In early October, Meta performance marketer Jason Yim (@jasonyimco) shared a short but critical update:

"Your customer feedback will be a much stronger signal to Meta in the ad auction… Beginning Oct, this feedback will be weighted more strongly in the auction."

He clarified three key points:

• Meta polls customers who bought from you via attributed Meta ads, and uses that feedback directly in the auction to reduce negative experiences and amplify positive ones.

• It only affects your own account – your competitors bad feedback can’t hurt you.

• This is one of many backend updates being rolled out, including "reducing self-competition within Meta ads auction" and the broader Andromeda retrieval overhaul.

In other words: Meta quietly turned the "customer feedback" dial up. If you ignore it, you're now paying a hidden tax in every auction.

What Is Meta's Customer Feedback Score?

Meta has tracked post-purchase satisfaction for years through short surveys sent to people who clicked ads and bought. These ask whether:

• The product matched expectations

• Shipping was timely

• Customer service, refunds, and exchanges were handled well

Those responses roll up into a 0–5 "Customer Feedback Score" for your Page/business. You typically see this under Account Quality / Business Support. Meta's own documentation confirms that ongoing survey results inform a business's Customer Feedback Score, and that those surveys explicitly measure whether product quality, shipping, and customer service met expectations.

From Side Penalty to Auction Signal: What Changed

The Old System (2019–Oct 2025)

When Meta rolled this out around 2019, feedback scores operated as threshold-based penalties:

Score collection: After a purchase from a Facebook/Instagram ad, users would see surveys asking about product quality, shipping speed, order accuracy, refunds/exchanges, and support. Responses accumulated over a rolling 30–90 day window.

Threshold penalties: The system was coarse and reactive:

• 4–5: No penalty

• 3–4: Room for improvement warnings

• 2–3: "Poor" status—visible warnings, CPMs up 20-30%, delivery capped

• 1–2: Aggressive throttling, high CPM, risk of page-level restrictions

• 0–1: Ads disabled until improvements made

The score felt like a separate "compliance meter" beside the auction. You optimized bids, creatives, and audiences first... and only looked at feedback when something broke badly.

Enter Andromeda: The Retrieval Engine Rewrite

In late 2024, Meta publicly introduced Andromeda — a new ML-heavy personalized ads retrieval engine. Instead of rule-based audience hacks, Andromeda uses large neural networks running on NVIDIA Grace Hopper and Meta’s own MTIA chips to pick ad candidates based on holistic user and ad signals

Key shifts:

• Retrieval now considers far more signals—creative content, behavioral patterns, and "ad quality" factors—before ranking

• Meta reports +8% "ads quality" improvement on selected segments after deploying Andromeda

• Creative and experience signals now matter more than micro-targeting

Meta's auction docs describe total ad value as:

Total Ad Value = Bid + Estimated Action Rate + Ad Quality/Relevance

What Jason Yim's update effectively says is: Meta has turned "customer feedback" into a stronger part of that Ad Quality/user-value term, starting October.

Your post-purchase reality is now a heavier, always-on input into how often—and how cheaply—you win auctions.

The Hidden Auction Tax: Before vs After

Before: Threshold-Driven Penalty

Your Page sits at 2.3/5:

• Meta marks you as "poor"; you see a warning in Account Quality

• CPM creeps up maybe 20–30% vs. a 4.5-score Page

• If you fix operations and rise above ~3, the explicit penalty eases

• It felt like a coarse lever: you were either fine, "under penalty", or in trouble

After: Continuous Auction Tax

Now the same Page at 2.3/5 after October's weighting change:

Retrieval: Andromeda must choose from millions of ads. If two ads have similar bid and expected action rate, it now leans harder into user-experience history. The higher-feedback advertiser gets picked first.

CPM: Case studies already show tight correlation between poor feedback scores and higher CPM/restricted delivery. With stronger weighting, a 2.3 score can mean a 50%+ CPM gap versus a 4.5 score at the same bid.

Creative half-life: Under Andromeda, creative is a core signal. When recent buyers complain "not as described" or "late", the retrieval system marks your asset as less safe to show to similar users—shortening its life even if CTR and CPA look fine initially.

