How to Rank in Microsoft Copilot in AI Answers: Bing Index, Entities, Freshness

Quick answer

Under this guide’s method, Canlah AI treats ranking in Microsoft Copilot as Bing work plus answer-shaped pages, because Copilot has no crawler of its own: bingbot builds the index and Copilot grounds its web answers in that index. The work runs in five steps in dependency order. Allow bingbot, get indexed and stay fresh through Bing Webmaster Tools and IndexNow, write for the short grounding query Copilot sends to Bing, structure every section so it can be extracted alone, then read the AI Performance report every month.

Canlah AI publishes the manual version here because Microsoft reports Copilot citations back to publishers for free: the Bing Webmaster Tools AI Performance report, in public preview since February 2026, names the grounding queries that retrieved your pages and the pages cited.

Treat any single Copilot answer as directional, because Copilot is a sampled system whose web grounding a tenant administrator or the user can switch off. Freeze a panel of eight to twelve buyer questions and repeat it by surface, market and language before deciding what to change.

Disclosure: Canlah AI publishes this guide and sells the managed AI visibility service described at the end of it. Microsoft product behaviour is taken from Microsoft’s own documentation, and tool details come from public pricing pages reviewed on September 23, 2026. Canlah AI’s published engine list does not name Microsoft Copilot, as the monitoring section states.

What does ranking in Microsoft Copilot actually mean?

Ranking in Microsoft Copilot means being cited in the source list of a written answer, and four conditions decide it: crawl access, Bing index membership, grounding-query match and the Copilot surface. Copilot returns an answer with sources rather than a numbered page of ten blue links, and a site owner controls the first two almost completely.

• Crawl access: whether bingbot may fetch the page in robots.txt, at the CDN and at the firewall.
• Index membership: whether the page is stored in the Bing index, the only web index Copilot grounds in.
• Grounding query match: whether the page answers the short query Copilot generates from the user’s prompt.
• Surface: whether the answer came from Web mode, which uses the public web, or Work mode, which reasons over tenant data through Microsoft Graph.

A page never crawled and a page never retrieved look identical in a Copilot answer; server logs and the AI Performance report separate them.

Which crawler decides whether a site can rank in Microsoft Copilot

One crawler decides whether a site can rank in Microsoft Copilot, and it is not a Copilot crawler: bingbot fetches the pages, and the index it builds feeds Bing Search, Yahoo, DuckDuckGo, Copilot and Microsoft’s Web IQ grounding APIs. That is the structural difference from ChatGPT and Perplexity, which run their own named search crawlers. One robots.txt line governs five destinations.

Microsoft’s crawler documentation, reviewed September 23, 2026, states that bingbot obeys robots.txt and honours crawl-delay values from 1 to 30 seconds. A genuine bingbot request resolves by reverse DNS to a hostname ending in search.msn.com, and Microsoft publishes the IP ranges at bing.com/toolbox/bingbot.json, first published on November 15, 2021.

The practical trap is a wildcard rule written in 2023 or 2024 to keep archives out of model training: a rule blocking every bot except Googlebot removes the site from Bing, Copilot and three other destinations at once. Firewall rules written after Cloudflare began blocking AI crawlers by default for new domains in July 2025 can catch bingbot too, so a clean robots.txt can still sit behind a closed CDN.

Why the Bing index decides Copilot visibility

Copilot visibility is Bing visibility, and the audience gap is why the work pays: Statcounter figures reported in 2026 put Bing near 4.7% of worldwide search and close to 12% of desktop search, while Microsoft-reported figures put Copilot at roughly 420 million monthly active users. A team that deprioritised Bing is also absent from an assistant inside Windows, Edge and Microsoft 365.

The mechanism most teams miss is the grounding query. Microsoft’s documentation on web search in Copilot, updated August 18, 2026, states that Copilot “generates a search query that it sends to the Bing search service” and that “this generated search query is different from the user’s original prompt — it consists of a few words informed by the user’s prompt.” The full prompt and tenant documents are excluded from it.

Optimising a page for a 40-word conversational prompt is therefore optimising for a string Bing never receives. Microsoft’s worked example turns a prompt about a possible acquisition of Fabrikam into two queries, “Fabrikam strategy” and “Fabrikam financials”, which is closer to classic query-document matching than most generative engine optimisation advice assumes. Microsoft also shows those queries beside a Copilot Chat response and keeps them in the thread for 24 hours.

