In early March 2026, advertisers opened Ads Manager and saw conversions down 20, 30, sometimes 40 percent. Same campaigns. Same budgets. Nothing touched. I got three messages that week that all said some version of "did Meta break?"
Meta didn't break. Meta changed what a click means.
| Click-through | Engage-through | View-through |
|---|---|---|
| An actual link click on the ad | Likes, comments, shares, saves | Ad shown, no interaction at all |
| Site, lead form, call, message | Short video views, per testing | 1-day window |
| 1-day or 7-day window | 1-day window only | Weakest evidence of influence |
| What most reports lean on | On by default for conversions | Report it as its own column |
What did Meta actually change in March 2026?
On March 3, 2026, Meta redefined click-through attribution to only count conversions after an actual link click — someone tapping through to your website, app, or lead form. Before this, "click-through" was really "any-click-through": a like, comment, share, or save counted as a click for attribution. Someone liked your ad Tuesday, bought from a Google search Wednesday? Meta could report that as a click-through conversion.
Those non-link interactions didn't vanish from reporting. Meta moved them into a new attribution category called engage-through. Here's how the buckets shake out now:
| Attribution type | What triggers it | Windows available |
|---|---|---|
| Click-through | An actual link click on the ad | 1-day or 7-day |
| Engage-through | Non-link interactions: likes, comments, shares, saves, and (per several practitioner writeups) short video views | 1-day only, on by default |
| View-through | Ad was shown, no interaction at all | 1-day |
Meta's stated reasoning was alignment: this brings its click-through definition closer to how third-party analytics tools like GA4 have always measured a "click" — a session that actually landed on your site. Worth saying plainly: this was a reporting change. Billing didn't change. Delivery didn't change. What changed is which column a conversion lands in.
One caveat: Meta's own documentation here has been thinner than the ecosystem coverage, and some details (like exact video-view thresholds for engagement) come from practitioner testing, not an official spec. If a detail matters for your account, verify it in your own Ads Manager.
- Same campaigns — Same budgets, same creative, nothing touched
- March 3, 2026 — Click-through now needs a real link click
- Likes and saves — Refiled into the engage-through bucket
- Reported drop — Practitioners saw 15-40% off click-through
- Backend revenue — Orders and CRM leads never moved
Why did my conversions drop if my sales didn't?
Because click-through attribution just got stricter, and that's what most Ads Manager reports lean on. Conversions that used to ride along on a like or a save now sit in the engage-through column — which many advertisers don't even have visible in their reports.
Practitioner reports after the change described click-through drops anywhere from 15 to 40 percent depending on the account, engagement-heavy accounts worse. The pattern that kept showing up: add click-through and engage-through together and the total lands close to the old "any click" number. The sales are still there. They got re-filed.
Who got hit hardest? Accounts whose ads generate lots of interaction but fewer link clicks — video-heavy creative, meme-style content, anything that racks up shares and saves. If your ads were engagement magnets, a big slice of your old "click-through" conversions were never link clicks to begin with. The new definition exposed that.
Check your own numbers before assuming anything. Pull up your Shopify or WooCommerce orders, your CRM, your Stripe dashboard for the weeks before and after early March. Revenue steady while Meta's reported conversions fell off a cliff? That's a definition change, not a performance problem.
- Annotate March 3, 2026 everywhere — Every dashboard and client deck gets the marker: attribution definition changed
- Decide on engage-through per ad set — It defaults to on for website and in-store conversion campaigns
- Split the reporting columns — Click-through, engage-through and view-through side by side, never blended
- Reset CPA and ROAS targets — Old targets were calibrated against any-click; recalculate from post-March data
- Verify Pixel and CAPI events — Confirm Purchase and Lead fire, deduplicate via event_id, check Event Match Quality
- Leave campaign structure alone — Delivery and the auction didn't change; restructuring resets learning for nothing
How do I re-baseline my Meta reporting after the change?
Treat March 3, 2026 as a hard line in your data. Everything after it is measured with different rules than everything before it. Here's the working checklist I've been running on client accounts:
- Annotate the date everywhere. Every dashboard, every Looker Studio report, every client deck gets a marker on March 3, 2026: "Meta attribution definition changed." Six months from now nobody will remember why the line dips.
- Check attribution settings per ad set. Engage-through defaults to on for website and in-store conversion campaigns. Decide deliberately whether you want it — for lead gen especially, a conversion "attributed" to someone who liked your ad but never visited your site is a stretch.
- Use the attribution comparison in Ads Manager. The compare attribution settings feature breaks conversions out by click-through, engage-through, and view-through side by side. This is where you see exactly where your "missing" conversions went.
