Meta now counts only link clicks as click-through, so re-baseline your CPA history
Meta now counts only link clicks as click-through conversions and shortened the reporting windows it exposes. How to re-baseline targets and reconcile Meta with your measurement partner.
Two Meta attribution changes landed within three months of each other this year, and together they make any Meta trend line that crosses January or March unreliable. If you set cost targets, judge creative tests or report to a finance lead from Meta's numbers, you need to mark those dates and re-baseline before the next planning cycle.
The two changes
January 12 in the Ads Insights API. Meta's developer announcement said the 7-day view and 28-day view windows would no longer be returned through the attribution windows parameter. The 1-day click, 7-day click, 28-day click, 1-day engaged view and 1-day view windows remain, along with defaults based on each advertiser's attribution settings. The same announcement limited historical data on some breakdowns. If you pull Meta data into a warehouse or dashboard, any chart built on a view window that no longer exists now has a gap.
March 3 in Ads Manager. Meta said click-through attribution would count only link clicks for website and in-store conversions. Likes, reactions, comments, shares and saves no longer qualify. Those conversions move to a category called engage-through, which replaces engaged-view, applies to all formats rather than only video, and keeps a 1-day window. Meta said billing is unchanged, and that the changes would roll out in stages from later in March, so the timing differs by account.
Two details deserve attention. First, a conversion that happened two to seven days after a non-link interaction used to count as click-through and, under the new definition, no longer counts at all. Second, the reported scope is website and in-store conversions. The coverage we reviewed does not say whether app conversions are treated the same way, and Meta's definition of a link click, as described by an independent Meta ads analyst, includes app store links and deep links. Do not assume your app campaigns are exempt or affected until you have compared your own before and after numbers.
Why a small definition change can move big numbers
The effect depends on how much of your volume came from people who interacted with an ad without clicking the link. Accounts with heavy retargeting or very engaging creative may see more of a shift. Accounts that run mostly direct-response ads where people tap the link may see little. Nobody can tell you your number in advance, and Meta's own guidance is that you may begin to notice changes in reporting.
The second-order effect is on optimization. Meta's delivery uses reported conversions as training signal, so a change in what is counted can change what the system learns from. Whether Meta changes anything in its optimization logic has not been spelled out in the coverage we reviewed, so watch delivery stability rather than assuming either way.
A re-baselining routine you can finish in a week
- Freeze the comparison. Pick a campaign set with stable budgets and creatives. Compare the four weeks before March 3 with the four weeks after, but remember the staged rollout means your account may have switched on a different day.
- Use Meta's own comparison tools. The compare attribution settings view and the breakdown by attribution setting show how conversions split across windows and settings. They will not give you a perfect restatement of history, because past conversions cannot be reliably re-sorted, but they show the direction.
- Split reported conversions into click-through and engage-through. If engage-through is a large share, your click-only numbers will look worse than before, even if the underlying results haven't changed.
- Reconcile against a source Meta does not control. Compare installs and in-app purchases by day against your measurement partner or backend. If Meta's reported number fell and your partner's total did not, the change is definitional. If both fell, look elsewhere.
- Reset targets once, with a note. If you hold cost-per-result caps or ROAS floors, adjust them to the new baseline. Moving them gradually over several weeks makes it impossible to tell what changed.
Choose one primary window and stick to it
Having several windows available invites cherry-picking. Pick one for decisions, for example 7-day click plus 1-day view, or a click-only view if you want the most conservative reading, and use the others for diagnosis. Write the choice in your reporting standards and apply it to every campaign, so comparisons between teams and months stay valid.
For a subscription app, the window question interacts with trial length. If most of your revenue lands several days after install, then a 1-day window shows only the first part of the story, and the better comparison is revenue or trial-start rate per install from your measurement partner over a fixed cohort age.
What to tell stakeholders
Say it plainly: reported Meta conversions changed because of definition changes in January and March, not necessarily because performance changed. Show the dates on the chart. Offer the reconciled view against a source outside Meta. Give a range rather than a single restated number, because historical conversions cannot be reliably re-sorted under the new definitions.
The takeaway: annotate January 12 and March 3 everywhere, re-baseline once using a partner-reconciled comparison, and keep decisions on one stated window.