How to rank in Microsoft Copilot, step by step

Ranking in Microsoft Copilot takes five steps in dependency order, because content work on a page bingbot cannot fetch is wasted.

Step 1: Allow bingbot, then check nothing downstream blocks it

Open yourdomain.com/robots.txt, find the group that applies to bingbot and replace any blanket disallow with an explicit “Allow: /” group. Then filter server logs for the bingbot/2.0 marker and look for 403 or challenge responses. If no verified bingbot request appears in 30 days of logs, access is the constraint.

Step 2: Verify the site in Bing Webmaster Tools and turn on IndexNow

Verify the domain, submit the XML sitemap and check that priority pages return as indexed. Then enable IndexNow, the free protocol Microsoft launched in October 2021 that notifies search engines the moment a page is added, updated or removed. It tells Bing what changed; bingbot still fetches.

Step 3: Write for the short grounding query, not the long prompt

List the buyer questions you care about, then write down the two to five word query each one collapses into, and match those strings in the title, the first sentence and one H2. Canlah AI treats this as the largest difference from ChatGPT work, because ChatGPT draws on several retrieval providers and Copilot draws on one keyword-shaped index.

Step 4: Make every section chunk cleanly, then keep the entity and the date consistent

Microsoft’s October 2025 guidance states that assistants parse content into smaller pieces and that “these modular pieces are what get ranked and assembled into answers.” Lead each section with a self-contained answer to its heading and put comparable facts in tables. Keep the legal name, category, address and founding year identical across the site, JSON-LD Organization markup and every directory Bing indexes, and pair each dated refresh with an IndexNow ping. Canlah AI’s September 2026 study of 92 pages cited by ChatGPT and Google AI Mode, not Copilot, found about 73% of copied facts in the first 30% of the page.

Step 5: Read the AI Performance report every month and act on grounding queries

The AI Performance report shows Total Citations, Average Cited Pages, Grounding Queries and page-level citation activity for Copilot and AI summaries in Bing. On June 16, 2026 Microsoft added Intents, Topics, Compare and Citation Share, the share of all citations for one grounding query that point to your site. Where a competitor holds the larger share, rewrite the first sentence of the matching section to answer that query, ping IndexNow and compare the next month’s sampled figure.

Microsoft Copilot ranking factors and how to verify each one

Seven factors decide whether a page ranks in Microsoft Copilot, and a site owner can check all seven for free with server logs, Bing Webmaster Tools, the Copilot Chat answer itself and the AI Performance report. Five are documented by Microsoft; structure and entity consistency rest on Microsoft guidance.

Seven Microsoft Copilot ranking factors in dependency order, with the evidence behind each, reviewed September 23, 2026.

Ranking factor	What it affects	How to verify it yourself	Evidence strength and caveat
bingbot access	Whether the site is eligible at all	robots.txt group plus verified bingbot requests in server logs	Documented by Microsoft
Bing index membership	Whether the page can be retrieved for a grounding query	Bing Webmaster Tools indexation report, plus 403s to bingbot at the CDN	Documented by Microsoft
Freshness signal	Whether Bing refetches a changed page quickly	IndexNow submissions in Bing Webmaster Tools and a visible updated date	IndexNow documented; effect on Copilot citations not quantified
Grounding-query match	Whether the page enters the candidate set	Web search query citations in the Copilot Chat answer, kept 24 hours	Documented by Microsoft, August 18, 2026
Extractable structure	Whether a passage can be lifted and assembled	First 60 words of each section read alone	Microsoft guidance, October 2025; measured on ChatGPT and Google AI Mode, not Copilot
Entity consistency	Whether Bing resolves the brand to one set of facts	Name, category, address and founding year compared across site, JSON-LD and directories	Guidance, not a documented Copilot factor
Citation share	Whether competitors own the same grounding query	AI Performance report, Citation Share metric, preview from June 16, 2026	Documented by Microsoft; grounding queries are sampled

Source: Microsoft documentation for the documented rows, Microsoft Advertising guidance of October 2025 for the structure and entity rows and the Canlah AI citation study of September 2026 for the measured evidence.

Why a website is not showing in Copilot: five common results

A website that is not showing in Microsoft Copilot is usually in one of five states, and the first two are access problems that no content change can fix.

Not indexed by Bing: This is the most common cause and the easiest to confirm in the Bing Webmaster Tools indexation report. A site can rank well in Google and still be absent from Copilot for this reason alone.

