Last updated: October 2026. This is a Copilot SEO guide for B2B SaaS teams: how Microsoft Copilot picks the pages it cites, a seven-step playbook built around that mechanic, and how to measure the result with Microsoft's own first-party data instead of guesswork. It draws on the Arobis AI Readiness Report and the buyer-prompt audits we run for clients.
To rank in Microsoft Copilot, you have to rank in Bing for the short search query Copilot writes from your buyer's prompt, and then give Copilot a passage it can quote. Copilot does not search with the buyer's full question, so Copilot SEO is Bing SEO for those short "grounding queries", measured in Bing Webmaster Tools.
Key takeaways
- Microsoft says Copilot takes a few words from the user's prompt, turns them into a short query and sends that query to Bing, so Bing rankings decide which pages Copilot can cite.
- The AI Performance report in Bing Webmaster Tools shows total citations, cited pages, grounding queries and trends for Microsoft Copilot and Bing AI summaries, but no click data.
- Microsoft reported more than 30 million paid Microsoft 365 Copilot seats in its fiscal Q4 2026 results on July 29, 2026.
- Microsoft 365 administrators can control whether Copilot uses web information; with web search off, Microsoft Copilot (formerly Microsoft 365 Copilot) answers only from work content.
- We found no Microsoft documentation of a Copilot-specific schema type or a special LinkedIn feed into Copilot.
What is Copilot SEO, and how is it different from ChatGPT SEO?
Copilot SEO is the work of getting your pages cited, and your brand named, in answers from Microsoft Copilot. That includes the Copilot app, Copilot in Edge and Windows, the AI answers inside Bing, and Copilot Chat inside Microsoft 365.
The difference from ChatGPT is where the answer comes from. ChatGPT blends what its model already learned with its own web search. Copilot leans on one search index you can actually see and control: Bing. When web grounding is on, Copilot queries Bing, reads the pages that come back, and builds its answer from them. The table compares Copilot with the other major engines.
| Factor | Microsoft Copilot | ChatGPT | Perplexity | Gemini |
|---|---|---|---|---|
| Primary Audience | Enterprise B2B Professionals & knowledge workers |
General audience | Researchers & technical buyers | Consumers & Google users |
| Primary Data Source | Bing + Microsoft Knowledge Graph + M365 data | Training data + optional browsing | Live web crawl (RAG) | Training data + Google Search |
| Unique Optimization Lever | LinkedIn + Bing + Microsoft Clarity | Entity strength in training data | BLUF structure + domain authority | Google Search performance |
| Enterprise Context Awareness | Very High M365 organizational data access |
Low | Low | Low |
| Citation Format | Bing-linked inline citations | Brand mentions with limited links | Always inline citations | Mixed by query |
| Competitive Density (2026) | Low Opportunity Most brands are not optimizing yet |
High | Medium — growing rapidly | Medium |
That makes Copilot the most inspectable AI engine for a marketer. You can see the query Copilot sent. You can see your Bing ranking for that query. And you can see, in a Microsoft report, how often your pages were cited. If you already work on ranking in ChatGPT, treat Copilot as the engine where you can finally check your assumptions. The wider picture of how Perplexity, ChatGPT, Claude and Gemini choose which brands to mention explains where the other engines differ, and our guide on how to rank in Google Gemini covers the Google side.
Why should B2B SaaS teams care about Copilot?
B2B SaaS teams should care about Copilot because Copilot sits where their buyers already work. Your buyer does not have to go looking for Copilot. Copilot is built into Outlook, Teams, Word, Edge and Windows, which is where vendor research, RFP drafts and shortlist emails happen.
- Business reach: Microsoft 365 Copilot is sold as paid business seats, so the people using it are often the people who research and buy software. Current seat counts are in the dated section below.
- Desktop search: Bing holds a small but real share of US desktop search, and desktop is where most B2B work happens. The latest StatCounter figure is in the dated section below.
- Measurability: Microsoft is one of the few AI engine owners that gives site owners first-party citation data, so you can see whether your work is paying off.
