Meta is getting ready to shut down Business Managers over unhappy customers, according to what its account managers have been telling large advertisers this September. Not for policy strikes or rejected creatives. If your buyers are frustrated enough to put you in the bottom 5% on customer experience, your Business Manager goes. The report comes from Artem Smirnov, who has spent 10 years in user acquisition and collaborates with Birch as an expert; it appeared first on his channel.
Some of what's behind that is documented by Meta and easy to check. The rest is what practitioners with the biggest accounts are seeing. You'd act on the two differently, so we've kept them apart.
What the score is: the documented layer
Meta has surveyed buyers after ad-driven purchases for years. The answers roll up into a feedback score from 0 to 5, and Meta documents two of them: a Page feedback score and a business portfolio feedback score. The portfolio is the current name for what most of us still call Business Manager, and the score used to be called the customer feedback score. The thresholds Meta publishes for the Page score are blunt, and practitioners apply the same bands to the portfolio score. Between 1 and 2, your ads get a delivery penalty and reach fewer people for the same budget. Below 1, advertising stops. When a Page score drops to 3 or below, Meta says it sends an email and an in-product notification as the Page nears a restriction. The survey asks about four things, and those four are the levers: whether the product arrived, its quality, delivery speed, and customer service.
Two details catch people out. If you sell through Commerce Manager, the surveys go out around the estimated delivery date in your shipping settings, so an optimistic window means buyers get asked about a parcel that hasn't arrived. And the portfolio score is shared by everything in the portfolio, so one struggling product line drags the rating every campaign inherits.
What's changed is visibility, and here the sources disagree. Practitioners have reported since late 2024 that the self-serve view left the account-quality page (now part of Business Support Home) and hasn't come back; what's left in Business Suite is a ratings and reviews section, which is something else (Ad Account Clinic). Meta's own Business Support Home page still lists reviewing your feedback score as one of its functions. Either the documentation is stale or the removal is narrower than reported. If you can't find the number, your performance data is the dashboard you have.
The reported layer: a score behind the score
What changed this month isn't the feedback score. It's what practitioners with the largest accounts say has been built on top of the same surveys.
Artem's account is the most detailed we've seen, and it stays his throughout this section. Since autumn 2025, he reports, Meta has scored all advertisers on customer experience, not only e-commerce. The metric started as ACE for e-commerce clients and picked up new names along the way: PQ for page quality, then CES for customer experience score. Until September it moved costs quietly—CPMs went up for advertisers whose buyers reported frustration and eased for those whose buyers didn't. Then Meta's managers started telling large advertisers that the bottom 5% would have their Business Managers blocked, with no date attached.
There's a dashboard, but only advertisers spending well into seven figures a month can see it; the score itself seems to be computed from about $300k in monthly spend. The scale runs from the bottom 5% to the top 30% of all advertisers. The post-purchase survey is worded through the negative, with the critical answers listed first. And the complaint categories the report surfaces are unexpected payment charge, low product quality, and bad support experience.
Why this is moving now
Two dated events sit behind the timing.
In November 2025, Reuters published an investigation based on internal documents that put 10.1% of Meta's 2024 revenue, about $16bn, down to ads for scams and banned goods. Meta called the figure rough and overly inclusive and said later assessments came in lower. In late August 2026, Poland asked the European Commission to fine Meta €250m over scam ads it says Meta moderates poorly.
Regulators and the press keep landing on the same question, and scoring advertisers on how buyers feel is the mechanism Meta can point to as its answer. You don't need to follow the politics. It's enough to notice which way enforcement is tightening: around the buyer's experience.
What it means for your business
The consequence is the same for everyone; the exposure depends on what you sell. Three situations cover most accounts.
Physical goods. If you sell through Commerce Manager, the survey lands on the delivery date in your shipping settings, so that window is a scoring input rather than a logistics detail. If checkout happens on your own site, Meta doesn't document how the survey is timed, and it's safest to assume it lands before the parcel does. Either way, because the score is per portfolio, one slow product line or one batch of damaged orders is inherited by every campaign in the account, including the ones doing well.
