Copilot SEO: How to Get Cited in Microsoft Copilot Answers

Copilot SEO is the work of getting your pages used as cited sources inside Microsoft Copilot's answers — and the single most useful fact about it is that there is no separate Copilot index to get into. Microsoft states that experiences like Copilot are “Powered by Bing's search index”. Everything follows from that. If a page is not in Bing's index, Copilot cannot ground an answer on it; if it is, the remaining questions are whether Microsoft's crawlers are allowed to use it, whether your snippet controls permit it to be quoted, and whether the page is written in pieces an assistant can lift. Since February 2026 you no longer have to guess at the outcome: Bing Webmaster Tools reports, in public preview, which of your URLs are cited in Copilot answers and which phrases retrieved them.

What Copilot SEO actually means

The phrase is doing two jobs at once, and it is worth separating them before spending any money. Search “copilot seo” and most of page one is products literally named SEO Copilot — AI writing and auditing tools that happen to own the brand name. That is a different market from the one this page is about. Here, Copilot SEO means optimising your own site so that Microsoft Copilot cites it, which is also what people mean when they search for seo for copilot, bing copilot seo or microsoft copilot seo.

Framed that way, the job is narrower than it sounds. There is no Copilot submission form, no Copilot-only ranking factor Microsoft has published, and no schema type that buys admission. What there is: an index, a set of access and display controls, a documented set of content-structure preferences, and — new since February 2026 — a citation report. The rest of this page works through those four in order, because that is the order in which they can actually stop you.

Does Copilot use Bing?

Yes, for web grounding, and Microsoft says so in its own words. Microsoft Advertising's guidance on AI search opens with the line: “Powered by Bing's search index, experiences like Microsoft Copilot, Microsoft Start, and others handle billions of queries each month, embedding Bing deeply into how people search, shop, and explore online.” (Documented — Microsoft Advertising, “Optimizing Your Content for Inclusion in AI Search Answers”, October 2025.)

The Bing team says the same thing from the other direction when it describes its snippet controls as governing “Bing Search and Copilot experiences powered by Bing”. (Documented — Bing Webmaster Blog, “Bing Introduces Support for the data-nosnippet HTML Attribute”.) Two separate Microsoft product teams, describing the same dependency. Put plainly: what people call copilot ai search is Bing retrieval with a language model composing the answer on top of what it retrieves.

The practical consequence is unglamorous and load-bearing: Bing indexation is the entry ticket. Plenty of sites that have spent a decade on Google have never once opened Bing Webmaster Tools, have never checked whether their pages are indexed there, and are surprised to find their Copilot visibility is zero. That is not an AI problem. It is an indexation problem wearing an AI costume, and it is fixed with the same sitemap and crawlability work as any other index.

The surfaces, and what admits or removes you from each

“Copilot” is a brand covering several places an answer can appear. Bing's AI Performance documentation groups the ones a publisher can be cited in as Microsoft Copilot, AI-generated summaries in Bing, and select partner integrations. They share the index, so they share the entry condition, but the display controls bite differently.

Surface What admits you What removes you
Microsoft Copilot answers Being in Bing's index and permitted to be quoted noindex; a robots.txt disallow; nosnippet; data-nosnippet on the specific block
AI-generated summaries in Bing Same index, same snippet permissions Same directives; Bing states it “respects all content owner preferences expressed through robots.txt and other supported control mechanisms”
Partner integrations using Bing grounding Same index Same directives, applied upstream at the index
Local and “near me” answers Index plus accurate business data; Microsoft names Bing Places for Business Stale or absent listing data

Note what is not in that table: a separate opt-in, a Copilot-specific crawler you must allow-list, or a paid inclusion route. Microsoft has published none of those.

The controls Microsoft documents

The one control here that genuinely rewards reading the documentation is data-nosnippet, which Bing adopted in October 2025. It does something the other directives cannot: it removes a specific block of a page from snippets and AI answers while leaving that block indexed and available for ranking. Bing's wording is exact — “any content marked with data-nosnippet is still indexed normally, but it will be excluded from snippets and AI summaries. The content tagged as data-nosnippet is available for ranking.”

Directive Scope Effect on Copilot answers
noindex Whole page Page cannot be cited at all — it is not in the index
nosnippet Whole page Blocks all snippet text and preview thumbnails
data-nosnippet One HTML element That block is excluded from snippets and AI summaries; still indexed, still ranks
max-snippet, max-image-preview, max-video-preview Whole page Caps the size or duration of preview content

This is the sharpest instrument on the page, and it cuts both ways. It is the honest answer for paywalled or legally required text you do not want quoted back at you. It is also, applied carelessly to a cookie banner wrapper or a whole content <div>, a fast way to make a page that ranks and is never quoted. If your Copilot citations fall off a cliff after a template change, this is the first thing to grep the template for.