To be fair about it: Google and ChatGPT are still bigger channels for most SaaS brands. The case for Copilot is not that it is the biggest engine. It is that Copilot reaches buyers at work, and it is cheap to improve because the work overlaps with SEO you already do. That fits the broader pattern of how AI search changes SaaS marketing: the channel is decided inside the answer, not on a results page.
How does Copilot decide which pages to cite?
Copilot cites pages that rank in Bing for the short query it writes and that contain a clear passage answering the question. Think of every Copilot answer as a chain with four links. You can lose the citation at any one of them.
- Link 1, the prompt. A buyer types something long and specific, often with work context: "We're a 300-person fintech moving off Zendesk, which help desk tools handle SOC 2 and Salesforce well?"
- Link 2, the grounding query. Copilot compresses that into a few words and sends it to Bing, something like "help desk software SOC 2 Salesforce integration". The long prompt is not what gets searched. The short query is.
- Link 3, the Bing results. Bing returns its results for that short query. If your page is not in what Bing returns, Copilot never reads it.
- Link 4, the passage. Copilot pulls the parts of those pages that answer the question and cites them. A page that ranks but buries the answer in paragraph nine can still lose the citation to a clearer page.
Most advice about Copilot focuses on link 4, writing "AI-friendly content". The bigger losses usually happen at links 2 and 3. Teams optimize for the long, conversational prompt, but Copilot searches with a short, almost old-fashioned keyword. If you do not rank in Bing for that keyword, your formatting never gets a chance. That is also why traditional SEO alone is no longer enough: the ranking still matters, but for a query your buyer never typed.
What is a grounding query, and how do you find yours?
A grounding query is the search phrase an AI engine sends to a search index to fetch sources for its answer. Microsoft uses the term in its own reporting: Bing Webmaster Tools defines grounding queries as the key phrases the AI used when retrieving content.
For Copilot, grounding queries are the new keywords. There are two free ways to find them.
- Watch Copilot write them. In Microsoft 365 Copilot Chat, when an answer uses the web, you can open the details and see the exact query Copilot sent to Bing and the sources it used, as Microsoft's web search documentation describes. Run ten of your real buyer prompts and write down every query. You will usually find they are shorter and more generic than you expect.
- Read them in Bing Webmaster Tools. The AI Performance report lists grounding queries that led to citations of your pages. Microsoft notes this is a sample of citation activity, not the full list, but it is real data from the engine itself.
Once you have the list, check where you rank in Bing for each query. Not Google. Your Bing rankings can be quite different, and Bing is the only one that matters for this chain.
How do you rank in Microsoft Copilot? A seven-step playbook
The seven steps below follow the chain in order. Do them in sequence, because the later steps are wasted if the early ones are broken. The table turns the playbook into a prioritized checklist with rough effort estimates.
| Priority | Action | Impact | Effort |
|---|---|---|---|
| 1 | Submit site to Bing Webmaster Tools + sitemap | Foundation Enables Bing-based Copilot retrieval. |
~1 hour |
| 2 | Install Microsoft Clarity | Behavioral Signals Feeds engagement data into Bing. |
~10 min |
| 3 | Enable IndexNow | Indexing Speed Accelerates discovery of new and updated content. |
1–2 hours |
| 4 | Optimize LinkedIn Company Page | Entity Strength Improves Microsoft Knowledge Graph signals. |
2–4 hours |
| 5 | Add Organization + FAQ Schema | Structured Data Improves entity clarity and extraction. |
Half day |
| 6 | Create Enterprise Use-Case Content | Query Match Aligns with enterprise Copilot searches. |
1–2 weeks |
| 7 | Build G2 & Capterra Presence | Trust Signals Strengthens third-party validation. |
Weeks–Months |
| 8 | Earn Editorial Mentions | Authority Creates high-trust Bing citations. |
Ongoing |
| 9 | Run Monthly Copilot Prompt Tests | Measurement Tracks visibility and competitor movement. |
2 hrs/month |
Step 1: Make sure Bing can index you properly
Verify your site in Bing Webmaster Tools, submit your XML sitemap, and check the indexing and crawl reports for errors. Confirm that your robots.txt does not block Bingbot. Microsoft states that Bing respects robots.txt and other content controls, so a stray rule written for Google can silently remove you from Copilot too. It is worth checking what AI crawlers actually read on your key pages while you are there.
