In early October 2025, Meta performance marketer Jason Yim noted that customer feedback would be weighted more strongly in the auction from that month.
Meta already polled people who bought from attributed ads. The change is that those answers now pull harder on delivery.
Three details matter. The surveys only affect your own account — a competitor's bad reviews do not tax you. The questions are about whether the product, shipping, and support matched what people expected. And this landed on top of Andromeda, Meta's retrieval rewrite, which already scores ads on a wider set of quality signals than interest stacks.
What the score is
Responses roll up into a 0–5 Customer Feedback Score on the Page or business. You typically see it under Account Quality. Meta's own documentation says ongoing survey results inform that score, and that the surveys measure product quality, shipping, and customer service against expectation.
For a GLP-1 or telehealth advertiser, “product” is the medication or the consult. “Shipping” is whether the vial, the compound, or the kit arrived when the landing page said it would. “Service” is whether someone could reach a clinician or a pharmacy tech when something went wrong. The survey does not know you are a clinic. It knows whether the person felt misled.
From a red bar to a continuous tax
From around 2019 until this change, feedback behaved like a threshold penalty. Roughly 4–5 meant no extra friction. 3–4 was a warning. 2–3 was “poor,” with CPMs up and delivery capped. Near 1, ads could be disabled until the score moved. You treated it as a compliance meter beside the auction, and only looked when something broke.
Andromeda changed retrieval: large models pick candidates from a much bigger pool using creative, behaviour, and ad-quality signals, not just the audience you typed in. Total ad value is still bid plus estimated action rate plus ad quality. The October update makes customer feedback a heavier part of that quality term.
A Page sitting at 2.3 versus 4.5 is no longer only a warning in Account Quality. Retrieval prefers the higher-feedback advertiser when bid and predicted action rate are close. Case work in the category already showed a tight link between poor feedback and higher CPM. With stronger weighting, that gap can be 50% or more at the same bid. Creative also dies faster: if recent patients say the offer was not as described, the system treats the asset as less safe to show to similar people, even if CTR still looks fine.
What Meta is actually asking
Strip the ML and the surveys keep returning to four things. Honesty: the ad and the site must match the consult, the compound, the dose, the timeline. Delivery: clear ETAs, a visible shipping and returns path, tracking when a pharmacy is slow. Support: advertised refunds and exchanges honoured, replies in hours not days. Capacity: do not scale spend past what the pharmacy, the 3PL, or the care team can fulfil.
What Meta is actually asking
Strip the ML and the surveys keep returning to four things. Honesty: the ad and the site must match the consult, the compound, the dose, the timeline. Delivery: clear ETAs, a visible shipping and returns path, tracking when a pharmacy is slow. Support: advertised refunds and exchanges honoured, replies in hours not days. Capacity: do not scale spend past what the pharmacy, the 3PL, or the care team can fulfil.
Most feedback disasters in this vertical do not start in Ads Manager. They start when a UGC ad promises a result the landing page cannot, when “ships in 3–5 days” becomes three weeks because the compounder is backlogged, when the inbox is a desert, or when the return path is buried. That is operations. Under Andromeda it is also auction input.
The account still sits underneath it
Fixing surveys does not replace account quality. A certified brand can run honest creative and still pay a trust tax if the ad account itself is treated as risky. Feedback is one lever. The HiVA tier on the Business Manager, the Page, and the ad account is another. They compound.
The practical order for a health advertiser is: tell the truth in the ad, keep fulfilment inside the promise, answer patients, and run the spend through infrastructure Meta already trusts in this category. Bid tricks will not outrun a 2.3.
