Skip to main content

Your AI visibility tool loses Microsoft Copilot on November 9. Here is how a Georgia B2B team keeps measuring the buyers who research there.

Semrush stops tracking Microsoft Copilot on November 9. What Georgia B2B teams lose, what buyer research stays hidden, and how to measure AI search pipeline now.

Four professionals in suits talking around a meeting table with a laptop and notebooks

Your AI visibility report says your brand shows up in Copilot answers about a third of the time. Your sales team says enterprise buyers arrive already holding a shortlist. Nobody can say which answer put you on it. On November 9, one of the biggest tracking tools stops looking at Copilot at all.

This is not your fault, and it is not a Semrush failure either. Microsoft now requires a signed-in account to use Copilot, and its terms ban bots and scrapers. For a Georgia B2B team selling into Atlanta's corporate headquarters, that means the assistant built into the Microsoft 365 desks of your best buyers is about to become the one your dashboard cannot see.

It is fixable. This article shows what changed and when, the six ways it hurts your pipeline, how the gap plays out in Georgia, Florida and North Carolina, which data you still have, and the measurement setup that keeps you honest about AI search after November 9.

Nov 9
Semrush Enterprise AIO stops tracking Microsoft Copilot
21.6%
of US work requests to ChatGPT are about making decisions
0.2%
of tracked web visits came from AI chatbots in a year
15
Fortune 500 headquarters in the Atlanta market in 2026

What changed with Copilot tracking in October 2026

Semrush told its Enterprise AIO customers by email on October 28, 2026 that it will stop tracking Microsoft Copilot on November 9, because changes to Copilot's access rules stop AI visibility platforms from collecting the answers. Tracking for ChatGPT, Gemini, Google AI Mode and the other assistants it covers carries on.

What the Semrush notice says

The notice is two short paragraphs. According to PPC Land's report on the Semrush notice, the company wrote that recent changes to Copilot's access requirements "prevent AI visibility platforms, including Semrush Enterprise AIO, from collecting the data needed to track Copilot performance." Note the plural. Semrush describes a limit on the whole category of tools, not a choice it made alone.

Customers get two options for the prompts they were tracking in Copilot. They can return those prompts to their overall balance and use the capacity on other assistants. Or they can keep the historical Copilot data, in which case the prompts keep counting against the balance even though no new Copilot data arrives after November 9. The notice sends customers to their Customer Success Manager and gives no deadline.

Copilot was a late addition for Semrush. It joined the product on December 1, 2025, pitched on its presence across work devices and software. Semrush's pricing page names nine tracked assistants for its AI Optimization product. Without Copilot, eight remain.

Why Microsoft's sign-in rule breaks the trackers

Microsoft's sign-in rule breaks trackers because AI visibility tools work by sending thousands of buyer-style prompts to each assistant and reading the answers, and that method needs open, repeatable, anonymous access. Microsoft's Copilot terms of use, effective August 18, 2026, say you must be logged in with a Microsoft account to use Copilot. The same terms tell users not to use bots or scrapers to access it.

The terms also draw a line you should notice. They do not apply to people who sign in with a work, school or organizational Entra account. Those are your enterprise buyers using Copilot Chat inside Microsoft 365, and no outside tracker ever saw their answers in the first place.

What this means for a Georgia B2B team

For a Georgia B2B company, the change lands on the buyers who matter most. The Atlanta market holds 15 Fortune 500 headquarters, tied with Minneapolis for seventh place in the country, according to RealPage's June 2026 count. Where a company runs Microsoft 365, its staff can meet Copilot Chat inside Outlook, Teams and Word. If you sell software, logistics services or professional services into Atlanta's corporate buyers, Copilot is a research tool on their desk that you can no longer sample from outside.

Six ways the Copilot blackout hurts your B2B pipeline

The Copilot blackout hurts your pipeline in six concrete ways, from a chart that misleads your board to buyer research you cannot attribute. Each one has a fix, and at least two bite harder in Georgia than elsewhere.

1. Your Copilot line drops to zero and reads like a loss

A share-of-voice chart that stops receiving Copilot data on November 9 will show a cliff, not a gap. Your CEO or board sees "Copilot visibility: 0%" in the December report and asks what went wrong. Nothing went wrong in your market. The data feed ended. Annotate every AI visibility chart with the November 9 date now, and remove Copilot from any blended score before the December numbers land.

