Short answer: AI engines cite pages they can extract, verify, and attach to a real entity. On Framer that means a quotable answer under the question, FAQ and Organization schema that matches visible copy, consistent names (brand, people, place), and `sameAs` links to profiles you actually own — LinkedIn, YouTube, Wikidata if it exists. The AEO scanner is a diagnostic. It is not a citation.
We wrote the primer as Framer AEO: Optimizing Your Site for AI Search. This is the layer underneath: structured data, entity linking, and the citation hygiene that primer assumed. The service that implements it is SEO optimization.
AEO vs SEO, one more time, then we move
SEO still earns the crawl: titles, speed, links, indexability. Framer SEO in 2026 is still the checklist. AEO earns the quote: a passage an assistant can lift without hallucinating who said it.
Overlap is large. Keyword stuffing does not get you cited. A self-contained answer, a named studio in Fountain Hills, AZ, and schema that points at real profiles — that is the job. Framer’s HTML is already parseable. Stop publishing essays with no answer and no identity.
What AI engines are actually looking for
We cannot see an assistant’s notes. We can see what survives contact with them:
Extractable answers. First paragraph under a question-shaped H2 is a complete sentence, not “It depends” followed by 400 words of throat-clearing.
Visible structure. Headings, lists, FAQs. If the answer only exists in a screenshot or a WebGL hero, it does not exist.
Named entities. “PixlForm,” “Connor ONeil,” “Framer Expert,” “Fountain Hills” — same spelling everywhere. Not “Pixl Form” on page two.
Corroboration. The same facts on your site, your LinkedIn, your YouTube About, your Google Business Profile. Engines notice disagreement more than they notice clever copy.
Dates and first-hand detail. “We ship in an average of 14 days” beats “we deliver rapidly.” We will not fabricate a citation rate to make this post feel scientific.
Do Framer websites rank on Google? is the ranking question. This post is the attribution question.
Schema that is worth putting on a Framer site
Framer does not have a full built-in schema studio. JSON-LD goes in Custom Code → Head, site-wide or per page. CMS detail pages can pull fields with `{{Title | json}}` so every post is not the same blob. Marketplace plugins exist if you want a form. We still read the output.
Google’s own docs (and Framer’s structured-data guide) are blunt: markup does not guarantee a rich result, a featured snippet, or an AI citation. FAQPage in particular no longer powers a Google FAQ dropdown. We still use it because assistants parse Q&A pairs. We do not promise a SERP toy that Google retired.
Use types that describe visible content:
Type | Where it belongs | Why it helps AEO |
|---|---|---|
`Organization` or `LocalBusiness` | Site-wide | Identity: name, url, logo, area, `sameAs` |
`Person` | About, authors | Ties Connor (or your founder) to the org |
`WebSite` + `SearchAction` | Optional | Sitename, search box if you have one |
`Article` | Blog CMS template | Headline, datePublished, author, image |
`FAQPage` | Pages with a real FAQ | Packaged Q&A for extractors |
`BreadcrumbList` | Nested URLs | Hierarchy, less ambiguous paths |
`Service` | Service pages | What you actually sell |
Do not stamp `Product` + fake `AggregateRating` on a studio homepage. Do not mark up FAQs that are not on the page. Inconsistent schema is worse than none — it is a lying machine-readable layer.
A tight Organization pattern
Keep it boring and true. High-level, not a dump of every Schema.org key:
`@id` for the org (your domain plus `#organization`) so other blocks can point at it
`name`, `url`, `logo`, `description` that match the homepage
`founder` as a `Person` with their own `url` / LinkedIn
`address` as locality and region — city and state, not a street number you do not want public
`sameAs` array of canonical profile URLs
`sameAs` is the AEO-specific part. It tells engines: this entity is the same as these other nodes.
sameAs, Wikidata, and YouTube (the part the primer skipped)
Assistants and Knowledge Graph-style systems resolve brands by corroborating URLs, not by your tagline. `sameAs` is how you hand them the corroboration.