Instead of a mostly static "red bar penalty", you now get a continuous, quietly compounding tax:

• Slightly fewer impressions per auction

• Higher CPM at the same bid

• More "limited" delivery and volatile scaling

• More sensitive response to any new complaints

What Meta Actually Measures (And Why It Matters)

If you strip away the ML complexity, Meta's surveys keep coming back to four pillars:

1. Product Honesty

• Ads and PDPs must accurately represent what you sell: quality, fit, colors, use-cases

• Sizing charts must match the markets you sell into

• No aspirational photography that oversells reality

2. Delivery Clarity

• Clear ETAs on PDPs and in order flows

• Visible shipping & returns policy before purchase

• Reliable tracking and proactive notifications on delays

3. Customer Service Transparency

• Honor advertised returns & exchanges

• Clear reply-time promises and meet them

• Proactive communication on issues (stockouts, split shipments, etc.)

4. Operational Capacity

• Don't scale spend beyond what your warehouse, 3PL, or support team can handle

• If you can't fulfill demand, throttle ads—don't hide the backlog

These are exactly the dimensions Meta asks about in surveys: "Did the product arrive on time?", "Was the product as described?", "How was the support?"

Knowing what to fix and actually fixing it at scale are different problems. Most brands understand these principles—the challenge is diagnosing which specific gaps are poisoning your feedback score, then systematically closing them without disrupting operations.

The Real Problem Isn't Media Buying. It's Operations.

Most feedback disasters don't start in Ads Manager. They start with:

Expectation gaps: TikTok-style ads promising premium quality; AliExpress-quality product arrives. Edited product photos that don't match color, fit, or size.

Shipping opacity: "Fast shipping" claims with 3–4 week delivery. No clear ETA and no tracking until the customer is already angry.

Support deserts: Slow first response (days, not hours). No clear SLA, no escalation path, templated non-answers.

Returns mazes: Hidden returns portal, surprise return shipping costs, long refund times.

None of that is a media-buying tweak. It's systems design. And in an Andromeda world, those system failures bleed directly into the auction.

How Scoreify Maps the Invisible Tax

For the last few years, Scoreify has been working in exactly this "hidden signal" layer—helping advertisers understand why Meta "doesn't trust" them, and how that trust (or lack of it) shows up in CPM, delivery, and scale.

We treat feedback and reputation as diagnosable infrastructure, not vibes.

What Scoreify Actually Does

Feedback Risk Audit

• Map where expectations break: ad → PDP → product

• Audit shipping transparency, ETAs, and tracking

• Assess support SLAs and returns friction

• Highlight where survey responses are likely to be "Below expectations" based on your actual flows

Truth-Aligned Creative Tests

• Replace one polished hero with a "what you actually get" asset (real lighting, real fit, real unboxing)

• Hold bids steady and observe CPM + stability vs. your previous best creative over 7–21 days

• Measure whether alignment with reality reduces your auction tax

Telemetry & Correlation

• Pipe WISMO ("Where is my order?"), first-response time, on-time dispatch %, and "not as described" data into a simple weekly view

• Overlay these with CPM and delivery stability so you can see which operational fixes actually move the auction

• Connect dots that Ads Manager never shows you

Controlled Rollouts

• Implement changes in sequences, with deliberate holdouts

• Know which adjustments dropped CPM vs. which were noise

• Build a repeatable playbook for your business

This October update is, from our perspective, the most important feedback-related change Meta has shipped so far: it finally aligns the algorithm with what we've been measuring for years.

Worried your feedback is quietly taxing every impression?

Scoreify's Feedback Diagnostic maps the exact operational and creative issues driving negative surveys — then ties them to your CPM and delivery patterns.

→ Book a Feedback Score Diagnostic
→ View a Sample Feedback Report

Bottom Line

Meta's always cared about post-purchase satisfaction, but now feedback is being weighted more strongly inside the auction itself—not just as a side-penalty.

That turns feedback into a hidden auction tax: if customers feel misled, delayed, or ignored, you quietly lose delivery and pay more per impression—often before Ads Manager surfaces anything obvious.

The fix isn't another hacky account structure or bid strategy. It's operational truth—accurate promises, transparent shipping, responsive support, and easy returns—tracked and tested the way you'd test creative.

Scoreify's entire reason to exist is to make those invisible penalties visible—and fixable—so you can actually enjoy what Andromeda was supposed to deliver: better performance for brands that treat customers well.

Stop Guessing. Start Fixing the Signals Meta Actually Uses.

Meta just made customer feedback a stronger auction input. If you keep treating it as a side metric, you'll keep paying more for worse traffic.

Scoreify gives you a concrete, evidence-first roadmap to improve feedback, reduce CPM, and unlock scale in the Andromeda era.

→ Get Started with a Feedback Score Assessment