Then run a site search in Bing for your most important pages. Pricing, product, comparison and integration pages should all be indexed. A page can rank fine in Google and still be missing or out of date in Bing.
Bing and Google are also often described as weighing some signals differently. The table summarises those commonly cited differences. Microsoft does not publish ranking weights, so read the table as directional and test it against your own AI Performance data.
| Bing vs Google Signal | Bing Weight | Google Weight | Copilot Implication |
|---|---|---|---|
| LinkedIn & Social Engagement | Very High | Low | LinkedIn activity can directly strengthen Copilot visibility. |
| Exact-Match Domain / Title Relevance | High | Medium | Brand names in titles and meta descriptions matter more. |
| Behavioral Engagement Data | High | Indirect | Microsoft Clarity may provide useful engagement signals. |
| Named Author Credentials | Very High | Medium–High | Author schema and LinkedIn profiles amplify trust. |
| Backlink Quantity vs. Quality | Quality-focused | Both | Fewer high-quality editorial links outperform volume. |
Step 2: Turn on IndexNow
IndexNow is an open protocol that lets you tell participating search engines, including Bing, the moment a URL is added, updated or removed. Bing's own guidance for improving AI citations recommends IndexNow, because Copilot can only cite the version of a page that Bing has.
Many CMSs and CDNs offer IndexNow as a setting or plugin. If yours does not, the protocol is a key file on your domain plus a simple request to an IndexNow endpoint when content changes. This matters most for pages with facts that change: pricing, integrations, feature lists and "best of" posts.
Step 3: Map buyer prompts to grounding queries
Write 20 to 30 prompts the way your buyers would type them at work. Use real roles, real company sizes and the tools they are leaving. Run them in Copilot, record the grounding query for each one, and group the queries into clusters.
For a typical B2B SaaS company the clusters look like this:
- Category: "best [category] software for [segment]"
- Comparison: "[competitor] alternatives", "[you] vs [competitor]"
- Requirement: "[category] with [integration or compliance need]"
- Cost: "[category] pricing", "[you] pricing"
- Problem: "how to [job your product does]"
This map replaces keyword research for Copilot. The map tells you which short queries decide whether you get named, and it usually exposes a few clusters where you have no page at all.
Step 4: Build one strong page per query cluster
Each cluster needs a page that deserves to rank in Bing for its main grounding query. For B2B SaaS that usually means a category page, honest comparison pages, integration pages, a pricing page with real numbers, and use-case pages by segment. We explain why AI search engines often prefer product and landing pages over blog posts for these queries.
Keep one page per intent. If three thin blog posts all target "[category] for startups", Bing has to guess which one to rank, and often picks none of them. One thorough page does better than three overlapping ones.
Step 5: Write passages Copilot can lift
Now link 4. Bing's published advice for AI citations is plain: improve structure and clarity with headings, tables and FAQ sections, keep content current and accurate, and support claims with evidence. In practice:
- Answer the query in the first two or three sentences under each heading.
- Phrase headings as the questions buyers ask.
- Keep paragraphs short enough to quote on their own.
- State facts precisely: plan names, prices, limits, integrations, certifications.
- Add a real FAQ section with short, direct answers.
Write each section so it still makes sense if Copilot quotes it with nothing around it. Clear, specific passages are one of the five signals AI engines use to recommend brands.
Step 6: Get onto the third-party pages Copilot already cites
For category queries, Copilot often cites roundups, review sites and industry publications rather than vendor sites. When you run your prompts in step 3, write down every domain Copilot cites. That list is your outreach and PR plan, and our research on the most influential domains in AI search rankings shows which kinds of sites tend to recur.
If a "best [category] tools" roundup is cited again and again, being in that roundup is worth more than another post on your own blog. Ask for inclusion, correct outdated facts about you, and give editors something useful to cite, such as original data or a clear pricing page.
Step 7: Keep the facts fresh
Copilot cites what Bing has now. When pricing, packaging or integrations change, update the page the same day, update the visible "last updated" date, and ping IndexNow. Stale facts are one of the quickest ways to lose a citation to a competitor, or worse, to have Copilot repeat your old price to a buyer.
