling MRR; Philips Avent, OiOi and Pilsan all rank well above their mention rates because they are named early when they are named at all. The score is a summary, not a substitute for the three numbers behind it, and every table in this report shows all three.

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: Chicco leads the general and open cuts on volume; ask for Turkish brands instead and Kraft takes over a shelf that is 14/18 domestic.

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

3.1 General leaderboard: 30 qualified entities

Average score of the 30 qualified entities: 58.4. Origin split: 17 Turkish, 13 foreign.

Reading. Chicco maxes the mention component and tops the table at 85.7, appearing in 30.7% of answers — the only entity above 20% in this cut. What follows is a cluster rather than a chase: Philips Avent (74.6), The Nest Kids (73.7) and TinyLamb (71.2) are separated by three and a half points in total, and all three rank on position rather than volume, each named in about 13% of answers but named early. Kraft (69.8) and Little Gusto (69.1) follow on the reverse profile — more mentions, later placement. Twenty-eight of the 30 qualified entities are seen by all five models; the two exceptions are the marketplaces, Trendyol (#15, breadth 4) and Hepsiburada (#25, breadth 3).

The origin balance is close to even, and the two groups interleave rather than clustering: seventeen Turkish entities and thirteen foreign share the table, and the top six alternate between them.

3.2 Brand-only view (marketplaces removed)

Only two of the 30 qualified entities are marketplaces rather than brands: Trendyol (#15) and Hepsiburada (#25). They are kept in the headline ranking because that is how shoppers meet the shelf, but neither sits high enough to distort the top of the table — removing them leaves the leading ten unchanged, with Chicco, Philips Avent and The Nest Kids still first, second and third. Chapter 5 explains why they rank as low as they do here.

ebebek is the genuinely difficult case. It is a specialist mother-and-baby retailer, a marketplace hosting third-party sellers, and a house of own-label brands, all at once — no single entity type describes it fairly. It is recorded here as a Retailer and kept in the brand views, because it sells its own brands alongside everyone else’s. At #17 in the general cut it is the only retailer in the study to qualify anywhere, and one of only four entities that qualify in all nine cuts.

LC Waikiki (#18) and the licensed brands Kanz (#13) and Carter’s (#23) are treated as brands, not retailers, and remain in this view.

3.3 Local leaderboard: 19 qualified entities (Turkish-only questions)

When the 21 questions that explicitly ask for Turkish / yerli / Türk malı brands are isolated, the shelf changes hands.

Average score of the 19 qualified local entities: 62.7. Origin split: 15 Turkish, 4 foreign.

Reading. Kraft (86.5) leads decisively, named in 20.0% of local-cut answers — a 13-point gap to Pilsan (73.2), which qualifies on the opposite profile entirely: 8.8% mentions but an MRR of 0.887, the highest position score in the cut. Trendyol rises to #3 here, and its 21.8% mention rate is actually the highest in the cut; a breadth of 3 and a low MRR are what hold it back. ebebek takes #12. Chicco, first in both the general and open cuts, falls to #8.

Four foreign brands survive the yerli filter: the licensed brands Kanz (#4) and Bambino (#18), plus Chicco (#8) and Lansinoh (#13). Unlike verticals where the local cut comes back entirely domestic, the assistants here occasionally answer a “Turkish brand” question with a foreign name — usually one manufactured or licensed locally. That is a real finding about how the constraint is interpreted, and worth watching across editions.

4. Segment Leaderboards: Three Markets, Not One

One-sentence takeaway: Clothing, travel gear and feeding are led by three different brands, and 61 of the 68 entities that qualify in any segment qualify in only one.

The general leaderboard describes an average that no shopper experiences. A parent asking about newborn bodysuits and a parent asking about i-Size car seats are shopping in different markets, and the data says so.