Documented, versus widely believed

Widely believed What Microsoft actually documents
Copilot is a separate engine that needs its own optimisation programme Copilot experiences are “Powered by Bing's search index”. Its own guidance says “traditional SEO fundamentals still matter. Crawlability, metadata, internal linking, and backlinks remain essential”
There is a special AI file or AI schema that gets you cited No such requirement is published. Microsoft's structural advice is ordinary: aligned title, H1 and description; real H2s and H3s; Q&A pairs; lists and tables; schema.org types that describe the content
Getting cited is a ranking position problem Microsoft describes selection at the passage level: assistants “break content down, a process called parsing, into smaller, structured pieces that can be evaluated for authority and relevance”, then assemble answers “often drawing from multiple sources”
Blocking snippets protects you without cost True for ranking, false for citation. data-nosnippet keeps the block indexed and rankable and removes it from AI answers — that is the documented trade, and it is a trade

That third row is the one worth sitting with, because it changes the unit of work. If answers are assembled from passages rather than pages, then a 3,000-word page with one clean, self-contained paragraph on the question is a better citation candidate than a 3,000-word page that answers the question across five paragraphs and a diagram. Ranking gets you considered. Structure gets you lifted.

How to rank in Copilot

Microsoft is unusually direct about what it wants, so this section is a summary of published guidance rather than folklore. The honest headline is that there is no trick: as Microsoft puts it, “there's no secret strategy for being selected by AI systems”. What follows is what its own product managers tell publishers to do, which is the closest thing to an answer for how to rank in copilot — or, if you prefer the engine-anchored phrasing, how to rank in bing copilot — that exists.

  1. Get indexed in Bing first. Verify the site in Bing Webmaster Tools, submit a sitemap, and fix crawl errors. Nothing below matters until this is true.
  2. Align title, H1 and description. Microsoft calls these “important signals AI systems use to interpret purpose and scope”, and says consistent alignment between them “improves both discoverability and confidence signals for AI systems”.
  3. Use headings as real boundaries. Its guidance describes H2s and H3s as marking “where one idea ends and another begins”, acting “like chapter titles that define clear content slices”. Vague headings cost you the slice.
  4. Write question-and-answer pairs. Microsoft notes assistants “can often lift these pairs word for word into AI-generated responses”. This is the cheapest available win and most sites skip it — our note on FAQ content covers the format.
  5. Use lists and tables for anything comparative. They “break complex details into clean, reusable segments”.
  6. Make sentences self-contained. The published criterion for snippet eligibility is phrasing that “make sense even when pulled out of context”. Pronouns that refer back three paragraphs are the enemy here.
  7. Do not hide the answer. Microsoft warns against putting key content in tabs or expandable menus, because “AI systems may not render hidden content”; against relying on PDFs for core information; and against putting key information only in images.
  8. Add schema that describes what the page is. JSON-LD marking a page as an article, product, review, FAQ or event “turns plain text into structured data that machines can interpret with confidence”.
  9. Anchor every claim. The AI Performance guidance asks publishers to “support claims with evidence” — examples, data and cited sources — which is also, not coincidentally, what E-E-A-T has always asked for.
  10. Keep it fresh, and say so quickly. Microsoft recommends IndexNow to notify participating engines “whenever content is added, updated, or removed”, so that AI systems reference the current version of a page.

Read as a list, that is not an AI strategy. It is competent on-page work with the parsing step made explicit — which is why treating copilot search optimization as a separate budget line usually produces worse results than fixing the on-page content the whole site shares. If you want to optimize for copilot and you have not yet made your answers liftable, start there rather than with anything branded AI.

What you can measure now

Until recently the honest answer to “are we cited?” was sampling: run prompts, record what comes back, accept the noise. Bing changed that on 10 February 2026 by putting AI Performance into public preview in Bing Webmaster Tools — a report that shows “how publisher content appears across Microsoft Copilot, AI-generated summaries in Bing, and select partner integrations”. (Documented — Bing Webmaster Blog, “Introducing AI Performance in Bing Webmaster Tools Public Preview”.)