2. You pay for prompts that return nothing

Keeping your Copilot history means those prompts keep eating your prompt balance with no new answers coming back. Returning them frees capacity but cuts your history. For most B2B teams the right call is to export the Copilot history to a spreadsheet first, then return the prompts and reuse them on ChatGPT and Google AI Mode, where your buyers also research.

Do this before November 9: export every Copilot prompt, answer and citation your tool holds. Once the prompts are returned to your balance, the history may go with them. A CSV in your shared drive is the only Copilot baseline you will have in 2027.

3. Your best Atlanta buyers were never visible anyway

Enterprise buyers in Atlanta use Copilot through work Entra accounts, and Microsoft's consumer terms do not cover those sessions at all. So even before November 9, the Copilot data you paid for described the consumer product, not the version a procurement lead at a Midtown or Alpharetta headquarters actually uses. If your Georgia pipeline leans on Fortune 500 accounts, your Copilot number was always a proxy at best.

4. Buyers decide inside the chat, then arrive as "direct"

Work use of AI assistants is mostly about deciding. OpenAI data published on October 26, 2026 shows making decisions and solving problems was 21.6% of US work requests to ChatGPT in the first quarter of 2026, per PPC Land's breakdown of the OpenAI Signals data. Requests for advice rose from 5.5% to 9.0% in a year. A buyer who builds a vendor shortlist in a chat window then types your name into a browser. Your analytics files that visit as direct or branded search, and the assistant gets no credit.

5. Referral traffic undercounts AI by a wide margin

AI referral traffic is too small to judge AI search by. OneLittleWeb's traffic dashboard estimates that AI chatbots sent about 0.2% of all visits to the sites it tracks over twelve months, while direct and search sent 92.0%. If your B2B team judges AI search by referral sessions, you will cut the program that shapes the shortlist because it sends almost no clicks.

6. Public-sector buyers research inside approved tools only

State agencies in Georgia must follow statewide AI rules before buying or regularly using any AI system, and staff request approval for each tool. For a Georgia B2B vendor selling to agencies in Atlanta, that means the buyer's research happens inside whichever tools the state approved, behind a government login. You will not see those answers in any third-party tracker, so your public-sector content has to win on the pages those tools read.

How the Copilot gap plays out in Georgia, Florida and North Carolina

The Copilot gap is widest where large employers and state agencies run Microsoft 365, and Georgia, Florida and North Carolina each have a different mix of both. The table below sets the three side by side, and the sections after it explain what each state means for your pipeline.

Georgia: Atlanta's headquarters buyers sit behind work accounts

Georgia is the sharpest case because its buyers are concentrated. The state holds 16 Fortune 500 headquarters in 2026, and 15 of them sit in the Atlanta market. For a B2B vendor in Atlanta or Marietta, a handful of corporate campuses can make up a large share of the pipeline, and those campuses research with Microsoft 365 work accounts that no AI visibility tool can sample.

The public side is just as closed. The Georgia Technology Authority's AI program tells agencies to follow the Enterprise AI Responsible Use Policy, PS-23-001, and the AI Responsible Use Standard, SS-23-002, before purchasing, developing or regularly using an AI system. Employees request approval to use an AI tool or pick from tools already approved for state use. If you sell to Georgia agencies, your buyer's AI research runs inside that approved list, not on the open web where trackers look.

The practical result for a Georgia team: build your measurement around what you control. That means your own site's citation data from Microsoft, deal-level questions in your CRM, and branded search trends in metro Atlanta. A Savannah logistics software vendor selling to port operators faces the same rule: the decision happens in a signed-in tool, and your proof has to come from the deal, not the dashboard. Our Georgia market page covers how we approach the state.

StateFortune 500 headquarters, 2026State agency AI ruleWhere buyer research hidesFirst fix
Georgia16 statewide, 15 in the Atlanta marketGTA policy PS-23-001 and standard SS-23-002, approved tools onlyWork Entra accounts at Atlanta headquarters and state-approved toolsBing Webmaster Tools AI Performance plus a "how did you find us" CRM field
Florida22No statewide AI law; SB 146 died on March 13, 2026Agency-by-agency tools and mid-market firms in Miami, Tampa and OrlandoTrack ChatGPT and Google AI Mode prompts by metro, export Copilot history
North Carolina12NCDIT Copilot for Microsoft 365 policy, licenses through agency CIOsLicensed Copilot seats in state agencies and Charlotte banksWrite pages Copilot can cite and watch grounding queries weekly

Florida and North Carolina: the same gap, different gatekeepers

Florida has more large headquarters than Georgia, with 22 Fortune 500 companies according to the 2026 state breakdown of the Fortune 500, but no single statewide rule for how agencies use AI. Florida Senate Bill 146, which would have required the Florida Digital Service to study state agency use of AI, died in the Governmental Oversight and Accountability committee on March 13, 2026. So each Florida agency picks its own tools. For a B2B vendor in Miami, Tampa or Orlando, that means public-sector buyers may use Copilot, ChatGPT or nothing, and you should track ChatGPT and Google AI Mode prompts by metro rather than assume one assistant.

