Copilot Rank Tracking: How to Track Your Brand in Bing Copilot

Copilot rank tracking answers a question that classic rank trackers cannot: when someone asks Microsoft’s AI in Bing for the best tool in your category, does the answer name you, and does it link to your site? To track your brand’s visibility in Copilot you need a different method from blue-link rank tracking, because the answer is generated text with a list of sources, and it changes from one run to the next.

What Copilot rank tracking measures

A Bing results page gives each URL a position. A Copilot answer gives you a paragraph, a few named products and a set of cited pages, so “rank” becomes a set of rates across many prompts and runs:

Metric	What it counts	Why it matters
Mention rate	Answers that name your brand, divided by all answers	How often buyers hear your name at all
Citation rate	Answers that cite a page on your domain	How often Copilot sends a reader to your site
Position in the answer	The order in which tracked brands are first named	The first brand named gets the most attention
Share of voice	Your mentions divided by all tracked brands’ mentions	Your standing against named competitors
Cited domains	Every domain in the sources, per prompt	The third-party pages that decide what Copilot says

Mention rate and citation rate move separately. Copilot can name you while it cites only a review site, or cite your page while it recommends a competitor. Track both, and keep the list of cited domains, because it tells you where to work next. Our AI share of voice guide has the formulas.

Which Copilot you can track

“Copilot” is the name of several Microsoft products. Only one of them gives the same public answer to anyone who asks.

Surface	Where	What it answers from	Can you track it from outside?
Copilot Search in Bing	bing.com/copilotsearch, the Copilot tab in Bing	The web, through Bing’s index	Yes. This is the logged-out public answer
Copilot app	copilot.microsoft.com, Windows, Edge	A consumer chat assistant with its own answers	Not with the API in this guide. Dashboards that list “Copilot” rarely say whether they mean this one
Microsoft 365 Copilot	Word, Outlook, Teams at work	Your organization’s files, mail and chats	No. Each tenant gets different answers

Microsoft has also tested renaming the Copilot search tab in Bing to “AI Mode”. Many vendors call the same page Bing Copilot, and our API calls it Bing AI Mode. When this guide says Copilot rank tracking, it means tracking Copilot Search in Bing.

Three ways to track your brand in Bing Copilot

Bing Webmaster Tools: free citation counts for your own pages

Microsoft announced the AI Performance report on 10 February 2026 as a public preview. The report has total citations, average cited pages, grounding queries, page-level citation activity and trends over time. Microsoft defines grounding queries as “the key phrases the AI used when retrieving content that was referenced in AI-generated answers”, and says the report holds a sample of overall citation activity.

The two sources we read disagree on scope. Microsoft’s post says the report covers “Microsoft Copilot, AI-generated summaries in Bing, and select partner integrations”. Search Influence, which analyzed its own export, wrote “This is Copilot only.” Search Influence also says any verified site can open the report, that it has no click-through data, and that you cannot see which other sites were cited beside yours. At launch it had no API, so the data came out as CSV exports.

Set it up first in any Copilot rank tracking plan, because it is free and the data comes from Microsoft. It only describes your own pages, though, so it cannot tell you that a competitor was named in an answer you missed.

Manual prompt checks: cheap, but you cannot repeat them

Open bing.com/copilotsearch, type the prompts your buyers ask, and note who is named and which pages are cited. This takes an afternoon and shows you what the answer looks like.

For Copilot rank tracking, manual checks have three problems. You see only one market, the one your connection is in. You cannot run the same prompt 50 times a week by hand. And one answer is a weak signal: in our paired runs, the same engine given the same prompt minutes apart repeated only about 40% of its cited domains (AI visibility sample size).

Programmatic sampling: a fixed prompt set on a schedule

Copilot rank tracking at scale works the same way whether you buy a tool or build one:

1. Fix a prompt set that covers your buyer intents.
2. Run it on a schedule, per country, and for US brands per state if your market is local.
3. Store the full answer text, the sources and the citation pills (the inline source chips) for every run.
4. Compute the metrics from the stored answers, never from memory of one check.

About 300 answers per period resolves a visibility rate to roughly 5 percentage points either way, so 50 prompts run daily gives you a sound weekly read. The sample size guide has the math.

Method	Cost	Your pages	Answer text	Competitors	Markets	Repeatable
Bing Webmaster Tools AI Performance	Free	Yes, counts and pages	No	No	Not stated	Yes, Microsoft collects it
Manual checks	Your time	Yes	Yes	Yes	One	No
API or dashboard sampling	Credits or a subscription	Yes	Yes	Yes	Per country and US state	Yes

Use the free report to see which of your pages Copilot already cites, and use sampling to see the answers your buyers read and who wins them.

