If you opened Meta Ads Manager this week and saw your conversions crater without touching a single campaign setting, you're having a very normal 2026 experience. Frustrating, but normal.
In March, Meta redefined what "click-through attribution" actually means. Link clicks still count. But likes, saves, shares, and comments — which used to count — now get shunted into a separate bucket called engage-through attribution, with a 1-day window instead of 7. The result hit a lot of accounts as a silent performance drop nobody could explain. Budget got cut. Campaigns got paused. Most of it was unnecessary.
What actually changed
Before March 2026, Meta's click-through attribution was generous. If someone liked your ad on Monday and bought on your website on Wednesday, that sale counted as a 7-day click-through conversion. Technically they never clicked a link. Didn't matter.
Now it does. Only actual outbound link clicks count as click-through. Everything else — the likes, the saves, the shares — goes into engage-through attribution, which only looks back 1 day instead of 7.
The 7-day window still exists. It just applies to a much smaller set of interactions than it did before.
Why your numbers look different even if performance didn't change
Remarketing campaigns got hit hardest. They target warm audiences who already know your brand and are likely to engage before eventually clicking. That whole pattern — engage, come back later, convert — used to be captured in the 7-day window. Now most of it isn't.
Cold prospecting with link-click objectives? Less affected. Those campaigns were already optimising for actual clicks, so the reclassification changes the label more than the underlying behaviour.
The tricky part is that the conversions still happened in the real world. Meta just moved them to a column most advertisers have never opened.
How to check if this is actually your problem
Open Ads Manager. Pull March through May 2026 and compare it against the same window a year earlier. Look at your reported website purchases or leads. Then open your Shopify dashboard, your CRM, or your payment processor and look at actual orders for the same period.
If your backend revenue is flat or growing while Ads Manager shows a drop, you're looking at an attribution reclassification, not a performance problem. If both dropped together, something else is going on.
That gap widening is the tell.
Why Meta says it made sense
The argument is consistency. Most third-party analytics tools — Northbeam, Triple Whale, GA4 — only count actual link clicks as attribution events. Meta's old definition was more inclusive, which meant advertisers were constantly reconciling two different sets of numbers and usually concluding that Meta was overcounting.
Whether or not you buy the reasoning, the billing didn't change. You're not paying more. You're just seeing fewer conversions in the column you've been staring at for years, with some of them now sitting in a different one.
Three things to do right now
First, don't cut budget until you've checked your backend. Seriously. A lot of accounts made this mistake in March and spent April trying to recover performance that was never actually lost.
Second, add the engage-through attribution column to your reporting view. It's not there by default. Go to Columns, then Customise Columns, search "engage-through" and add it. You'll get a much clearer picture of what's actually happening.
Third, recalibrate your ROAS benchmarks. If you were hitting 3.5x and you're now seeing 2.8x, that's not necessarily a failure — it might just be a different counting method on the same sales. Figure out which it is before making any changes.
For remarketing specifically: the 1-day engage-through window means you need more direct-response touchpoints. Stronger offers. Clearer CTAs. More reasons to click the link rather than just react to the post.
One honest take
Meta's attribution was always a bit flattering to Meta. This change makes it less flattering and more accurate. That's probably better for you long-term, even if the transition is messy. The advertisers who'll get hurt are the ones who make decisions off Ads Manager reported ROAS without ever cross-referencing actual revenue. If that's you, this is a good time to fix that habit.
If you want to see how your Meta account is actually performing — attribution health, ROAS against industry benchmarks, structural issues that might be costing you more than reporting quirks — the Gromerce free audit takes about 3 minutes.
Your conversions didn't disappear. Meta moved them. Check your backend before doing anything else.
Related articles: Meta Just Gave Claude and ChatGPT Access to Your Ad Account. Here's What Actually Changes. · Meta Advantage+ Is Quietly Becoming the Default — Here's What It Means for Your ROAS · Meta's AI No Longer Needs Your Audience Targeting. Here's What It Needs Instead.
Sources: Meta for Business attribution announcement, Search Engine Land, ALM Corp, Dataslayer, May 2026