North Carolina is the most Copilot-heavy of the three on the public side. The state's Copilot for Microsoft 365 policy, version 1.1 from February 26, 2025, tells agencies to buy licenses through the purchase requisition process, with agency CIOs requesting seats, and to train users first. North Carolina holds 12 Fortune 500 headquarters in 2026. If you sell into Raleigh agencies or Charlotte banks, Copilot is a licensed, sanctioned research tool for your buyer, which makes Microsoft's own citation data the most useful signal you have there.

The Copilot data you can still get after November 9

You can still get Copilot data after November 9, but only from Microsoft itself and only for your own site. Third-party tools lose the view of the answer, while Microsoft keeps showing you when its AI cites your pages.

Bing Webmaster Tools AI Performance

Bing Webmaster Tools added an AI Performance dashboard in public preview on February 10, 2026. It reports total citations, cited pages, page-level citation activity and grounding queries across Copilot and AI summaries in Bing. Grounding queries are the phrases Microsoft's AI used to retrieve your content. On October 13, a beta called Topics Portfolio began showing queries where rivals are cited and you are not. Microsoft calls the data a sample and the figures directional, so treat it as a trend line, not a census.

Why Bing still matters while its traffic falls

Bing still matters because it is the retrieval layer behind Copilot, even as its own web visits fall. OneLittleWeb estimates Bing lost 20.13 billion browser visits in the twelve months to September 2026, a 45.64% drop, per PPC Land's report on the ranking. Microsoft disputes the direction. Either way, the B2B mistake is to drop Bing indexing work because Bing sends fewer clicks. If Bing cannot crawl and index a page, Copilot cannot cite it.

What no tool can show you

No tool can show you the answer a buyer got inside a signed-in work account. Signed-in sessions add memory and personal context, so even the assistants that trackers still sample return answers that differ by user. Accept that AI search measurement is now a sample plus a deal-level question, and build your reporting around that pair.

Should you keep paying for AI visibility tracking

You should keep paying for AI visibility tracking if you use it to find content gaps, and cut it if you only use it as a board metric. The tool still covers the assistants your Georgia buyers use outside work, and it still shows which competitors get named for which questions.

When tracking still earns its fee

Tracking earns its fee when someone acts on it every month. That means a list of buyer questions where a competitor is named and you are not, turned into a page brief, published, and checked again 30 days later. If your team does that, the loss of Copilot reduces coverage but keeps the loop working.

When it does not

Tracking stops earning its fee when the only output is a share-of-voice percentage on a slide. Only 16% of brands systematically tracked AI visibility when the IAB published its measurement framework in August, per the same PPC Land report on the Semrush notice. Being in that 16% helps only if the number changes what you publish.

A simple test: list the last three pages your team published because of an AI visibility report. If you cannot name three, your tracking budget is paying for a chart, and the November 9 change is a good moment to move that money into content and CRM measurement.

What works for B2B AI search measurement now

What works now is measuring AI search from three sides at once: the citations your own site earns, the questions trackers can still sample, and what buyers tell you in the deal. Each side covers a blind spot of the others.

Measure citations at the source

Connect every domain you own to Bing Webmaster Tools and Google Search Console this week. Check AI Performance grounding queries weekly. When a grounding query matches a buyer question, such as "best freight audit software for mid-size shippers", you know which page Copilot pulled and can strengthen it. This works for enterprise buyers too, because a work Copilot answer that cites you still sends grounding data about your page.

Write pages an assistant can lift

Assistants cite passages, not whole sites. A B2B page that answers one buyer question in its first sentence, names the buyer's industry and city, gives a number with a unit, and compares options in a table is far easier to cite than a features page. Build one page per real shortlist question your sales team hears, and give Atlanta, Charlotte and Tampa buyers their own examples where the answer genuinely differs.

Ask the buyer, and log it

Add a required "how did you first hear about us" field to your demo form with an option for AI assistants, and a matching field in your CRM opportunity. Ask the question again on the first discovery call. Self-reported attribution is noisy, but it is the only signal that reaches inside a signed-in work account. Tie the answer to pipeline value, not lead count.

