PublishedVerifiedEvery 6 weeksSources6 namedAuthored bySquareRank Team
Bakers · § 4.17.1 · How-to · The wedge
Baker AI Search Citations
A parent typing "vegan birthday cake brooklyn under $300" into ChatGPT or a couple asking Perplexity "who makes the best custom wedding cakes in the Hudson Valley" gets back three to six named cake artists with style notes and inquiry guidance1. The list is not random. It is the model's best guess at the closest specialty cake artist for the [city] [style] [dietary] combination — and the artist whose Squarespace site ships Bakery schema, a [city] [style] content page, and a populated sameAs graph linking the Instagram bio is the artist the model reaches for.
This leaf is the citation install for custom-cake bakers on Squarespace 7.1 — the [city] [style] page template, the Bakery + Product schema graph, the Instagram cross-link pattern that ties the Squarespace entity to the off-platform discovery surface, and the 5-step build. It covers how buyers actually ask AI engines for a custom-cake artist, why specialty bakers cite disproportionately well in the food-vertical answer surface, the page template that lands citation inside 2-6 weeks of publication, and the honest measurement reality.
§01Query shape
How buyers actually ask AI engines for a custom-cake artist
The query shape buyers send to ChatGPT, Claude, and Perplexity for custom cakes is consistent across the three engines that drive most of this discovery traffic. The buyer types a conversational query that names the city, the style or occasion, and a dietary or budget constraint where one applies — 'vegan birthday cake brooklyn under $300', 'custom wedding cake hudson valley buttercream', 'gluten-free smash cake denver'. The model parses the constraints and answers with a list of three to six named bakers, often with one-line style notes and inquiry guidance. The specialty artist who matched their content shape to the conversational query wins this surface; the retail bakery optimised for 'bakery near me' does not.
The pattern is consistent and the asymmetry favours specialty artists. Retail bakeries with a wide-but-shallow product catalog tend to write generic homepage and category copy that the model finds nothing concrete to attribute. A custom-cake artist with a clearly named specialism (the styles, the dietary specialties, the served cities) and a content page per primary [city] [style] combination is materially easier to cite, because the citation has somewhere specific to land. The model picks the closest specialist. The artist who installs the pattern below is competing in a surface the wide-catalog retail bakery structurally cannot dominate.
The food-vertical AI-routing math for cake artists
800M
weekly ChatGPT users routing discovery queries, including custom-cake and dietary-specialty recommendations.
Why specialty bakers cite disproportionately well in food-vertical AI answers
Three structural features of the custom-cake business make AI engines treat specialty bakers as good citation candidates more readily than they treat retail bakeries. First, the buyer's question is constraint-rich (style, dietary, city, budget) and the specialty baker's content is constraint-organised in the same way, which matches the model's preferred extraction shape. Second, the personal-brand model means each cake artist's portfolio is anchored to a named human entity AI can disambiguate via the Instagram cross-link in sameAs. Third, the food-vertical local pack on Google is already image-rich, which signals to the model that cake-vertical recommendations should be visually-anchored — pointing back at specialty artists with portfolios over retail chains with stock imagery.
The third point compounds the others. When the model is deciding whether to surface a recommendation for "vegan birthday cake brooklyn", it is implicitly weighing whether the site it cites can be visually verified by the user after clicking through. A specialty artist with an organised portfolio gallery passes that test; a retail bakery whose page surfaces a stock cake image fails it. The model picks the answer the user will not regret clicking. The specialty artist with [city] [style] content, real portfolio imagery, and a verifiable Instagram cross-link reads, to the model, like the closest thing to a trusted local recommendation. That mechanism is the wedge.
The mechanism is the same one that drives citation in any other specialty service vertical, but the food-vertical asymmetry is sharper because visual verification is non-negotiable for the buyer. A buyer choosing a wedding cake or a child's birthday cake will not book without seeing the artist's prior work, which puts visual portfolio depth on the must-have list and puts retail-style stock imagery on the disqualifier list. The honest framing: this is one of the few verticals where AI citation and Instagram strength reinforce each other, because the model is implicitly checking the Instagram-shaped quality of the visual portfolio before recommending.
§03The page
The [city] [style] page template — the citation hook for custom-cake queries
Every custom-cake artist who wants to be cited in AI answers needs three to five dedicated content pages, each targeting one [city] [style] or [dietary] [city] combination the artist actually serves. The page is structured as a working answer to the conversational query: the title carries the query verbatim ('Vegan Birthday Cakes in Brooklyn'), the first 200 words are a 134-167 word self-contained answer that names the style, the dietary specialism, the typical price range, and the inquiry flow. The rest of the page goes deeper — typical timeline from inquiry to delivery, what the consultation covers, three to five portfolio images with descriptive alt text. 1,000 words total.
The structural template is consistent across [city] [style] and [dietary] [city] variations. H1 is the conversational query verbatim. The first paragraph answers the question in 30-50 words — short enough for a citation card, complete enough to stand alone. The next paragraph (the 134-167 word band) expands with the practical details: typical guest count range served, typical price range, typical lead time, typical inquiry process, dietary substitutions available. One or two named adjacent sources — a local food publication that has reviewed the bakery, a Squarespace Help reference for the inquiry-form pattern, a dietary-authority reference for the allergen handling — ground the page in citation-grade attribution.
