How does Microsoft Copilot pick what to surface?

Copilot retrieves web results for a question, reads the pages behind them, and composes an answer it can attribute back to sources. A structured page with a direct answer is easier to draw from than one where the point is buried. The selection logic is Microsoft's and changes over time.

Is Copilot the same as ranking on Bing?

They are related but not identical. Classic results point a person at a link. Copilot reads pages and answers in place, often without a click. A page can do well in results and still be hard for Copilot to read cleanly, so the two can move apart.

Can a tool guarantee Copilot will surface my site?

No. Copilot is probabilistic and Microsoft changes how it retrieves and reads. What you can control is readability: clear structure, direct answers, and machine-readable labels. Citedon measures that readiness and reports whether engines name you as proof, never as a promise.

← Back to answer engines

Answer engines · 3 min read

How Microsoft Copilot picks what to surface

Copilot answers from live web results and reads the pages behind them. Here is how it reads the web, what a page needs to be readable to it, and how to check yours.

Copilot lives inside the tools your buyers already work in, the browser, the search box, the workplace assistant. It answers there from pages it could read. So the question is not whether your buyer uses Copilot. It is whether Copilot could read you when they asked.

Copilot is a reader, not a results page

Classic web results pointed a person at a list of links. The person did the reading and the clicking.

Copilot inserts a reader between the question and the links. It pulls web results, reads the pages behind them, and writes an answer it can attribute to sources. The person often gets the answer without opening a single link.

So a page can surface well in results and still read poorly to the assistant that now sits in front of them. Doing well in the index does not mean Copilot can parse you into an answer.

How it reads the web and decides what to surface

Copilot gathers web results relevant to a question, reads the pages they point to, and composes a grounded answer from the ones it could actually parse. The general mechanics are in how AI engines decide what to recommend.

This fetch-and-ground pattern is retrieval-augmented generation: the engine reads live pages and builds an answer it can stand behind, leaning on the ones that read into clear statements.

We will not invent Microsoft's selection rules. They are proprietary and they shift. What stays constant is that a page Copilot cannot read cleanly is a page it cannot use cleanly, whatever its result position.

What a page needs to be readable to it

To be usable by Copilot, a page has to state its meaning in a form a machine can read without a person reading along.

A clear heading that names the topic, so it is not inferred from layout.
A direct, plain-text answer near the top, not buried below the fold.
Structured data that labels what the page is, what it offers, and what it costs.
Crawlable HTML, so the content is reachable rather than locked behind script.

Old way: win the result and assume the assistant inherits it. New way: keep the result work and add a machine-readable layer so Copilot reads you as clearly as the index does.

This does not force you into the answer. It makes you legible enough to be eligible, which is the part you can influence.

The damaging admission

Readable is not surfaced. We measure whether Copilot and the other engines can read and name you, and we report it plainly, but we will never promise Copilot will surface you.

The system is probabilistic and Microsoft changes how it works. It can read two clear pages and surface only one. Anyone guaranteeing you a spot in Copilot is selling certainty no one holds.

Readiness is the floor. If Copilot skips a page it cannot parse, that is fixable. If it reads you and surfaces a competitor anyway, at least you were in the running.

The automated fix is WordPress-only. On Shopify, Wix, Webflow, or a headless stack, the scan still diagnoses the gaps, but you apply the changes yourself.

Why readiness drifts on Copilot specifically

Copilot leans on the Bing index and on Microsoft's own surfaces, and both keep moving. Bing recrawls and re-ranks, Microsoft ships changes to how Copilot pulls and reads sources, and meanwhile you publish new pages that mostly ship without structure. Any one of those can turn a page that surfaced last quarter into one that does not.

So keeping Copilot able to read you is ongoing work, not a one-time setup. Nothing on your side has to change for the result to change.

And because Copilot sits inside Windows, Edge, and Microsoft 365, the answer often happens where someone is already working, not on a search page they chose to visit. Being readable at that moment is the whole game.

How to check whether Copilot can read your page

Pick the page you would most want Copilot to surface and find out if it can read it today.

Run a free scan on the URL to see whether Copilot and the other engines can parse it and which structural pieces are missing, or read how Citedon makes a site agent-ready and keeps it that way as the engines change.

See if Microsoft Copilot can read your site, free.

Run a free scan. No signup. You get a readiness score and the gaps to fix, in about a minute.

Run a free scan How it works