Blocked at the edge despite robots.txt: The robots.txt file allows bingbot while the CDN or firewall returns 403 responses to it. Server logs filtered for verified bingbot IP ranges show the pattern, and the fix is a firewall allow rule rather than a content change.

Cited only when the question names the brand: A branded question puts the brand inside the grounding query, so retrieval finds the site easily. Canlah AI’s September 7, 2026 self-audit of canlah.ai, sampled on ChatGPT and Gemini rather than Copilot, showed the gap: 39 of 39 branded answers named the brand against 7 of 177 grounded answers to open buyer questions, a 4.0% rate.

Visible in Bing Search but absent from Copilot answers: The page is indexed but does not answer the short grounding query in an extractable form. Read the web search query citations the Copilot answer lists, then check whether your page states that answer in its first lines; one rewritten sentence is usually enough.

Absent from Copilot at work but present at home: Work mode grounds in tenant data through Microsoft Graph, and an administrator can disable web grounding with the “Allow web search in Copilot” policy. No amount of public-web work reaches a tenant where web search is off.

Which tools report Microsoft Copilot visibility

Five tools report Microsoft Copilot visibility on pages reviewed September 23, 2026: Bing Webmaster Tools AI Performance for free, then Otterly.ai, Peec AI, Ahrefs Brand Radar and Scrunch at published entry prices from $29 to $250 a month. All four paid platforms name Copilot, so the buying question is sampling depth and price.

Five tools that report Microsoft Copilot visibility, from public pricing pages reviewed September 23, 2026.

Tool	Copilot coverage on reviewed pages	Published entry price	Best fit	Main point to verify
Bing Webmaster Tools AI Performance	First-party citations, grounding queries and citation share	Free	Every site owner, before any paid tool	Grounding queries are sampled, not complete
Otterly.ai	In the base set with ChatGPT, Google AI Overviews and Perplexity, daily	$29 a month for 15 prompts	Small budgets that want Copilot checked daily	Gemini, Google AI Mode and Claude are paid add-ons
Peec AI	One of six self-serve models, three tracked at a time	$95 a month for 50 prompts	Agencies that need unlimited seats	A fourth model costs extra below Enterprise
Ahrefs Brand Radar	30.8 million monthly Copilot prompts in a 454 million prompt index	$199 a month for the index	Category benchmarking	Copilot sets refresh monthly on a 90-day window
Scrunch	On Core, with ChatGPT, Perplexity and Google AI Overviews	$250 a month for 125 prompts	Teams that also want agent-facing site delivery	Repeat sampling per prompt: not found on reviewed page

Source: vendor pricing pages and Bing Webmaster Tools help, reviewed September 23, 2026. “Not found on reviewed page” means the provider’s public materials did not name that capability when reviewed on September 23, 2026. It is not proof that the capability is absent.

What ranking in Microsoft Copilot cannot promise

Microsoft Copilot ranking work cannot promise five outcomes, because Microsoft controls the grounding query, retrieval and the final source list.

• It cannot guarantee a citation for any single question, however closely the page matches the grounding query.
• It cannot prove that one content change caused a new citation without a frozen panel and dated repeat runs; before-and-after movement is association, not causation.
• It cannot reach Copilot Work mode or a tenant where an administrator has turned off the “Allow web search in Copilot” policy.
• It cannot be read from one answer, because repeat runs of the same question return different source lists.
• It cannot be bought, because Microsoft Advertising places labelled ads apart from the organic source list.

When to move from a manual panel to ongoing monitoring

A manual Microsoft Copilot panel becomes hard to sustain above ten questions run monthly, or once more than one surface, market or language is in scope. Monitoring pays when Copilot answers influence real buying decisions. The AI Performance report covers Copilot for free and nothing else, so a brand that also tracks ChatGPT or Gemini needs a second layer regardless.

Canlah AI is a Singapore-based SEO + GEO agency that measures and improves brand visibility inside ChatGPT, Gemini, Google AI Overviews and Google AI Mode, reporting AI visibility as re-verifiable ranges with timestamped evidence. Its pricing page describes a free AI visibility report within 48 hours and a GEO service priced per AI platform: Starter from $1,500 a month (ChatGPT, 4 buyer questions), Growth from $3,000 a month (adds Google AI Overviews, 8 buyer questions) and Full customised (adds Google AI Mode and Gemini, 16 buyer questions), each re-measured monthly.

