Home / Feedback Score

Why Did My Feedback Score Drop? Finding the Cause

Published July 10, 2026· Feedback Score

Why did my feedback score drop? It’s the right question asked at a hard time — because since late 2024 the score is invisible, so you’re usually asking it after the damage showed up somewhere else: CPMs drifting, delivery softening, rejections ticking up. The score dropped weeks ago; you’re just receiving the letter now.

The good news: score drops have a short list of causes, they’re findable with a timeline exercise, and each has a specific fix. Here’s how to trace yours.

First, understand the lag

Meta’s feedback score is built from post-purchase surveys, and surveys trail experience. A buyer orders, waits for shipping, tries the product, maybe fights for a refund — and answers Meta’s survey somewhere along that arc. Then enough surveys have to accumulate to move a rolling score.

Practical consequence: the cause of today’s drop is two to six weeks old. Whatever you changed this week isn’t showing yet, and whatever’s hurting you now happened around a month ago. So the diagnostic isn’t “what’s wrong today” — it’s “what did my business do in the past four to eight weeks that a buyer would report.”

The usual suspects, ranked

A shipping slip. The most common cause by far. A supplier holiday (Chinese New Year is the classic), a customs backlog, a carrier change, a viral sale that outran fulfillment — anything that pushed delivery past what buyers were told. “Arrived late” alone is reported to be survivable; the trouble is it rarely travels alone, because late orders generate support tickets, refund requests, and “never received” complaints — the account-killer category. Dropshippers live closest to this cliff, and the shipping-times problem has its own playbook.

A product batch. The supplier changed materials, a new SKU shipped with defects, quality drifted while nobody checked. Suddenly a slice of buyers holds a product that doesn’t match the ad — and “not as advertised” plus “low quality” are the most damaging categories operators report. Batch problems are sneaky because you changed nothing; the score dropped anyway.

Refund friction. A policy tightened, a support inbox backed up, or a new VA started fighting refund requests. The survey specifically asks how refunds went: refused or difficult refunds, and buyers forced through their banks, are among the worst answers possible. If your timeline shows a support change before the drop, you’ve likely found it.

A surprise-charge wave. Launched a subscription offer, added a post-purchase upsell, or a rebill cycle hit customers who forgot they’d subscribed. “Unexpected charges after payment” is a heavyweight complaint category, and it arrives in waves timed exactly one billing cycle after the offer change.

An expectation-setting change. New creative that oversells, a landing page rewrite with bolder claims, a new angle that attracts buyers the product will disappoint. The ads got better; the survey answers got worse.

External causes. A copycat store confusing buyers into complaining about you, a carrier meltdown, a platform glitch double-charging. Rarer — but real, and worth ruling in or out because the fix is different: copycat situations need takedowns, not operational surgery.

The timeline exercise

Take a sheet and two columns. Left: every operational event of the last eight weeks — new products, supplier changes, sales spikes, offer changes, support staffing, shipping notices. Right: the performance timeline — when CPMs started drifting, when delivery softened, when ads slowed their spending.

Now shift the left column forward by two to four weeks and look for the collision. In our experience the match is usually embarrassing in its obviousness: the sale that doubled order volume sits three weeks before the CPM inflection; the new subscription offer sits one billing cycle before it.

Cross-check against the complaint categories Meta is known to track — product not as advertised, low quality, unexpected charges, poor support, late arrival, never received — and ask which one your candidate event would generate. If you have any Meta rep access, this is the moment to ask for your customer feedback standing; the category breakdown turns your hypothesis into a confirmation.

Can’t find the match? Send us your two timelines — free feedback score audit on Telegram, and a second pair of eyes usually spots it fast: Message us on Telegram.

Fixing it without making it worse

Once you’ve found the cause, the sequence is standard: fix the specific operational failure, make refunds easy while the bad cohort works through the system (a refund given repairs signal; a refund fought deepens the damage), and keep reduced-but-real volume running so positive surveys refill the rolling window. Expect the recovery timeline to run two to three weeks before performance responds, one to two months to stabilize fully.

And two don’ts, both tempting mid-drop. Don’t scale into it — spending harder against an unfixed cause manufactures negative surveys at a faster rate. And don’t burn the setup — the score follows the business entity, so new ad accounts and fresh pages inherit the problem while forfeiting your history.

The drop had a date, and that date has an event. Find it, fix it, refund generously through the transition, and let the window turn over. The score system is brutal about lag, but it’s honest: it only ever reports what your buyers actually experienced.

Tags: feedback score, score drop, diagnosis, meta ads