- Rebuild your reporting columns. Add engage-through as its own column instead of letting it blend into one conversions number. Blended totals fool you twice — once when the number drops, again when you can't explain the recovery.
- Reset CPA and ROAS targets against the new baseline. Targets set against any-click attribution are now miscalibrated. Recalculate from post-March data only, or from your backend numbers (better — more on that below).
- Verify your pixel and Conversions API events in Events Manager. The change didn't touch event collection, but a drop in reported conversions is the moment people discover their tracking was already broken. Confirm Purchase and Lead events fire, deduplicate via event_id, and show healthy Event Match Quality. Don't let a real tracking problem hide behind a reporting change.
Should I restructure my campaigns because of this?
No. This is the most expensive mistake available right now, and plenty of advertisers are making it.
The delivery system didn't change. The auction didn't change. Your ads reach the same people and drive the same sales they did in February. Start duplicating ad sets, swapping optimization events, or slashing budgets because a reported number dropped, and you're resetting learning phases to solve a problem that exists only in a column definition.
The one decision genuinely on the table is the attribution setting itself: whether to keep engage-through on. That's a measurement choice, defensible either way depending on your funnel. Everything else — creative, budgets, audiences, structure — should be judged against backend revenue, not a metric whose definition changed mid-stream.
Why is first-party server-side tracking the real fix here?
Here's the deeper lesson from March 2026: platform-reported numbers are rules, and the rules change under you. Meta rewrote what a click means overnight. Before that, iOS 14.5 and ATT gutted the pixel. Before that, attribution windows shrank from 28 days to 7. Each change moved advertisers' numbers without moving a single sale. If Ads Manager is your source of truth, your source of truth has an editor you've never met.
The way out is owning your own measurement. That's what first-party server-side tracking is, in plain language: instead of relying only on Meta's pixel in the visitor's browser (where ad blockers, Safari's tracking prevention, and iOS privacy features eat it), your site sends conversion data from your own server, on your own subdomain, through a server-side Google Tag Manager container to Meta's Conversions API and every other platform you use. You collect the event once, on infrastructure you control, and you decide what gets shared.
For this specific mess, that buys you three things. First, a stable baseline: server-side Purchase events reconciled against actual orders don't re-file themselves when Meta renames a column, so you can immediately tell "definition change" from "real problem." Second, recovered signal: server-side events with strong customer parameters survive the ad blockers and browser limits that silently drop pixel events, which lifts your Event Match Quality — and better-matched data is what Meta's delivery system optimizes on, whatever the reporting bucket is called. Third, one source of truth feeding every platform: the same server event, deduplicated with event_id, goes to Meta, Google, TikTok — so your numbers agree with each other and with your backend, instead of every platform telling a different story under its own shifting rules.
Meta will change its reporting again. That's not cynicism, it's pattern recognition. The advertisers who shrugged off March 2026 were the ones whose real numbers already lived on their own infrastructure.
Frequently asked questions
Why did my Meta conversions drop in March 2026?
Meta redefined click-through attribution on March 3, 2026 to count only actual link clicks. Conversions previously attributed to likes, comments, shares, and saves moved to a new engage-through category, so the click-through column dropped even though actual sales didn't.
What is engage-through attribution on Meta?
Engage-through credits a conversion to someone who interacted with your ad without clicking the link — a like, comment, share, or save. It uses a 1-day window only and is on by default for website and in-store conversion campaigns.
Did Meta's 2026 attribution change affect billing or delivery?
No. It changed how conversions are categorized in reporting, not how ads are delivered or how you're charged. Only the numbers in the click-through column moved.
Should I turn off engage-through attribution?
Depends on your funnel. For lead gen, a conversion from someone who never visited your site is weak evidence — turning it off keeps reporting honest. For ecommerce with social-proof-heavy creative, engage-through captures real influence. Either way, report it as its own column.
How do I compare Meta performance before and after March 2026?
Mostly, don't — the two periods use different counting rules. If you must, add click-through and engage-through together post-March to approximate the old any-click definition, and annotate March 3, 2026 on every chart. Better: compare backend revenue, which never changed definitions.
Are my conversions actually gone after Meta's attribution change?
No. Check your store or CRM: if orders held steady while Meta's reported conversions dropped, the sales are still happening. They were reclassified into engage-through or fell outside the stricter definition.
Does server-side tracking fix Meta's attribution changes?
It doesn't change how Ads Manager categorizes conversions, but it gives you an independent first-party record of every conversion on your own infrastructure — the baseline that separates a reporting-rule change from a real problem — and it feeds Meta better-matched data through the Conversions API.
- meta ads
- attribution
- 2026 changes
WhatsApp