Keep Bing and Copilot in your content plan

Keep Bing indexing healthy with IndexNow or a clean sitemap, and fix crawl errors in Bing Webmaster Tools. A buyer at an Atlanta headquarters who asks Copilot for a shortlist gets an answer grounded in what Bing indexed. Losing the tracker does not mean losing the channel.

How Leadnox helps B2B teams measure AI search

Leadnox helps B2B teams in Georgia, Florida and North Carolina rebuild AI search measurement around data they control, then publish the pages assistants cite. We start with what your buyers ask and where your pipeline comes from, not with a visibility score.

A measurement setup that survives tracker changes

We connect Bing Webmaster Tools, Google Search Console and your CRM, export your Copilot history before November 9, and set up a monthly report that pairs AI citations with self-reported attribution and pipeline value. When a tool drops an assistant, the report keeps working because it never depended on one vendor.

Content built for shortlist questions

Our B2B marketing team turns your sales calls into a list of shortlist questions, then writes answer-first pages with tables, named places and defensible numbers. We plan Atlanta, Charlotte and Florida metro pages where buyers genuinely ask different questions, and we track whether assistants cite them.

Paid and search working together

AI answers shape the shortlist, and branded search and paid media catch the buyer afterwards. We align your branded search, LinkedIn and Google Ads so the buyer who arrives as "direct" after a chat meets a page that confirms the answer they read.

How to measure AI search for a B2B pipeline

Measure AI search with four numbers: Copilot and Bing citations, tracked answer share on the assistants still covered, self-reported AI attribution, and branded search demand. Together they show whether assistants name you and whether that turns into pipeline.

Citations and grounding queries

Track total citations and the count of buyer-intent grounding queries in Bing Webmaster Tools AI Performance each week. Good looks like a rising count of grounding queries that match your sales team's real questions, not just your brand name.

Answer share on covered assistants

Track how often ChatGPT, Gemini and Google AI Mode name you for 30 to 50 buyer questions, split by metro where your tool allows it. Good looks like steady gains on the questions you wrote pages for, measured 30 days after each page goes live.

Self-reported AI attribution

Track the share of new opportunities where the buyer says an AI assistant was part of their research, and the pipeline value of those deals. Good looks like that share holding or rising after November 9, which tells you the channel kept working while the tracker went dark.

Branded search demand

Track branded search impressions in Google Search Console and Bing, filtered to Georgia, Florida and North Carolina where you can. Good looks like branded demand rising in metro Atlanta after you publish shortlist pages, because buyers who read your name in an answer search for it next.

Frequently Asked Questions

Semrush says changes to Copilot's access requirements stop AI visibility platforms from collecting Copilot answers. Microsoft's terms, effective August 18, 2026, require users to log in with a Microsoft account and ban bots and scrapers. Semrush told Enterprise AIO customers on October 28 that tracking ends November 9, while tracking for ChatGPT, Gemini and its other assistants continues.

A Georgia B2B company should combine Bing Webmaster Tools AI Performance citations, tracked answers on ChatGPT and Google AI Mode, and a required "how did you hear about us" field in the CRM. Atlanta's 15 Fortune 500 headquarters research in work Microsoft accounts no tracker can see, so deal-level attribution matters more there than any dashboard score.

Yes. Bing Webmaster Tools added an AI Performance dashboard in February 2026 that reports citations, cited pages and grounding queries across Copilot and AI summaries in Bing. It covers only your own verified site and Microsoft calls the data a sample, but it still works after November 9 because it comes from Microsoft rather than an outside tracker.

AI search matters more than its traffic suggests. OneLittleWeb estimates AI chatbots sent about 0.2% of tracked web visits in a year, yet OpenAI data shows 21.6% of US work requests to ChatGPT are about making decisions. Buyers build shortlists in the chat and then visit you directly, so referral sessions undercount the channel badly.

Export the Copilot history to a spreadsheet first, then decide. Keeping the history means those prompts keep counting against your balance with no new data after November 9. Returning them frees capacity for ChatGPT and Google AI Mode. For most B2B teams, export and return is the better trade, but confirm the steps with your Customer Success Manager.

Is your AI search reporting about to go blind in Georgia?

Leadnox rebuilds AI search measurement and content for B2B teams in Georgia, Florida and North Carolina, so you can see which answers name you and which buyers they send. Start with a free audit.

Get My Free AI Visibility Audit