What we saw when we tracked Copilot answers about this topic

On 8 November 2026 we sent three prompts about this topic through our Bing AI Mode API, from the US. Each is a single run on one day, so read them as examples of how Copilot works, not as stable rankings.

Prompt 1: “How can I track my brand’s visibility in Copilot Search in Bing?” The answer named two tools, Rankscale and Finseo. All 10 cited sources were Copilot-specific tracking guides or tool pages: searchfit.ai, geoscout.pro, thepromptinsider.com, bing.ly, orchly.ai, rankscale.ai, ai-search-visibility-tools.com, presenc.ai, finseo.ai and trackmyvisibility.com. Both named tools had their own page among the sources. The answer also told the user to “Ensure Bing indexes your site, since Copilot relies on Bing’s index”.

Prompt 2: “How do I get my website cited in Bing Copilot Search answers?” The answer cited 12 how-to guides and called the AI Performance report “currently the only official way to see Copilot citations”.

Prompt 3: “What is the best API to track brand visibility and citations in AI search engines like ChatGPT, Google AI Mode and Bing Copilot?” The answer listed cloro first, as “Best API-first visibility platform”, then Profound, Peec AI, OtterlyAI and Semrush. None of the 12 cited sources was one of our pages. They were third-party roundups, such as pages on semrush.com, unite.ai and alignify.co.

The lesson for Copilot rank tracking: Copilot recommended what the pages Bing retrieved for the prompt recommended. In prompt 1 the named tools were the ones with pages in the sources. In prompt 3 we were named only through other sites’ roundups.

A tracker that stores only “mentioned: yes or no” cannot tell you why you were named or why you dropped out. Store the sources with every answer. The domains that keep appearing for your prompts become your list of pages to get onto.

We also timed the API on 10 everyday prompts (software, travel, finance, health). All 10 returned an answer, in 15.0 to 26.5 seconds (median 19.4 seconds). Each answer had 9 to 11 sources and 14 to 39 citation pills. For Copilot rank tracking, that is enough sources per answer to show patterns after a few weeks, and slow enough per call that a large prompt set belongs in an async batch.

How to track your brand’s visibility with a Copilot rank tracking API

The API opens Copilot Search in Bing in the market you choose, parses the page and returns the answer and its sources as JSON. Your script checks it for your brand and domain. The setup has five steps:

1. Write a fixed prompt set.
2. Learn the three response fields that matter.
3. Run the prompts and write one CSV row per prompt.
4. Move the schedule to the async API.
5. Read the results weekly and act on the cited domains.

Step 1: write the prompt set

Start with 30 to 50 prompts across four kinds: category (“best payroll software for startups”), problem (“how to run payroll in two states”), comparison (“Gusto vs Rippling”) and brand (“is Acme payroll any good”). Keep the wording fixed between runs, so a change in the numbers comes from Copilot and not from you.

Step 2: know the fields you need

Three fields in the Bing AI Mode response matter for tracking:

• result.text is the answer. Search it for your brand and competitor names.
• result.sources lists the pages Bing read, then any other page a citation chip links, each with position, label, url and description.
• result.citationPills has one entry per chip and source pair, with position, label, url, domain and citationPillId. This tells you which pages the answer linked inline.

Add "include": {"markdown": true} if you also want result.markdown, which keeps headings, tables and Bing’s follow-up question.

Step 3: run the prompts and write a CSV

import csv
import os
from datetime import date
from urllib.parse import urlparse

import requests

API_URL = "https://api.cloro.dev/v1/monitor/bing/aimode"
API_KEY = os.environ["CLORO_API_KEY"]

BRAND = "Acme"
DOMAIN = "acme.com"
COMPETITORS = ["Gusto", "Rippling", "Deel"]
PROMPTS = [
    "best payroll software for startups",
    "how to run payroll for employees in two states",
]


def on_domain(url):
    host = (urlparse(url).hostname or "").lower()
    return host == DOMAIN or host.endswith("." + DOMAIN)


def ask(prompt, country="US", state=None):
    body = {"prompt": prompt, "country": country}
    if state:
        body["state"] = state  # US only, adds 2 credits
    resp = requests.post(
        API_URL,
        headers={"Authorization": f"Bearer {API_KEY}"},
        json=body,
        timeout=120,
    )
    resp.raise_for_status()
    return resp.json()["result"]


def row(prompt, result, country, state):
    text = result["text"].lower()
    sources = result.get("sources", [])
    pills = result.get("citationPills", [])