4.1 Clothing (Giyim) — 29 qualified

The most domestic segment in the study: 21 of 29 qualifiers are Turkish, and the top three — The Nest Kids (91.7), TinyLamb (89.5) and Little Gusto (88.4) — are Turkish specialists, all named in roughly 40% of clothing answers. LC Waikiki (#4) is the first mass-market name and Carter’s (#6) the highest foreign one. This is also ebebek’s strongest cut, at #8. Chicco, which leads the study overall, sits at #14 here on an MRR of 0.152, the lowest position score of any qualified entity in the segment.

4.2 Travel & Gear (Seyahat Aksesuarları) — 30 qualified

The mirror image: 17 of 30 qualifiers are foreign, and this is where the international stroller and car-seat houses live. Chicco (98.5), Maxi-Cosi (97.2) and Kraft (93.6) finish within five points of each other, all named in more than 43% of segment answers — the densest top of any cut in the study. Kraft is the standout Turkish performer, and Prego (#6), Bagi (#9), Baby Plus (#11), Babyhope (#12) and Baby2go (#13) give the segment a substantial domestic middle.

4.3 Feeding (Beslenme Ürünleri) — 22 qualified

The most concentrated segment. Philips Avent (91.6) and Lansinoh (84.4) are named in 42% of feeding answers each, and Mamajoo (73.2) is the leading Turkish name. Below them sit two brands that illustrate the position mechanism at its most extreme: OiOi (#5) and Pilsan (#6) are named in only 11.8% and 13.2% of answers, but with MRRs of 0.97 and 0.89 they are named first almost every time they appear. Both are concentrated in two or three product questions — suction plates, weaning sets and high chairs — so read them as category leaders rather than market leaders.

4.4 The finding: almost nothing travels

Sixty-nine distinct entities qualify somewhere across the three segments. Sixty-one of them qualify in exactly one. Only four qualify in all three — Chicco, the retailer ebebek, and the two marketplaces, Trendyol and Hepsiburada — and just four more (Kanz, Kraft, Baby Plus, Elele) qualify in two. Every entity that spans the whole aisle is a shop, except Chicco.

This is the central structural fact of the vertical. “Mother and baby” is a retail aisle, not a market: the assistants treat clothing, travel gear and feeding as three separate recommendation problems with three separate answer sets. A brand’s overall rank therefore says much less here than in a vertical with a single shelf, and a strong general-cut position that comes from one segment should not be read as market-wide presence.

5. Differences Between Models

One-sentence takeaway: All five models name Chicco most or nearly most, but they differ enormously in breadth — Grok names 9.5 brands per answer to GPT-4o-mini’s 4.0 — and one of them answers with shops rather than brands.

5.1 Per-model behaviour summary

Web search was near-universal: three of the five models searched on every answer, Grok on 99.5% and Gemini on 96.4%. The clearest difference is breadth. Grok lists about 9.5 brands per answer and reaches the widest vocabulary (171 distinct entities, 2,081 total mentions), while GPT-4o-mini names 4.0 across 129 entities and 878 mentions — a 2.4× spread in how much of the market a single answer exposes. Claude Haiku (4.4) and Perplexity (5.1) sit near GPT-4o-mini; Gemini (6.4) is in between. Across the run, 1,084 of 1,098 answers named at least one brand.

5.2 Each model’s most-named brands

The tables below give each model’s five most-mentioned entities across its ~220 responses (rate = % of that model’s answers naming the brand).

Chicco is the most-mentioned brand in four of the five models — Claude 28.2%, Gemini 36.5%, GPT-4o-mini 27.7%, Perplexity 31.4% — and fourth in Grok at 29.7%. It is the closest thing to a universal name in the study. Just two brands appear in all five models’ top tens: Chicco and Little Gusto.

The models also disagree about origin. Claude returns seven foreign names in its top ten against three Turkish, and GPT-4o-mini six against four; Grok runs the other way, seven Turkish to three foreign; Perplexity splits evenly. Which assistant a parent asks materially changes whether they hear domestic or imported brands.

