Meta Click-Through Attribution in 2026: Only Link Clicks Count Now

Your Meta account reported 1,000 click-through purchases in February and 740 in March. Spend held. Creative held. Actual revenue, the number in Shopify, didn't move at all. Nobody touched the campaigns.

That gap isn't a performance problem. It's a definition change. On 3 March 2026 Meta published Simplifying Ad Measurement for a Social-First World, and the headline of it is narrow and consequential: click-through attribution for website and in-store conversions now counts only link clicks. Everything that used to sneak into that bucket on the back of a like, a share or a save is still being counted, just somewhere else. If you read the click-through column the way you read it in January, you're reading a different metric and calling it the same name.

We run Meta next to Google and Amazon on the same profit and loss for brands doing $5M and up, which is the only vantage point from which this is obviously a reporting artifact rather than a collapse. If you want that read on your own numbers, we manage Meta and Google together with Amazon as one channel mix and reconcile the three against each other every month. Prefer to just ask? Book a call and we'll look at the before and after with you.

What Meta actually changed

Per Meta's own announcement, dated 3 March 2026:

• Click-through attribution for website and in-store conversions counts only link clicks.
• Conversions that follow a like, share, save or other non-link engagement moved out of click-through and into a separate bucket.
• What used to be called engaged-view attribution is now called engage-through attribution.
• The engaged-view threshold for video ads dropped from 10 seconds to 5.
• Meta's stated reason: its old click definition didn't match third-party tools, and the change should line its reporting up better with tools like Google Analytics.

Read that last one twice, because it's the part most commentary skipped. Meta didn't tighten the definition to be conservative. It tightened it to stop being the outlier. For years the complaint about Meta reporting was that it claimed conversions Google Analytics never saw, and a chunk of that discrepancy was exactly this: engagement-based conversions sitting in a column labelled clicks.

Signal that preceded the conversion	Before March 2026	After March 2026
Link click to your site	Click-through	Click-through
Like, share, save, comment	Click-through	Separate engagement bucket
Video watched past the threshold	Engaged-view, 10 seconds	Engage-through, 5 seconds
Impression only, no interaction	View-through	View-through

One caveat on timing, and we'd rather state it than paper over it. Meta's post carries a 3 March date and doesn't declare the rollout complete, and we haven't found a later help-centre article restating it. Treat the change as in flight rather than settled, and check your own account's reporting notes before you attribute a specific week's drop to it.

The January change that set this up

This didn't arrive out of nowhere. On 19 January 2026 Meta removed the 7-day-view and 28-day-view attribution windows outright. So brands had already lost the longest view-based windows, and then two months later the click column itself got narrower.

The sequence matters when you're reading a trend line. A year-over-year Meta comparison now spans two separate definition changes, which means almost nobody's 2026 versus 2025 chart is comparing like with like. If you're going to run that comparison anyway, and most finance teams will, the honest version carries an asterisk and names both dates.

What engage-through is actually worth

Renaming engaged-view to engage-through and halving the video threshold from 10 seconds to 5 sounds cosmetic. It isn't. A 5-second threshold catches a much larger population than a 10-second one, so the engage-through bucket gets bigger at the same time the click bucket gets smaller, and the two movements are easy to mistake for a single shift of real performance.

So what do you do with it? Treat engage-through as a mid-funnel read, not a sales number. Somebody who watched five seconds of a video and later bought is a genuine data point about whether the creative holds attention. They are not somebody who clicked your ad and converted, and a report that blends the two will overstate what your link-click creative is doing. We keep them on separate lines and we don't sum them into a headline figure, because the moment you do, you've rebuilt the exact ambiguity Meta just spent a release removing.

The practical test: if you cut every ad whose engage-through numbers look weak, would you be cutting your best link-click performers? On a few accounts, yes. Video that converts on a five-second watch is often different creative from video that drives clicks, and the brands that optimize to one metric tend to discover the other one was carrying the account.

Why did my Meta conversions drop?

If click-through conversions fell while spend, creative and actual revenue stayed flat, the most likely answer is reclassification, not performance. Three checks separate the two in about twenty minutes.

1. Compare the total, not the column. Add click-through, engage-through and the engagement bucket back together. If the combined number is roughly where it was, nothing was lost, it was sorted differently.
2. Check your actual revenue over the same dates. Shopify, your ERP, whatever holds the real number. A reporting change can't move it. A performance change can.
3. Look at when the line moved. A reclassification shows up as a step on or around a platform date. A genuine decline is usually a slope, and it usually shows up in upstream metrics like click-through rate or cost per click first.

If the combined number held and revenue held, you have your answer and you should stop there. If revenue actually fell, the attribution change is a distraction and the problem is somewhere else.

The expensive mistake this change invites

A brand running Meta in a silo sees reported conversions fall, concludes the channel broke, and cuts the budget. That's the predictable wrong move, and it's only identifiable as wrong if somebody is reading Meta next to the analytics number and next to the rest of the mix.

Across the accounts we run, the gap between attributed and actual is the normal state of things rather than an anomaly. One of our own case studies makes the point cleanly: a 109% sales increase in one month alongside a 190% increase in attributed sales. Those two numbers describe the same month. Attribution moved almost twice as far as the business did, because attribution credits the window and the business only counts the till. A brand that budgets off the first number and a brand that budgets off the second will make different decisions, and only one of them is reading reality.

Cutting a channel on a redefined column is how a profitable channel gets starved. We've seen the reverse too: brands that pour budget in because attributed numbers look spectacular, while blended efficiency quietly degrades. Both errors come from trusting one platform's self-report as though it were accounting.

What to change in how you report

Nothing about your campaigns needs to change because of this. Your reporting does.

• Stop reporting Meta's click-through number as the channel's result. Report it next to actual revenue for the same period, every month, in the same document.
• Name the window. Pick one attribution setting, write it on the report, and don't change it mid-year. Most arguments about whether Meta works are two people using different windows.
• Treat engage-through as its own line. It's a real signal and it's not a click. Blending them back together to protect a number is just the old problem with extra steps.
• Put a dated note on your 2026 charts. 19 January and 3 March. Future you will need it.

If you want the measurement question answered properly rather than reported around, lift testing is the instrument that doesn't care how a platform labels its columns. It's more work than reading a dashboard. It's also the only method that survives a definition change.

If you run Amazon too, you've seen this movie

Amazon has its own version of this problem, and it's worth understanding separately rather than by analogy, because the windows and the rules are genuinely different. We've written up how Amazon's attribution windows work and where they mislead, and the short version is that each platform grades its own homework on its own timetable.

That's the structural argument for running the channels together rather than as three vendors reporting separately. When Meta redefines a column, a brand with one operator across Meta, Google and Amazon finds out in a weekly review and adjusts a report. A brand with three agencies finds out when one of them defends a number.

What we'd do with your account this month

Pull February and April, add the three conversion buckets together, and lay them next to actual revenue. If the combined total held, write the note on the chart and move on. If it didn't, you have a real performance question and the attribution change just happened to be standing nearby when it showed up.

Meta made its reporting more honest and in doing so made a lot of dashboards look worse. The brands that lose money on this are the ones that act on the dashboard. If you'd like somebody to read the before and after against the rest of your mix before you touch a budget, that's a conversation worth having, and we'll tell you which of the two it is.