On January 12, 2026, Meta permanently removed the 7-day view and 28-day view attribution windows from its Ads Insights API. Reported conversions dropped 15–40% overnight, with no change to campaigns, budgets, or targeting. If your ROAS looks worse this quarter, your ads almost certainly aren’t the problem — Meta changed how it counts, not how your ads perform. For WooCommerce stores making budget decisions from post-January dashboards, the wrong response is pausing campaigns. The right response is fixing measurement: click-match quality, incrementality testing, and a conversion record Meta can’t redefine on you.
Why Your WooCommerce ROAS Looks 15–40% Worse (And Your Ads Are Fine)
Before the change, Meta credited a conversion if a customer had simply seen your ad up to 28 days before buying — even if they never clicked. That’s the 28-day view window. Pair it with the 7-day view window (same logic, shorter horizon) and you had two attribution lanes stacked on top of standard click-based attribution. Every reported conversion included all three buckets combined.
Both view buckets are gone. The replacement is a single category called 1-day engage-through, and the rules are tighter: only actual outbound link clicks trigger an engagement signal now. Likes, profile visits, and passive impressions no longer count. The change hit the Ads Insights API directly — which means Ads Manager, Looker Studio connectors, third-party dashboards, and automation tools all saw the drop at the same moment (Conversios, 2026).
Meta also tightened the reporting horizon: historical data is now capped at 13 months for unique counts and 6 months for frequency breakdowns (Conversios, 2026). Year-over-year benchmarks built on older Meta data are partially walled off now.
The Before/After Math on an $80 AOV Store
Numbers make the shift concrete. Here’s a hypothetical WooCommerce store running $2,000/week in Meta Ads at an $80 average order value — same campaigns, same creative, same week-over-week budget across the January 12 boundary:
| Metric | Week of January 5 | Week of January 19 | Delta |
|---|---|---|---|
| Ad spend | $2,000 | $2,000 | 0% |
| Reported conversions | 50 | 32 | -36% |
| Reported revenue | $4,000 | $2,560 | -36% |
| Reported ROAS | 2.0x | 1.28x | -36% |
| Actual WooCommerce orders | 50 | ~48 | ≈0% |
The campaign spent the same. The store shipped roughly the same. Only the attribution lane narrowed — and the dashboard registered that narrowing as a ROAS collapse.
Your first diagnostic question: did your WooCommerce admin show the same revenue drop Meta reported? If WooCommerce says “48 orders this week” and Meta says “32 conversions,” the 16-order gap is exactly the view-through credit Meta used to take and doesn’t anymore. Nothing broke on your side. Meta just stopped claiming credit for impressions that turned into sales without a click.
You may be interested in: Why Google Ads, Meta, and GA4 Show Different Conversion Numbers for the Same WooCommerce Sale
Why Pausing Campaigns Is the Wrong Response
The most expensive reaction to a measurement change is an action taken on the new numbers as if they were comparable to the old. Three things WooCommerce store owners have been doing wrong across Q1 2026:
- Pausing “underperforming” campaigns. The ads didn’t get worse. The yardstick got shorter.
- Cutting budgets to defend an old ROAS target. ROAS targets set against old 28-day-view-inclusive measurement are no longer valid. You are comparing two different things.
- Switching from Advantage+ to manual bidding because “the AI broke.” Meta’s optimisation runs on the same conversion signals regardless of what the reporting surface displays. The AI didn’t break; the display did.
This is a measurement problem, not a performance problem. Treat it like one.
The Three-Part Recovery Framework
The question isn’t whether your reported ROAS is lower. It is. The question is what visibility you can rebuild given Meta’s new, narrower credit assignment. Three moves, in order.
1. Event Match Quality Above 7.0
With view-based credit gone, click-based attribution is the entire Meta budget. Whether Meta matches a conversion event to a click now depends almost entirely on Event Match Quality (EMQ). A strong EMQ score requires hashed email, phone, first name, last name, city, state, zip, country, external ID, and client user agent on every Purchase event.
Advertisers with Conversions API setups see 17.8% lower cost per result versus those without (Meta for Business, via PPC Land 2026) — and that gap widened after the window change, because CAPI is how you get the PII into the hashed-match pipeline in the first place. Pixel-only setups top out at what the browser leaks before ad blockers take it; pixel-only tracking can miss 30-60% of actual conversions due to browser restrictions, ad blockers, and iOS limitations (Conversios, 2026).