Include only profiles you control and that are alive:
LinkedIn company page (canonical `/company/…` URL)
X / Instagram / GitHub if you actually post
YouTube channel URL (`/channel/…` or `@handle`, not a random upload)
Crunchbase if it exists and is yours
Wikidata item (`https://www.wikidata.org/wiki/Q…`) if one exists
Wikipedia article only if it actually exists — do not invent a wiki URL
Skip dead Facebook pages and directory spam. Three clean links beat eleven stale ones.
Wikidata, without the LARPing
Wikidata is a public knowledge graph a lot of systems read. If your organization already has an item, add it to `sameAs` and keep labels consistent with your site (name, instance of, official website, inception, headquarters locality).
If you do *not* have an item, do not treat “create a Q-number” as a growth hack. Empty, self-sourced items get ignored or deleted. For most studios the honest path is: consistent NAP and website, LinkedIn, YouTube, then Wikidata when there is actually something to describe. We are not going to pretend a new Wikidata row is a ranking lever we can sell.
YouTube as a citation surface
AI engines quote video transcripts. That does not mean “start a media company.” It means:
One channel, same brand name as the site
That channel URL in `sameAs`
Videos that answer the same questions as your key pages (pricing, migration, what a Framer Expert does) with a spoken short answer up front
A description + title that use the entity name, not only a joke
Optionally `VideoObject` schema on the page that embeds the video, matching the visible title
A ramble with no on-screen answer is not AEO. A short video that matches the blog post is. YouTube does not replace the article; it gives engines a second attributable source with the same facts.
FAQs that get quoted (and FAQs that waste HTML)
A useful FAQ is four questions a buyer actually types, with answers that stand alone if the heading is stripped.
Write answers like pull quotes:
“Landing pages start at $2,750. Multi-page sites start at $4,750.”
“Average delivery across our projects is 14 days.”
“We are an Official Framer Expert studio in Fountain Hills, AZ.”
Do not:
Repeat the H2 inside the answer with no payload
Answer “It depends” and then hide the number
Mark up a FAQ that is visually collapsed into unreadable accordion garbage *and* different text in JSON-LD
Framer implementation: visible FAQ on the page, then JSON-LD that copies those strings. CMS-bound FAQs should generate both the UI and the schema from the same fields so they cannot drift.
Run Framer’s AEO scanner on the money pages after. Treat flags as a punch list (missing answer-first paragraph, weak headings, thin entity naming). Re-scan. It is PageSpeed for answerability, not a trophy.
On-page habits that make schema less lonely
Schema without the page is a costume.
One H1, question H2s where the query is a question
Short answer near the top (you are reading one)
Authors with a real Person page or a consistent byline — “PixlForm Team” is fine if that is true; random ghost writers are not
Internal links to the service you want cited (website migration, full websites, this SEO page)
Updated dates on posts that still earn queries
The same stats everywhere: 100+ sites, 14-day average, 10+ industries, 3M+ visitors. Pick the numbers you will defend. Repeat them. Do not rotate in a new percentage for a pitch deck.
How we implement this at PixlForm
On client builds, AEO is not a plugin we toggle at the end. It is:
Information architecture with one intent per URL
Answer-first copy on the pages that can actually win a question
Organization + Article/FAQ JSON-LD that matches the canvas
`sameAs` collected in discovery (send us the URLs)
Scanner pass + Search Console after launch
We do not sell “we will get you cited in ChatGPT this quarter.” We sell a site that is easy to quote and hard to confuse with someone else.
FAQ
Does FAQ schema still matter if Google killed FAQ rich results? For the old SERP dropdown, mostly no. For AI extraction, yes — if the questions and answers are visible on the page and the JSON-LD matches.
Do I need a Wikipedia page to show up in AI answers? No. Wikipedia helps when it exists and is accurate. Most businesses will never have one. Consistent site + LinkedIn + YouTube + honest schema is the adult path.
Can Framer add schema without code? Not fully native. Custom Code in the head, CMS variables, or a marketplace plugin. Someone still has to make sure it is true.
Will the AEO scanner get me cited? It will tell you if a page is hard to extract. Fixing that is necessary, not sufficient. Citations still need demand, uniqueness, and a resolvable entity.
If you want this implemented on a live Framer site — schema, FAQs, entity links, not a vibe — book a 15-minute intro or email c@pixlform.com. Start at SEO optimization.