    # Order in which tracked brands are first named in the answer
    found = {b: text.find(b.lower()) for b in [BRAND] + COMPETITORS}
    named = sorted((pos, b) for b, pos in found.items() if pos >= 0)
    order = [b for _, b in named]

    own = [s["position"] for s in sources if on_domain(s["url"])]
    return {
        "date": date.today().isoformat(),
        "country": country,
        "state": state or "",
        "prompt": prompt,
        "mentioned": BRAND in order,
        "mention_rank": order.index(BRAND) + 1 if BRAND in order else "",
        "brands_named": ";".join(order),
        "source_position": min(own) if own else "",
        "pills_on_domain": sum(on_domain(p["url"]) for p in pills),
        "cited_domains": ";".join(sorted({p["domain"] for p in pills})),
    }


def main(country="US", state=None, path="copilot_tracking.csv"):
    rows = [row(p, ask(p, country, state), country, state) for p in PROMPTS]
    new_file = not os.path.exists(path)
    with open(path, "a", newline="") as f:
        writer = csv.DictWriter(f, fieldnames=list(rows[0]))
        if new_file:
            writer.writeheader()
        writer.writerows(rows)


if __name__ == "__main__":
    main()

Each row records whether Copilot named you, your order among the tracked brands, where your first page sits in the sources, how many inline chips point at your domain, and every domain the chips cite. Group the CSV by week for mention rate, citation rate and share of voice. Count the cited_domains column to find the sites that shape the answer.

Step 4: move the schedule to the async API

The script above uses the synchronous endpoint, which waits for each answer and costs 6 credits a request (4 plus a 2-credit sync charge). For scheduled Copilot rank tracking, send the prompt set to POST /v1/async/task/batch with taskType: "BING_AIMODE" and collect results by webhook or polling, at 4 credits a request.

On the Hobby plan ($100 a month for 250,000 credits) that is $1.60 per 1,000 answers. Running 50 prompts daily in one market comes to 350 answers and 1,400 credits a week. See pricing for the other plans.

Step 5: read it weekly and act on the sources

Compare each week with the last on the same prompts. When your mention rate drops, look at cited_domains for those prompts first: a new roundup that leaves you out, or a page of yours that fell out of Bing’s index, is the usual reason. For the same prompts in Google, read how Bing’s citations differ from Google AI Mode’s.

Copilot search rank tracking tools

If no one on your team wants to own a script, several dashboards say they track Copilot. The table lists what each vendor’s own page says. We did not test them for this guide, and most of the pages do not say which Copilot they read.

Tool	What its page says it tracks for Copilot	Which Copilot	Starting price (public page)
Rankscale	Search terms, brand mentions and citations in Bing Copilot, with visibility scores and share of voice against competitors	Page does not say which Copilot surface	Pro $99/month, 7-day Pro trial
Finseo	Brand mentions, cited Bing URLs per prompt, competitors, sentiment and prompt coverage, by country and language	Page does not say which Copilot surface	Starter from €70/month, Copilot is one of 3 models you pick
Omnia	Brand mentions in AI answers; Microsoft Copilot available on request	Page does not say which Copilot surface	Growth €79/month, Copilot on request
Otterly.AI	Lists Microsoft Copilot among its tracked AI engines	Page does not say which Copilot surface	From $29/month
TrackMyVisibility	A one-off report: visibility score, Bing index check, top queries, missed queries and a competitor leaderboard	Page does not say which Copilot surface	First report free, no signup; paid plans for ongoing tracking
cloro (API)	Answer text, sources and citation pills, per country and US state, for your own pipeline	Copilot Search in Bing	Free plan 500 credits/month; Hobby $100/month for 250,000 credits

Before you buy a Copilot rank tracking tool, ask the vendor which Copilot surface it queries and from which country. If it is Copilot Search in Bing in your market, the numbers match what your buyers see. If you want the raw answers for your own reports or product, use the API row. Our LLM visibility tracking tools roundup covers dashboards across more engines.

Start Copilot rank tracking with ten prompts

Pick ten prompts your buyers ask, run them through Copilot Search in Bing on our free plan, and check the CSV for two things: whether you are named, and which domains are cited when you are not. Those domains are where your Copilot visibility work starts. Then verify your site in Bing Webmaster Tools, so the AI Performance report can show you which of your pages Copilot already cites. For a ChatGPT version of the same setup, see our ChatGPT visibility tracker guide.