Meta Ads Attribution Window Change 2026: Why Conversions Dropped

The Meta Ads attribution window change in 2026 is why your reported conversions fell 15–30% even though spend, targeting, and creative never moved. Meta removed the 7-day-view and 28-day-view attribution windows in January, then shifted ad delivery to outcome-based optimization in March. Both changes shrank what gets counted. Before you cut a single budget, you need to know whether the drop is a measurement artifact or a real performance problem — because the fixes are opposite.

This post is the diagnostic. It tells you how to separate the two in about 20 minutes, and what to do once you know.

Why this matters

The dangerous move right now is the reasonable-looking one: a campaign's reported conversions drop 25% week over week, so you pause it or cut its budget.

If that drop is the attribution change, you just killed a campaign that was still working. Spend goes elsewhere, the "fix" looks like it helped because the comparison baseline is also broken, and the account quietly gets worse while every number on the dashboard says you made a smart call.

The opposite mistake is just as costly. If the drop is real — a fatigued creative, a broken pixel, a landing page regression — and you write it off as "just the attribution change," you let a genuine leak run for weeks.

There is no shortcut around the diagnosis. You have to tell artifact from reality before you touch the budget.

What actually changed (and when)

The first thing to get straight: "the 2026 attribution change" is three separate events stacked on top of each other, shipped months apart, all surfacing in your reports now.

| Change | When | What it did | Acts on your account? | |---|---|---|---| | 7-day-view and 28-day-view windows removed | January 12, 2026 | View-through conversions from days 2–28 stopped being counted in the Ads Insights API | Yes — already applied to all reporting | | Click-through redefined + engage-through created | March 2026 | Only link clicks count as click-through; likes, shares, saves, and engaged-views moved to a separate 1-day engage-through bucket | Yes — changes what each metric means | | Delivery shifted to outcome-based optimization | Early March 2026 | Meta's AI began optimizing for predicted downstream conversions instead of clicks; CPMs rose | Yes — affects real auction costs |

According to Supermetrics' change log for January 12, 2026, Meta removed the 7d_view and 28d_view attribution windows along with several combined and first-conversion variants — any request for those windows now returns empty data. Per DataSlayer's breakdown of the removal, some advertisers had 30–40% of their reported conversions coming from that 8-to-28-day view window that no longer counts. The supported defaults are now 1-day-view and 7-day-click.

The March layer is different in kind. Per Search Engine Land's reporting on the engage-through rollout, only link clicks now count as click-through, and every other interaction moved into engage-through on a 1-day window. The full mechanics of that split are in the engage-through attribution explainer — this post assumes it and focuses on the diagnosis.

The part most "why did my conversions drop" posts miss is the delivery change underneath all of it. Per Digital Applied's analysis of the March 2026 drop, Meta transitioned its AI delivery system from auction-based placement optimization to outcome-based optimization that predicts downstream conversions rather than clicks. The same report documents CPM increases of 15–40% across retail, lead-gen, and ecommerce in the first two weeks of March, and notes that campaigns generating fewer than 50 weekly optimization events lost significant algorithmic priority.

So you have two reporting changes (fewer conversions counted) sitting on top of one real change (higher CPMs). That combination is exactly why the drop is so easy to misread.

The diagnostic: artifact or real drop?

Run this before any budget decision. The goal is one verdict per campaign: measurement artifact, real regression, or both.

Signals it's a measurement artifact

If most of these are true, the conversions probably didn't disappear — the credit did.

• The drop is dated to January 12 or early March, not to a change you made. Line up the conversion line against those two dates.
• It hits view-heavy campaigns hardest — awareness, video, top-of-funnel, broad prospecting. These leaned on 7-day-view and 28-day-view for credit. Bottom-funnel, click-driven campaigns barely move.
• Spend, CTR, CPC, and frequency are flat, but reported conversions and ROAS fell. If the front of the funnel is unchanged and only the back-end count dropped, the measurement moved, not the performance.
• Your destination platform disagrees. GA4, Shopify, or your backend shows roughly the same orders or leads as before. The point of the March change was tighter alignment with downstream analytics, so the Meta–GA4 gap should narrow, not the real revenue.
• A new engage-through line appeared alongside click-through, absorbing interactions that used to be counted as clicks.

Signals it's a real performance drop

If these show up, something actually broke and the attribution change is cover for it.

• CPM is up but so is frequency, and CTR is falling — that's creative fatigue, not measurement. The creative refresh cadence framework for 2026 covers the trigger metrics.
• Your backend confirms the drop. Shopify or your CRM shows fewer real orders or leads, not just fewer Meta-attributed ones.
• The drop is sudden and campaign-specific, dated to a creative swap, an audience edit, or a pixel/CAPI change — not to January 12 or early March.
• Add-to-carts or landing-page views fell, which the attribution change does not touch. Those are upper-funnel events; if they dropped, demand or the page changed.

Putting it together

| What you see | Likely verdict | First move | |---|---|---| | Conversions down, backend flat, view-heavy campaigns hit | Artifact | Rebuild the baseline, don't cut budget | | Conversions down, backend also down, frequency rising | Real (fatigue) | Refresh creative, then re-measure | | Conversions down, backend down, dated to a pixel/audience edit | Real (tracking or targeting) | Audit the change you made | | CPM up, conversions flat in backend, ROAS down on paper | Artifact + real cost | Hold budget, account for higher CPM in targets |

The hardest cases are "both." Higher CPMs are a real cost increase, and the conversion undercount is an artifact — they stack, so a campaign can look catastrophic on the dashboard while the real damage is a manageable cost bump. Separating the two is the entire job.

What to do, in order

Once you have a verdict per campaign, act in tiers.