5.3 One model answers with shops, not brands

The single largest per-model divergence in the study is not about breadth but about kind of entity. Grok’s three most-named entities are not brands at all — they are places to buy. Trendyol appears in 187 of its 219 answers (85.4%), Hepsiburada in 165 (75.3%) and the specialist retailer ebebek in 93 (42.5%). Chicco, the brand that leads the whole study, is only Grok’s fourth-most-named entity.

No other model behaves this way. Neither Trendyol nor Hepsiburada reaches any other model’s top ten. ebebek is the one retail name that travels — it is GPT-4o-mini’s tenth most-named entity — but at 9.1%, roughly a fifth of the rate Grok names it.

That explains the shape of the general leaderboard. Trendyol’s raw mention count (212) is the second-highest in the study, which would normally place it near the top; a breadth of 4 out of 5 and a low MRR are what hold it to #15. Hepsiburada, with breadth 3, sits at #25. In the Travel & Gear segment both drop to breadth 2 and fall to #25 and #26.

The practical reading is that four of five assistants answer a mother-and-baby question with brands, and one answers substantially with the shops that sell them. Any conclusion about “where AI sends parents to shop” is a conclusion about which assistant the parent used.

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

One-sentence takeaway: 99.1% of answers used web search, so there is no meaningful memory-only segment to compare against — which is itself the finding.

Some Herm.io editions split answers into web-grounded and memory-only segments that produce different leaders. That split does not exist in this run. 1,088 of the 1,098 answers returned at least one citation or search result, and the “own knowledge” segment is ten responses — far too small to support any comparison. Three of the five models (GPT-4o-mini, Claude Haiku, Perplexity) searched on 100% of answers, Grok on 99.5%, and only Gemini fell below, at 96.4%.

Because a separate chapter would compare roughly 1,088 grounded answers against about 10 ungrounded ones, it is left out this edition. The takeaway is that for Turkish mother-and-baby questions these assistants reach for live web content essentially every time. The rankings in this report therefore describe what these systems could retrieve on 26 July 2026, not what they had memorised — which makes them more responsive to changes in web content, and more volatile between editions, than a memory-driven ranking would be. 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: Chicco leads all three question types, but the runners-up rotate completely — The Nest Kids owns discovery, Philips Avent attribute, TinyLamb use-case.

Alongside the product segments, the 44 questions carry three behavioural types: discovery (open “which brands” questions), attribute (specific qualities such as organic fabric, i-Size certification or anti-colic design) and use_case (situation-driven, such as a hospital discharge set or a travel system for flying).

Reading. Chicco leads all three, most heavily on discovery questions (40.4%), where an open “which brands are good” prompt pulls the broadest and most familiar names. But the field behind it reshuffles entirely. On discovery, The Nest Kids is second on a strong 26.7%; on attribute, Philips Avent is second and Dr. Brown’s eighth, both anti-colic feeding specialists that surface when the question names a property; on use_case, TinyLamb is second at 16.0% with the highest position score in the cut (MRR 0.779), because hospital-discharge and newborn-set questions name it first.

Discovery is also by far the widest cut — 41 brands qualify, against 25 on use-case. Open questions produce long, inclusive lists; situation-driven questions produce short, specific ones. For a brand, that means visibility on discovery questions is cheaper to obtain and worth less, while a use-case placement is rarer and more likely to reach a parent who has already decided what they need.

8. When the Question Is Ambiguous

One-sentence takeaway: Two questions in the set are genuinely ambiguous in Turkish, and which market the assistant thinks it is answering depends entirely on which assistant you ask.

The questions in this study are written the way people type them, which means they occasionally contain the same ambiguities that real queries do. Two did. Rather than rewrite them — which would have measured a market nobody actually queries — both were retained in every cut, and the entities from the adjacent industries were marked out of scope so they never reach a leaderboard. What follows is a result, not a defect.

8.1 “Oto koltuğu” — child seat or vehicle seat?