What about Copilot inside Microsoft 365, where it reads work files?
Microsoft 365 Copilot can answer from two places: the buyer's work content (email, chats, documents) and the web through Bing. Microsoft lets organizations control whether Copilot uses web information at all. In a company that turns web search off, Copilot can only answer from internal content.
That has a practical consequence nobody talks about. In those tenants, the only "content" about you that Copilot can see is what already sits in the buyer's inbox and shared drives: your proposal, your one-pager, your security questionnaire answers, the call notes. So sales collateral is part of Copilot SEO too.
- Send documents as real text, not flattened images, so they can be read.
- Put the key facts on page one: what you do, who it is for, pricing model, integrations, compliance.
- Use the same product and plan names everywhere so internal and web sources agree.
You cannot measure this part from the outside, so treat it as good hygiene rather than a channel you track.
What does not move Copilot visibility?
Posting volume on LinkedIn, special schema, llms.txt and Google rankings on their own do not reliably move Copilot visibility. A lot of Copilot advice repeats claims nobody has shown. Be careful with these:
- "LinkedIn posts feed Copilot directly." Microsoft owns LinkedIn, but we found no Microsoft documentation saying Copilot has a special LinkedIn feed. Public LinkedIn pages can be indexed by Bing like any other page, which is a good enough reason to keep your company page accurate. Just do not expect posting volume to buy you citations.
- "There is a special Copilot schema." There is not. Structured data helps search engines understand your pages, and FAQ sections help with clear answers, but no markup type is reserved for Copilot.
- "llms.txt gets you into Copilot." In the Arobis AI AI Readiness Report (1,450 site scans, June to September 2026), 45.6% of sites had an llms.txt file and 75.3% shipped JSON-LD. Sites with both averaged a readiness score of 93.2 against 47.2 for sites with neither. That sample is self-selected, since the sites that scan themselves are already AI-aware, so it shows correlation, not cause. Bing's own AI citation guidance does not mention llms.txt. Fine to have, not a Copilot lever.
- "Rank well in Google and Copilot will follow." Sometimes. But Copilot uses Bing, and the two indexes and rankings differ. Check Bing directly.
How do you measure Copilot visibility?
You measure Copilot visibility with the AI Performance report in Bing Webmaster Tools, a fixed monthly prompt panel, and referral and self-reported attribution. Copilot is unusual among AI engines because its owner gives site owners first-party citation data. Use that data, then fill the gaps.
1. Bing Webmaster Tools AI Performance report
The AI Performance report covers Microsoft Copilot, AI summaries in Bing and select partner integrations. The report shows:
- Total citations: how often your pages were shown as sources in AI answers.
- Average cited pages: how many unique pages of yours were cited per day.
- Grounding queries: the phrases the AI used when it retrieved your content.
- Page-level activity: citation counts per URL.
- Visibility trends: how citations change over time.
Later additions brought Intents, Topics, Citation Share and Compare. Citation Share is the one to watch: your site's percentage of all citations shown for a given grounding query. Compare overlays a previous period, so you can see whether an update actually moved anything.
Know the limits. The report shows citations, not clicks or traffic, and grounding query data is a sample. The report tells you whether Copilot uses your pages. It does not tell you whether Copilot recommends your brand when your pages are not the source.
2. A fixed prompt panel
Keep the 20 to 30 prompts from step 3 and rerun them monthly, in the same way each time. Record three things per prompt: is your brand named, is your site cited, and which competitors appear. A prompt panel catches what the Bing report cannot, such as Copilot naming you based on a third-party roundup.
If you want to automate this, we compared the tools that actually cover Copilot in our guide to Microsoft Copilot rank trackers, and our Searchable vs Profound vs AthenaHQ comparison covers the broader AI visibility platforms. For a quick first read across engines, the free AI visibility checker shows whether AI engines mention your brand today.
3. Referral traffic and pipeline
In GA4, look for referral sessions from Copilot and Bing hostnames, and ask "how did you hear about us?" on demo forms with an AI assistant option. Copilot will under-report in analytics because many buyers read the answer and never click, so self-reported attribution matters.
What results should you expect from Copilot SEO, and how fast?