Four metrics are worth knowing before you open it, because two of them are easy to misread:

  • Total Citations — how often your content is shown as a source. Bing is explicit that this is “without indicating placement or presentation within a specific answer”. It is a count, not a rank.
  • Average Cited Pages — the average number of unique pages cited per day. Bing again warns it “does not indicate ranking, authority, or the role of any page within an individual answer”.
  • Grounding queries — the key phrases the AI used when retrieving the content it cited. Bing calls the data “a sample of overall citation activity”.
  • Page-level citation activity — citation counts per URL, over a date range.

Grounding queries are the genuinely new artefact, and the one worth building a process around. They are not the user's prompt. They are the retrieval phrases the system generated on the way to an answer, which means for the first time you can see the vocabulary that fetched you rather than the vocabulary you hoped would. When those phrases do not match the terms your keyword universe is built on, the keyword set is the thing that is wrong. Treat copilot citations as a coverage signal, not a scoreboard: the useful question is not “how many”, it is “which pages, for which phrases, and what did we never get asked about at all”.

One limit to set expectations honestly: this is a public preview reporting on a sampled system, and Bing says it will “continue to refine this metric as additional data is processed”. Read the trend, not the day. Tooling that samples answer engines directly still has a job — we cover the category in LLM visibility tools — but for Copilot specifically, first-party citation data now exists and it is free.

Copilot versus ChatGPT and Gemini

The three read differently enough that a single “AI SEO” checklist misleads on at least one of them. Copilot's grounding index is Bing's, and its controls are Bing's robots and snippet directives. Google's Gemini surfaces split across Search features and the Gemini app, with a separate Google-Extended token governing only the app. ChatGPT runs its own crawlers with their own robots.txt tokens. The shared layer is real — be indexed, be parseable, be quotable — but the control surfaces are not interchangeable, and a robots.txt written for one of them says nothing about the other two.

One vocabulary note, since old advice is still circulating: bing chat seo was the phrase for this work before Microsoft retired the Bing Chat name in favour of Copilot in late 2023. The mechanism did not change with the rename — the product was grounded in Bing's index then and Copilot is now — but any checklist still written against the old name predates data-nosnippet, predates AI Performance reporting, and should be re-read against current documentation rather than trusted.

Questions we hear

How do I appear in Microsoft Copilot?

Be in Bing's index, allow Bing's crawler, avoid page-level or block-level snippet suppression, and write the answer as a short self-contained passage under a heading that names the question. There is no submission step and no Copilot-only ranking factor Microsoft has published. If you are wondering how to appear in microsoft copilot and you have never verified the site in Bing Webmaster Tools, that is the first move, not the last.

Does Copilot use Bing?

For web grounding, yes. Microsoft describes Copilot as “Powered by Bing's search index”, and the Bing team describes its snippet controls as applying to “Bing Search and Copilot experiences powered by Bing”. That is why Bing indexation, not any AI-specific step, is the prerequisite.

Is Copilot SEO different from ordinary SEO?

Mostly no. Microsoft's own position is that “traditional SEO fundamentals still matter — crawlability, metadata, internal linking, and backlinks remain essential”. What differs is the unit of selection: answers are assembled from parsed passages rather than whole pages, so structure and self-contained phrasing carry more weight than they do for a blue link.

Can I stop Copilot quoting part of a page without losing the ranking?

Yes, and this is the one genuinely Copilot-specific control. Bing supports the data-nosnippet HTML attribute on any element: the marked content “is still indexed normally, but it will be excluded from snippets and AI summaries”, and remains “available for ranking”. Use it for paywalled, sensitive or promotional blocks — and audit your templates for it if citations drop unexpectedly.

How do I know whether Copilot is citing me?

Open AI Performance in Bing Webmaster Tools, which entered public preview in February 2026. It reports total citations, average cited pages per day, page-level citation counts, and the grounding query phrases the AI used to retrieve your content. The grounding queries are the most actionable field, because they show the retrieval vocabulary rather than the vocabulary you targeted.

Do I need a Copilot-specific schema or an AI text file?

Microsoft has published no such requirement. Its schema guidance points at ordinary schema.org types — product, review, FAQ, event — described as turning “plain text into structured data that machines can interpret with confidence”. Structured data earns its place on its own merits; it is not an entry ticket to Copilot.

Go deeper

Related: generative engine optimization · ChatGPT SEO · Gemini SEO · LLM visibility tools · augmented intelligence SEO.

Want the Bing indexation check, the passage rewrite and the citation sampling run for you? See how AI search visibility works as a module, or request a free AI-powered SEO audit as a first read.

Back to blog

Leave a comment

Please note, comments need to be approved before they are published.