Question 9 asks “iyi bir yerli oto koltuğu markası var mı” — “is there a good Turkish car-seat brand?” Without a child-specific qualifier, oto koltuğu reads just as naturally as automotive seating. Of 25 responses, 10 returned commercial-vehicle seat manufacturers, 6 were mixed, and only 9 were clean child-seat answers.

The control sits inside the study. Question 22 asks the same thing with one standards term added — “i-Size güvenlikli yerli oto koltuğu var mı” — and is 0 of 25 contaminated. A single certification keyword fully disambiguates the query.

8.2 “Süt sağma makinesi” — breast pump or milking machine?

Question 42 asks “çalışan anne için sessiz yerli süt sağma makinesi var mı” — “is there a quiet Turkish breast pump for a working mother?” In Turkish, süt sağma makinesi is used both for a breast pump and for livestock milking equipment. Despite the phrase “çalışan anne” (working mother) in the question, the models resolved it very differently:

Seventeen of the 25 responses are genuine breast-pump answers. Several models explicitly asked which sense the user meant before answering.

8.3 What this means for a brand

The implication is direct: assistant choice determines which market a colloquial Turkish query lands in. A brand competing on ambiguous vocabulary is not competing on the same terms across assistants — on question 42, Perplexity showed a parent no baby brands at all, while Claude and GPT-4o-mini showed only baby brands.

The fix is visible in the data. Question 22 shows that a single standards or category term — i-Size, bebek, anne — collapses the ambiguity completely. Brands whose category vocabulary overlaps with an adjacent industry have an unusually cheap intervention available: make sure the disambiguating term appears alongside the product name wherever the assistants read about it.

9. Open Market: Turkish vs. Foreign

One-sentence takeaway: On origin-neutral questions the shelf tilts foreign — 54.5% of mentions and 21 of 37 qualified entities — while the yerli questions return an almost entirely Turkish shelf.

This chapter relies only on the 23 open questions with no origin restriction (574 responses), the only fair basis for comparing Turkish and foreign brands.

9.1 Origin share

Foreign brands take 2,302 mentions to 1,924 Turkish — 54.5% against 45.5%. But breadth points the other way: 106 distinct Turkish entities surface in the open market against 74 foreign. Turkish brands are more numerous and individually less visible; foreign brands are fewer and more concentrated. That is the signature of a market where a handful of international houses hold the recognised positions while a long domestic tail competes below them.

9.2 Open-market top 15 brands (any origin)

9.3 The pattern: foreign leads gear, Turkish leads clothing

The individual ranking shows what the aggregate hides. The open-market top three are foreign — Chicco, Philips Avent, Maxi-Cosi — and they are a stroller-and-car-seat house and two feeding houses. The first Turkish name is The Nest Kids at #4, followed by TinyLamb (#5) and Little Gusto (#7), all three baby-clothing specialists.

That is the same split visible in Chapter 4, arriving from a different direction. Where the product is safety-regulated and capital-intensive — car seats, strollers, bottles and pumps — the assistants reach for international brands. Where it is apparel, they reach for Turkish ones, and the Clothing segment is 20 of 28 domestic. Turkey’s strength in the mother-and-baby aisle, as these systems describe it, is textile.

The local cut confirms the assistants can distinguish the constraint: asked for yerli brands, they return a 15-of-19 Turkish shelf led by Kraft. So the foreign tilt on open questions is a genuine market signal about what these systems retrieve by default, not an artefact of the instrument.

10. The Discovery Ecosystem: Where AI Learns About Brands

One-sentence takeaway: AI’s picture of this market rests on specialist retailers and editorial content as much as on brand sites — e-bebek.com is the single most-cited domain, and editorial carries 20.6% of citations.

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 mother and baby brands. Across the run the models cited 484 distinct domains.

10.1 The most-cited domains

The most-cited domain is e-bebek.com, a specialist mother-and-baby retailer, cited by 29.6% of all responses — ahead of the general marketplace trendyol.com (28.5%). A category specialist, not a general marketplace or a brand’s own site, is where these assistants read about this market most often.