Act today

• Stop auto-pausing on reported conversions alone. Any rule that cuts budget on a week-over-week conversion drop is now firing on a measurement artifact. Pause those rules before they pause your campaigns.
• Pull the backend number. Open Shopify, your CRM, or GA4 and compare real orders/leads to the same period last month. This is the single fastest way to confirm whether anything actually changed.

This week

• Rebuild your comparison baseline from mid-January forward. Anything that straddles January 12 compares two different measurement systems. Per the Meta Business Help Center, the supported windows are now 1-day-view and 7-day-click — standardize every rolling report on a defined stack and document it.
• Reset conversion targets for higher CPMs. The 15–40% CPM rise is a real cost change, so a CPA that looks "up" may just reflect more expensive impressions. Move the target, don't kill the campaign.
• Separate engage-through from click-through on the scorecard. Don't add them — different windows, different meaning. The updated report template for 1-day view + engage-through has the column setup, and the fewer-clicks diagnostic walks the symptom on your own numbers.
• Brief stakeholders once. If a founder or CMO sees a 25% drop in the Monday report without context, the conversation becomes about Meta competence instead of account performance. Get ahead of it.

Monitor

• Watch the Meta-vs-backend gap as your signal. If it widens again later, that's a leading indicator of a tracking break — not another attribution change.
• Feed under-fed campaigns. Campaigns under 50 weekly optimization events lost delivery priority, per Digital Applied; consolidate audiences or budgets so optimization events clear that floor.
• Re-pull anything that optimized on 28-day-view. That history is gone from the API, so the signal Meta optimizes on today differs from what you budgeted against last year.

A worked example

Take a hypothetical DTC brand spending $80,000 a month on Meta, split across a broad prospecting campaign and a retargeting campaign.

In December, the prospecting campaign reported about 900 conversions a week on the old default stack, which still credited 7-day-view. After January 12, the same spend on the same ads reports closer to 650 — a roughly 28% drop, concentrated entirely in the campaign that depended on view-through credit. The retargeting campaign, which converts on clicks, barely moves.

Then March hits. CPMs across the account rise about 20%, so the prospecting campaign's reported CPA climbs further. On the dashboard, prospecting now looks like a disaster: conversions down, CPA up, ROAS halved.

But the brand's Shopify orders for the month are flat versus December. The backend says the customers still showed up. The "collapse" is two artifacts (the view-window removal and the engage-through reclassification) stacked on one real cost increase (higher CPMs). The correct response is to raise the CPA target to reflect CPMs and leave the budget alone — not to cut prospecting, which is what the dashboard was begging for.

This pattern — a measurement change masquerading as a performance cliff — shows up repeatedly across high-spend Meta accounts after January and March 2026. The account that cuts prospecting here loses the working top of its funnel and won't see the damage for weeks, because the broken baseline keeps validating the bad call.

Common mistakes

• Cutting budget on reported conversions without checking the backend. The dashboard number changed definitions twice this year. Your Shopify or CRM number didn't. Trust the one that didn't move.
• Comparing 2026 to 2025 without an asterisk. Numbers that span January 12 compare two measurement systems. The drop is partly definitional, and the comparison is meaningless without annotation.
• Blaming a real CPM increase on attribution, or an attribution artifact on CPMs. They're different problems with opposite fixes. The 15–40% CPM rise is a real cost change; the conversion undercount is a counting change. Separate them.
• Treating engage-through as a click-through equivalent. Different windows, different interactions. Merging them produces a number that represents nothing.
• Assuming the 28-day-view conversions can be recovered. Per Supermetrics, those windows return empty data now. There's no toggle to bring them back.

FAQ

Why did my Meta Ads conversions drop in 2026 with no other changes? Most likely the attribution window change. Meta removed the 7-day-view and 28-day-view windows on January 12, 2026, and redefined click-through in March, so conversions that used to count via view-through or social interactions no longer appear. Per DataSlayer, some advertisers had 30–40% of conversions in that lost view window. Reported numbers fell 15–30% with no change in actual performance.

How do I know if my conversion drop is real or just the attribution change? Compare your backend — Shopify, your CRM, or GA4 — to the same period last month. If real orders or leads are flat while only Meta-attributed conversions fell, it's a measurement artifact. If the backend also dropped, or frequency is rising while CTR falls, it's a real performance issue.

Did Meta really remove the 28-day attribution window? Yes. Per PPC Land and Supermetrics, Meta removed the 7-day-view and 28-day-view options from the Ads Insights API on January 12, 2026. The supported windows are now 1-day-view and 7-day-click.

Why did my CPMs go up at the same time? That's a separate, real change. Per Digital Applied, Meta shifted delivery to outcome-based optimization in early March 2026 and advertisers reported CPM increases of 15–40%. Higher CPMs are a genuine cost change, not a reporting artifact, so reset your CPA targets accordingly.

Should I change my campaign optimization because of this? Not reflexively. The attribution change is a measurement change, not an optimization instruction. Keep optimizing on click-based conversions for most performance campaigns, make sure each campaign clears roughly 50 weekly optimization events, and reset targets for higher CPMs rather than rebuilding your whole structure.

---

Stop guessing whether a drop is real. Good Morning reads the new attribution surface — engage-through, 7-day-click, 1-day-view — and hands you a pre-diagnosed weekly action list that already separates a measurement artifact from a real regression. Action items, not analysis: Act today / This week / Monitor, ranked before you open it. For a single read on an account before you start untangling it, the Meta Ads audit framework and a single account health score collapse the noise into something you can act on, and the May 2026 Meta Ads update tracks what else shifted this spring. $50/mo per account, 14-day free trial, read-only Meta access.