Getting EMQ above 7.0 is no longer a nice-to-have. It’s the recovery lever that replaces view-through credit with better click-through credit. See Why Your Facebook Event Match Quality Score Is Below 7.0 (And What It’s Costing You) for the step-by-step.
2. Incrementality Testing for Real Lift
Attribution windows — any attribution window — answer “what did Meta take credit for?” Incrementality answers “what sales would not have happened without Meta?” The second question is the one your CFO actually cares about, and it’s the only one that’s robust across reporting changes.
Standard incrementality design for WooCommerce: hold out 10–20% of a well-defined audience, run identical creative and offers on the rest, measure the delta in WooCommerce sales between holdout and test. If Meta is driving real lift, the test group outperforms the holdout regardless of what the Ads Insights API reports. If it isn’t, you find out quickly — on the right number, not on an attribution lane Meta can redefine.
3. First-Party Attribution in BigQuery
Here’s the structural problem: every store that uses Meta’s reporting surface as its single source of conversion truth gets re-measured every time Meta changes its windows. January 12 wasn’t the first time, and it won’t be the last.
The durable answer is a conversion count stored on infrastructure Meta doesn’t control. A WooCommerce order that hits your BigQuery table at checkout is there forever, at order-ID precision, with the full UTM plus referrer plus timestamp trail intact. When Meta redefines view windows next time, the BigQuery count doesn’t move. That’s the asymmetry: reporting is a subscription to somebody else’s definition of your sales; a warehouse is a record of your sales.
Why This Pattern Will Keep Happening
Platform-side attribution surfaces aren’t stable measurement frameworks. They’re optimisation aids for Meta’s ad auction, tuned for Meta’s commercial and regulatory position in a given quarter. View-through made sense when cookies were load-bearing; the current world runs on 1-day engage-through because that’s what survives Safari’s restrictions, iOS opt-outs, and the GDPR consent layer. That direction is one-way.
The same logic drove Meta removing the AEM 8-event cap in June 2025, and it will drive the next change too. Planning your measurement around what Meta reports today is planning to rebuild every six to twelve months.
Transmute Engine™ is a first-party Node.js server that runs on your own subdomain (e.g., data.yourstore.com) and writes every WooCommerce event into BigQuery as the authoritative conversion record — while separately feeding Meta’s CAPI with enriched EMQ data. When Meta changes its windows again, your BigQuery count stays put. You stop needing to rebuild your measurement every time a platform reshuffles its reporting surface.
Key Takeaways
- The date was January 12, 2026. Meta removed the 7-day view and 28-day view attribution windows from its Ads Insights API. Only 1-day engage-through remains.
- Reported conversions dropped 15–40% overnight. That’s a measurement artifact, not a performance collapse.
- Diagnostic question: did WooCommerce revenue drop by the same percentage Meta is reporting? If not, your ads are fine.
- Recovery is three moves: EMQ above 7.0 via CAPI, incrementality testing for real lift, and first-party attribution in BigQuery so Meta’s reporting surface isn’t your only source of truth.
- CAPI setups cost 17.8% less per result than pixel-only setups, and that gap widened after the window change.
Frequently Asked Questions
Meta permanently removed the 7-day view and 28-day view attribution windows from its Ads Insights API. Only 1-day engage-through attribution remains as the view-based category, which counts actual outbound link clicks instead of passive impressions, likes, or profile visits.
No. Reported conversions dropped 15-40% overnight with no change to campaigns, budgets, or targeting. That is a measurement artifact, not a performance collapse. Check whether actual sales in your WooCommerce admin dropped by the same percentage — they almost certainly did not.
Nearly. The old 1-day view and 28-day view windows are gone. The replacement, 1-day engage-through, only credits conversions preceded by an actual outbound link click. Likes, profile visits, and passive impressions no longer count toward attribution.
Raise Event Match Quality above 7.0 via Conversions API so click-based matches fire at full strength, run incrementality tests to measure real lift, and store raw WooCommerce conversion events in BigQuery so you have a conversion record independent of Meta’s reporting surface.
Extremely unlikely. The change aligns with platform-wide privacy direction across Google, Apple, and Meta, and the industry has moved toward clicks plus incrementality as the accepted measurement basis. Meta has been quietly reshaping its measurement surface every six to twelve months; planning on reversal is planning on the past.
Put your WooCommerce conversion data somewhere Meta can’t redefine it. Start at seresa.io/product.