Two of the next three are editorial: anneysen.com (19.6%) and onedio.com (16.0%), followed by eniyisinde.com.tr (13.5%) and eniyilerden.com (10.2%). A parenting forum, annelertoplandik.com, sits at #9 with 12.7%. Brand-owned domains appear from #5 — thenestkids.com.tr (15.0%), littlegusto.com.tr (14.9%), tinylamb.co (12.3%) — and 34 of the top 50 domains were cited by all five models.

10.2 Source-type mix

10.3 The editorial finding

The striking pattern is how little of this market’s evidence base is brand-controlled. Brand-owned sites account for 37.4% of citations, while editorial and media carry 20.6%, retailers 10.2%, marketplaces 9.7% and forums and social 6.0%. Better than a quarter of the citation base — 26.6% — is third-party commentary and community content.

That distribution says something about how the category is researched. Baby products are bought on comparison rather than on sight — parents read about anti-colic valves, i-Size approval and fabric certification before they buy — and the citation mix mirrors that. Four of the twelve most-cited domains are editorial, and a parenting forum sits at #9. The assistants are reading what parents read.

There is a visible correlation between own-site citation and segment standing, but only in one segment. The three most-cited brand domains — thenestkids.com.tr, littlegusto.com.tr and tinylamb.co — belong to exactly the three brands topping the Clothing leaderboard. Yet Philips Avent, which leads Feeding and ranks second overall, is cited on philips.com.tr in only 5.0% of responses. Its visibility is carried by retailers and editorial coverage, not by its own domain.

Read as description rather than advice: in this vertical, being written about appears to matter at least as much as publishing well.

11. What the Patterns Suggest

One-sentence takeaway: The most visible brands share broad model coverage and a defined product segment; a strong own-site presence turns out to be neither necessary nor sufficient.

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 or safety.

1) Breadth is the entry condition. 28 of the 30 qualified general-cut entities appear in all five models, and the only two exceptions are marketplaces. Below the qualification line, breadth is what separates a brand that ranks from one that does not: in the Travel & Gear segment, UPPAbaby, Joolz and Nuna all clear the 5% mention bar but sit in the bottom four because only two or three models know them.

2) Position and frequency are independent signals. Chicco is named around 1.6 times as often as the next entity but ranks first by a smaller margin than that implies, because it is named mid-list. Philips Avent, OiOi and Pilsan are named far less often but almost always first. A brand can lead a segment either way, and the two profiles suggest different situations — high mention rate reads as general familiarity, high MRR as being the obvious answer to a specific question.

3) Segment definition beats overall presence. With 61 of 69 segment qualifiers appearing in only one segment, the brands that rank are the ones the assistants can attach to a specific product problem. The three Clothing leaders are pure baby-apparel specialists; the Travel & Gear leaders are stroller and car-seat houses. Breadth of catalogue does not appear to help; legibility of category does.

4) The evidence base is third-party. With 99.1% of answers web-grounded, brand-owned sites at 37.4% and editorial plus forums at 26.6%, more than a quarter of what these assistants read is content no brand controls. The clothing segment shows own-site citation tracking with rank, but Philips Avent shows it is not required. Where a category is researched through comparison content, presence in that content is closely associated with visibility.

5) Origin is contested, and the question decides it. 54.5% foreign on open questions, 15 of 19 Turkish on local ones. Both are true, and quoting either without its cut is misleading. For a Turkish brand the practical reading is that yerli framing is a reliable route to visibility, and origin-neutral discovery is where the contest actually happens.

6) One assistant is a different market. Grok names more than twice as many brands per answer, puts three retailers and marketplaces above every brand, and skews domestic; GPT-4o-mini names the fewest and skews foreign. A brand invisible in one may be prominent in another, and a single-assistant check will mislead in either direction.

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; which segment it qualifies in, if any; how often it is mentioned within that segment; and how early it is named when mentioned. Those four readings locate a visibility gap without saying anything about whether the product is good.