The limitation matters most on this page. Microsoft Copilot is not named in Canlah AI’s published engine list, and its September 7, 2026 self-audit sampled OpenAI and Gemini at the API layer across 39 questions and 234 runs, producing 317 raw answer files with no Copilot leg in them. A buyer who needs Copilot tracked should treat the AI Performance report plus one Copilot-covering tool from the table above as the baseline.

Do not choose a tool or an agency from this guide alone. Verify your own domain in Bing Webmaster Tools first, read the grounding queries Microsoft already reports for your pages, then ask any shortlisted vendor to rerun your frozen Copilot panel in front of you with prompts, dates and cited URLs visible.

Frequently asked questions

How do I rank in Microsoft Copilot?

Ranking in Microsoft Copilot starts with allowing bingbot in robots.txt and at the firewall, because Copilot has no crawler of its own and grounds in the Bing index. Then comes indexation and freshness through Bing Webmaster Tools and IndexNow. Last, each section’s first sentence should answer the short grounding query sent to Bing, not the long prompt a user typed.

Why is my website not showing in Copilot?

A website is usually missing from Copilot because it is not indexed by Bing, or because bingbot is blocked at the CDN even though robots.txt allows it. Check the Bing Webmaster Tools indexation report, then server logs for 403 responses to verified bingbot IP ranges. If both are clean, the fix is usually one rewritten first sentence that answers the grounding query.

Is Microsoft Copilot SEO different from Bing SEO?

Microsoft Copilot SEO is Bing SEO plus passage structure. The access and indexation layers are identical, because one bingbot index serves Bing Search and Copilot. What is added is chunking: Microsoft parses pages into modular pieces and retrieves them against a few keywords.

Can I pay to appear in Microsoft Copilot answers?

No. There is no product that places a website inside the organic source list of a Copilot answer, and any vendor guaranteeing a citation is promising something it does not control. Microsoft Advertising places labelled ads reviewed by AdIdxBot, and buying them does not change organic grounding.

Does Canlah AI track Microsoft Copilot?

No. Microsoft Copilot is not part of Canlah AI’s standard tracking, which covers ChatGPT, Gemini, Google AI Overviews and Google AI Mode by tier (Starter covers ChatGPT, Growth adds Google AI Overviews, Full adds Google AI Mode and Gemini), and its September 7, 2026 self-audit contains no Copilot leg. Confirm current engine coverage with Canlah AI or any other vendor directly before signing.

Method and sources

This guide was written on September 23, 2026. Crawler behaviour, grounding-query mechanics, admin controls and the AI Performance report come from Microsoft’s own documentation; tool coverage and prices from public pricing pages reviewed the same day. Audience figures are Microsoft-reported or Statcounter-derived. The Canlah AI figures come from API-layer samples of OpenAI and Gemini on September 7, 2026 and a September 2026 citation study of ChatGPT and Google AI Mode; neither contains a Copilot leg.

• Data, privacy, and security for web search in Microsoft Copilot, updated August 18, 2026. Generated search queries, citations held for 24 hours, the “Allow web search in Copilot” policy and the Web content toggle.
• Optimizing your content for inclusion in AI search answers, Microsoft Advertising, October 2025. Parsing into modular pieces, and JSON-LD schema.
• New AI visibility insights in Bing Webmaster Tools, June 16, 2026. Intents, Topics, Citation Share and Compare.
• Bing Webmaster Tools AI Performance report and IndexNow. The four metrics, in preview since February 2026, and the protocol launched October 2021.
• bingbot documentation and bing.com/toolbox/bingbot.json. User-agent marker, crawl-delay range of 1 to 30, reverse DNS verification and the IP ranges published November 15, 2021.
• Vendor pricing pages for Otterly.ai, Peec AI, Ahrefs Brand Radar and Scrunch, reviewed September 23, 2026.
• Canlah AI pricing. Free AI visibility report within 48 hours, per-platform tiers with their buyer-question counts, monthly cadence and the engine list that does not name Copilot.
• Canlah AI self-audit archive, anonymised, September 7, 2026. Thirty-nine questions, 234 runs, 317 raw answer files; 7 of 177 grounded open answers named the brand. Canlah AI citation study, September 2026: sixty queries on ChatGPT and Google AI Mode, 526 citations, 92 cited pages.

See where your brand stands in AI answers. Run the free AI visibility audit →

Related articles

• How to rank on ChatGPT in search results
• Why a website is not showing in AI Overviews
• How AI engines choose which sources to cite
• Best AI search monitoring tools