The page count is the underbuilt half of most baker sites. A specialty artist serving three primary cities and four primary styles can ship up to twelve [city] [style] pages — and a thoughtful baker ships the three to five combinations that produce the highest-converting inquiries first, then expands the matrix over the next two quarters. The pages reinforce each other in the internal linking graph: each [city] [style] page links to the adjacent [city] [other-style] page and to the master portfolio gallery, building the topic-cluster shape that compounds rank for the long-tail query universe.
§04Entity wiring
Bakery + Person schema and the Instagram cross-link in sameAs
The entity graph for a custom-cake artist on Squarespace runs Bakery schema sitewide (with serviceType carrying the specialism), Person schema on the founder or about page (with knowsAbout carrying the style and dietary specialism tags), and sameAs linking Instagram, Pinterest, Google Business Profile, and any local press features. The Instagram cross-link is the most underbuilt of the three on most baker sites and the most consequential for AI citation, because the model uses sameAs to verify that the Squarespace entity is the same business the buyer is seeing on the Instagram surface they discovered the artist through.
The Person schema5 knowsAbout array should mirror the style and dietary terms the artist actually works in. Specific tags ("Buttercream wedding cakes", "Vegan birthday cakes", "Gluten-free smash cakes", "Fondant sculpted cakes") carry materially more disambiguation weight than generic ones ("Cakes", "Baking"). The same tags also drive the topic-cluster internal-linking pattern from the [city] [style] pages, so investing in specificity early compounds across the next quarter's content work.
The sameAs array5 is the cross-platform verification surface AI engines read to confirm a single entity across multiple surfaces. The Instagram link is the most important entry for cake artists specifically, because Instagram is the upstream discovery surface and the buyer is almost always coming from there. Add the Pinterest link, the Google Business Profile link, and any local press features — local food magazines, wedding publications, dietary-authority writeups. Each verified surface strengthens the model's confidence that the artist is a real, attributable specialist rather than a content-only site.
JSON-LDPerson schema for a custom-cake artist — Page Settings > Code Injection > Header on /founder/
Pair this Person entity with the sitewide Bakery schema4 shown on the bakers hub, with worksFor pointing back at the bakery. Code Injection is gated to Business plan and above on Squarespace; on Personal plan the workaround is the in-body author bio with the same content, which is less effective but still readable to AI parsers.
§05The install
The 5-step install, in order
The install runs in five sequential steps. Crawler access first, because the model has to be able to read your site. Schema graph second, because Bakery + Product schema is the food-vertical semantic anchor. [city] [style] content pages third, because those pages are the citation hooks for the conversational queries. Entity graph fourth, because the Instagram cross-link in sameAs is the verification surface. Manual tracking fifth, because food-vertical AI traffic mostly arrives without referrer data and the only honest measurement is direct query logging.
Step 1 — Crawler access. Open Settings > Crawlers in Squarespace. Confirm the AI exclusion toggle is unchecked3. Verify in a private window that yourbakery.com/robots.txt does not disallow GPTBot. ChatGPT-User does not follow robots.txt2 but OAI-SearchBot does, and OAI-SearchBot decides ChatGPT Search citations.
Step 2 — Schema graph. Inject Bakery schema sitewide via Settings > Advanced > Code Injection. Pair with Product schema per tier on Commerce-style catalog pages or Service schema on inquiry-page offerings. The full Bakery + Product example lives on the bakers hub.
Step 3 — [city] [style] content pages. Ship three to five pages, one per primary [city] [style] or [dietary] [city] combination you actually serve. 1,000 words each. Title carries the query verbatim, first 200 words carry the citation-shaped answer, two named adjacent sources.
Step 4 — Entity graph. Inject Person schema with knowsAbout and sameAs on /founder/ or /about/. The Instagram link is the most consequential single sameAs entry for cake artists; add Pinterest, Google Business Profile, and any local press features.
Step 5 — Manual tracking. Build a 10-query tracking spreadsheet across [city] [style] and [dietary] [city] combinations. Run weekly across ChatGPT, Claude, and Perplexity. Log results. Pair with the discovery-call question on every inbound inquiry. The full GA4 setup lives in the dark-traffic leaf.
§06Measurement
Measuring whether AI engines are actually citing your bakery
AI-citation measurement for custom-cake bakers is harder than Google measurement because the routing is conversational, most cited visits arrive without referrer data, and mobile-app traffic strips referrers entirely. The honest 2026 stack is a manual query log run weekly across the engines that matter, a GA4 custom channel grouping for the visibly tagged subset, and a discipline of asking every inbound inquiry where they heard about the bakery. The discovery-call question is the most reliable single signal because food-vertical AI traffic almost always names ChatGPT, Perplexity, or Instagram explicitly in the answer.
The manual query log is the floor. Pick 10 conversational queries that map to your styles, dietary specialisms, and cities — "vegan birthday cake brooklyn under $300", "custom wedding cake hudson valley buttercream", "gluten-free smash cake denver near me", "where can I get a fondant sculpted cake in portland". Run each one through ChatGPT, Claude, and Perplexity on Friday morning. Log whether your bakery name surfaces. Screenshot when you appear. Quarterly review aggregates the pattern.
The GA4 layer captures the visible fraction. Create a custom channel grouping called AI Search using a regex that matches chatgpt.com|chat.openai.com|perplexity.ai|claude.ai. OpenAI started tagging some citation links with utm_source=chatgpt.com in 2024 and extended the tagging in mid-2025; conversational inline links remain untagged. Expect GA4 to undercount actual AI traffic by an order of magnitude for food-vertical queries specifically, because mobile-app routing is the dominant pattern for buyers shopping for cakes on their phone.