Expect indexing fixes to show first, Bing ranking gains over weeks to months, and third-party inclusion last. Nobody has published verified timelines for Copilot, so be wary of anyone who promises one. What we can say:
- Indexing and freshness are the fastest fixes. With IndexNow, Bing can pick up a changed page quickly instead of waiting for a recrawl.
- Bing rankings for grounding queries move on normal SEO timelines, which means weeks to months, faster for low-competition queries.
- Third-party inclusion depends on other people's editorial calendars and is usually the slowest part.
Start from an honest baseline. In recent Arobis AI audits, a public BI company appeared in 0 of 9 buyer prompts we tested across AI engines, and an automation consultancy scale-up appeared in 0 of 8. Being absent from AI answers is the normal starting point, not a sign that something is broken beyond repair, and the usual causes are the ones in why your brand does not appear in AI answers and the 10 reasons brands never appear in ChatGPT.
If you want an outside view of where you stand in Copilot and the other engines, and which pages to fix first, an AI visibility audit maps your buyer prompts, grounding queries and cited sources in one place. For ongoing work tied to pipeline, see our AI visibility program for SaaS companies. Arobis AI is an AI Search Demand Generation agency for B2B SaaS, and this is the done-for-you work we run for clients.
What has changed in Copilot and Bing recently? (Checked October 2026)
Copilot and Bing facts that change often are kept in this one section so they are easy to re-check. We review this section every month.
- Seats: Microsoft reported more than 30 million paid Microsoft 365 Copilot seats in its fiscal Q4 2026 results on July 29, 2026. Those are paid business seats, not casual consumer users.
- Search share: Bing held 12.94% of US desktop search in August 2026, according to StatCounter, against 81.84% for Google.
- February 2026: Microsoft launched AI Performance in Bing Webmaster Tools as a public preview.
- June 2026: Microsoft added Intents, Topics, Citation Share and Compare to the AI Performance report.
Frequently asked questions
Does Microsoft Copilot use Bing or Google?
Microsoft Copilot uses Bing. When Copilot grounds an answer in the web, it generates a short query from a few words of the prompt and sends that query to Bing. Your Bing indexing and Bing rankings are what count for Copilot citations, not your Google rankings, so check both separately.
Can I see how often Copilot cites my website?
Yes. The AI Performance report in Bing Webmaster Tools shows total citations, average cited pages, grounding queries, page-level activity and visibility trends for Microsoft Copilot, Bing AI summaries and select partners. The report does not show clicks, and its grounding query list is a sample rather than a full record.
What is a grounding query?
A grounding query is the short search phrase an AI engine sends to a search index to find sources for its answer. In Microsoft Copilot, a grounding query is usually a few words taken from a much longer prompt, and it works like a keyword you need to rank for in Bing.
Is Copilot SEO different from Bing SEO?
Copilot SEO is mostly Bing SEO with two extra jobs: targeting the short grounding queries that Copilot writes, and writing passages clear enough to be quoted on their own. Getting named on the third-party pages that Copilot already cites for your category matters too, especially for "best tools" questions.
Is Copilot SEO part of GEO?
Yes. Copilot SEO is one engine-specific part of generative engine optimization, the practice of earning citations and mentions in AI-generated answers. It also draws on answer engine optimization, which is about writing direct, quotable answers that engines like Copilot can lift.
Does IndexNow help with Copilot?
IndexNow helps Copilot see your latest content. IndexNow tells Bing when a page is added, updated or removed, and Bing recommends it in its guidance for AI citations. IndexNow does not make a weak page rank, but it stops a good page from being cited with old facts.
Does Microsoft 365 Copilot use the public web?
Microsoft 365 Copilot can search the web through Bing, but each organization controls whether Copilot uses web information. When web search is off, Microsoft 365 Copilot answers only from the organization's own work content, such as email, chats and documents, so your sales collateral becomes the source.
How is ranking in Copilot different from ranking in Perplexity?
Copilot searches Bing, an index you can inspect, and Microsoft gives site owners first-party citation data in Bing Webmaster Tools. Perplexity runs its own index and crawlers, so the Bing playbook only partly carries over. We cover the differences in how to rank in Perplexity.