Subscriptions and apps. Two of the three reported complaint categories, unexpected payment charge and bad support experience, describe a subscription business with an unclear paywall and a cancel flow that needs a support ticket. Artem's view is that most of his industry sits in the bottom 5% by default for exactly that reason. If a trial converts to a charge the buyer didn't expect, the survey collects it.
Lead generation and B2B. There's no purchase to survey, so there's no documented score and never was. Artem's report says the advertiser-level scoring applies to everyone, not only e-commerce, but doesn't say how an account with no purchases gets measured. So a clean-looking account isn't evidence of a good rating, and performance is your reading here too.
Each of the three categories maps to something your team controls, and each is worth a look this week:
🔴 Unexpected payment charge
🟢 The price, the currency, and the recurrence are visible before the payment step, not after. Trial terms say what happens when the trial ends. The charge descriptor matches the brand the buyer remembers
🔴 Low product quality
🟢 The creative promises what actually arrives. If the ad oversells, the survey collects the difference. For Commerce Manager sellers, the shipping window in the settings matches reality, because the survey lands on the estimated delivery date
🔴 Bad support experience
🟢 A refund policy that can be found, a cancel flow that works without a ticket, and response times a buyer would call reasonable
What others are doing about it
Almost nobody outside Meta can see the number, so what practitioners describe is what they changed on their own side.
Artem fixes the paywall and the cancel flow first, on the reasoning that a buyer who can leave doesn't answer the survey angrily. He expects Meta to widen dashboard access and lower the spend thresholds over the coming months, and to adjust the scoring so advertisers who do the basics stop landing in the bottom band. He also reads the moment as the case for moving some spend to Google and AppLovin.
Jason Vaught, Director of Content and Marketing at SmashBrand, treats the feedback score as a measure of how well creative sets expectations and runs a three-point visual accuracy check—color, scale, texture—before any campaign launches, so the product image matches the final packaging.
Karina Tymchenko, founder of Brandualist, treats feedback as an auction variable that keeps affecting both CPMs and delivery quality, and says aligning ad messaging with the actual post-purchase experience is the first lever her team reaches for.
What to check on your accounts this week
With the number hidden from almost everyone, the question that pays off isn't "what's my score" but "what would tell me it moved". Two habits and three rules cover it.
Monthly: read the survey categories against your own operations: refund rate, the share of support tickets about charges or cancellations, delivery times against the window in your settings. None of these feed Meta's score directly—that's built from survey responses—but they're the closest proxies you can see for what the survey measures.
Continuously: watch for delivery getting more expensive for no creative reason. A sentiment problem can show up as CPM drifting up on ads that haven't changed. So can auction seasonality, frequency and audience saturation, a shift in placement mix, or a new competitor. CPM drift is a hypothesis to rule in, not a diagnosis: check whether the market moved with you first, and only then whether the drift is yours. You can also ask Meta MCP inside Birch AI whether that CPM move is market-wide.
The rules that make this workable are the ones in our Meta ads automation guide; automated rules are how the pattern stops being tedious to watch by hand.
- CPM drift alert. Notifies when CPM over the last 3 days sits above CPM over the last 14 days by a margin, scoped to ads older than a chosen number of days so new creative doesn't trip it. Take the margin from your own account.
- Refund-rate condition. Refunds aren't a Meta metric, but they are the kind of thing custom metrics exist for: if your team records them in a spreadsheet, a Google Sheets custom metric brings refund rate per campaign into a rule, so a campaign whose refunds cross a threshold gets flagged or paused before the survey responses arrive.
- Downtrend watch. Explorer's Downtrending Ads report flags ads with declining website purchases and conversion rate over the last 14 days, and you can create a rule from the report in one click.
Before switching any of these on, run the condition against last month's data. If it would have paused half the account, or fired every week, the threshold is a guess, not a signal.
What to keep on hand:
| Signal | Where it lives |
|---|---|
| Feedback score thresholds | Penalty below 2, ads off below 1; warning at 3 or below (Meta Help) |
| Survey timing | Around the estimated delivery date, for Commerce Manager sellers |
| Score visibility | Reported hidden since late 2024; Meta's docs still describe a view |
| Reported layer | ACE, bottom 5% facing Business Manager blocks, per practitioner reporting |
| Your early signal | Refund and cancellation share; CPM drift once the market is ruled out |