12. 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, point-in-time measurement. Collected in one run on 26 July 2026, a snapshot rather than a trend. With 99.1% of answers web-grounded, movement in the search index will move these rankings. This is the first (baseline) edition; quarterly editions will enable trend analysis.
• Never compare Turkish and foreign brands across all questions. Twenty-one of the 44 questions explicitly constrain the answer to Turkish brands. Chapter 9’s open cut is the only valid basis for that comparison; the general-cut origin split is a property of the instrument, not the market.
• Small-sample position inflation. A few brands clear the 5% bar but rank high mainly on position, on thin mentions. Naynes (#10 Clothing) qualifies on 6.4% mentions with the highest MRR in its cut and a breadth of 3; Esas Tekstil (#7 Clothing) on 5.9%. OiOi and Pilsan rank high in Feeding on visibility concentrated in two or three product questions and should be read as category leaders, not market leaders. Read these ranks cautiously.
• Small score gaps are not meaningful. In the packed middles of these leaderboards — particularly Travel & Gear, where the top three finish within five points — differences of a point or two should not be over-interpreted.
• Model-specific visibility. Seven qualifying entities are named by fewer than three of the five models: Akakçe, Andy, Hepsiburada, Joolz, LC Waikiki, Trendyol and UPPAbaby. Their standing is a property of one or two assistants and should not be read as market-wide.
• Do not compare distinct-brand counts across segments. Feeding carries 14 questions against 15 for the others, and four discovery questions against five. It is the smallest segment at 349 responses — still well above the ten-question floor, but carrying more sampling noise.
• Sub-brands under-count structurally. Where a model names a parent rather than a product line, the line records nothing. This affects licensed and sub-branded ranges most.
• Some near-duplicate entities remain unresolved. Kinderkraft and Kraft are genuinely different companies and were deliberately kept apart. Civil and Civilim, Huggy and Huggyberry, and Bebbek / Nenny Bebbek / Nenny Baby were left separate after inspection but could not be resolved with confidence and may each represent a single commercial entity.
• Visibility is not quality or safety. This report measures only mention and position. It does not assess whether a brand was described accurately, whether a product meets any safety standard, or whether a recommendation was sound. In a category containing car seats and feeding equipment this distinction matters: a score is not a safety signal and 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 and should be treated as indicative. Citation payloads also differ by provider: Gemini’s real source domains are read from its search-result titles rather than its redirect URLs, and Grok returns unusable citation titles, so its domains are taken from URLs.
• Origin classification involves judgment. Origin follows the trademark, so brands manufactured or licensed in Turkey under a foreign mark — Kanz, Carter’s, OshKosh, Pierre Cardin, Bambino — are recorded as foreign. Eight entities whose trademark country could not be established are excluded from origin shares.
• Marketplaces and retailers are included as entities per study scope, and compete differently from direct brands; §3.2 gives a marketplace-free view and explains why ebebek, which is retailer, marketplace and own-label brand at once, is kept in it.
• 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

13. Appendix & Data

13.1 All questions (44) and their categories

Distribution: 14 discovery, 15 attribute, 15 use_case; 15 Clothing, 15 Travel & Gear, 14 Feeding; 21 local and 23 open.

13.2 All tracked brands and retailers (226)

13.3 Data availability

The study tracked 233 entities — 226 brands and retailers, 6 marketplaces and 1 out-of-scope entity. Thirty cleared the 5% threshold in the general cut, with 19 qualifying in the local cut and 37 in the open cut. To support scrutiny and reproduction, the underlying data is available on request: response-level brand mentions, all nine qualified leaderboards, full unfiltered entity metrics, per-model and per-question-type breakdowns, the open-market breakdown, the full 484-domain source list, and the methodology and data dictionary.

All figures reflect a single point-in-time run (26 July 2026) and are subject to the limitations in Chapter 12. This is the first (baseline) measurement; the study will be repeated quarterly.

13.4 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 or safety; 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.