Platform Watch: What the January 2026 View-Through Window Removal Really Cost You
On January 12, 2026, Meta flipped a switch inside the Ads Insights API. Two attribution windows, 7-day view and 28-day view, stopped returning data.
Not an error. Not a warning banner.
The fields just came back empty, and reported conversions across the industry dropped 15 to 40 percent overnight with zero change to campaigns, budgets, or creative.
Meta had actually given advertisers roughly three months of notice, publishing the change on its developer blog back on October 13, 2025.
Most advertisers never saw it. What they saw instead was a dashboard that looked broken on January 12, and what a lot of them did next is the actual subject of this post.
The window removal itself was a measurement change. What it cost some accounts was a decision problem, and that part is avoidable.
What actually disappeared
Before January 12, Meta credited a purchase to an ad under three separate windows: a 7-day click, a 7-day view (someone saw the ad, did not click, converted within a week), and a 28-day view (same idea, much longer runway).
The 7-day and 28-day view windows are gone. What is left is 7-day click and 1-day view. If your reported conversions used to lean on that longer view-through credit, and for many awareness and remarketing campaigns they did, that credit simply stopped showing up.
Meta also tightened how much historical detail its API will hand back.
That distinction matters more than it sounds like it should: any team still pulling detailed year-over-year comparisons through the API, not just the top-line number, lost access to the exact data they needed to prove the January drop was a definition change and not a decline.
The real cost: three ways this actually hit budgets
Panic-pausing campaigns that were still working.
This is the expensive one. A prospecting campaign’s reported conversions fall 25 percent week over week, and the reasonable-looking move is to pause it or cut its budget.
If that drop is the attribution change and not a performance change, the campaign gets killed while it is still profitable, spend shifts somewhere the broken baseline says looks better, and the account gets quietly worse for weeks before anyone notices, because the comparison baseline used to catch the mistake is the same broken baseline that caused it.
This pattern showed up repeatedly across high-spend Meta accounts through the first quarter of 2026, and it is the single costliest response an advertiser could have to a measurement change.
Automated rules and targets that kept firing on the old baseline.
Any budget rule, bid cap, or CPA target built on pre-January conversion volume did not know the volume had shifted for reporting reasons rather than performance ones.
Those rules kept making decisions against a number that no longer meant what it used to mean, which is a quieter but no less real cost than an outright pause.
A second change stacked on top before the first one settled. On March 3, 2026, Meta redefined click-through attribution to count only actual link clicks, moving likes, shares, saves, and comments into a new engage-through category.
Layer that on top of CPM increases of 15 to 40 percent that also landed around the same window, and accounts that had not yet adjusted to the January change were now trying to separate three things happening at once: a measurement artifact from January, a second measurement artifact from March, and a real cost increase in CPMs.
Teams that treated all three as one undifferentiated performance collapse made worse decisions than teams that separated them, because the correct fix for a real CPM increase (raise the CPA target) is the opposite of the correct fix for a reporting artifact (change nothing).
The one diagnostic that would have prevented most of this
Before touching a budget on any campaign whose numbers moved around January 12 or March 3, pull the same period from your own backend, Shopify, WooCommerce, GA4, or CRM, and compare it to what Meta reported.
A useful shortcut here is a health ratio: Meta-reported conversions divided by your independently tracked conversions for the same window.
A ratio landing roughly between 0.9x and 1.4x is normal drift between two measurement systems. A ratio well outside that band is a signal something in your tracking setup, not your campaign, needs attention.
If your own numbers were flat while Meta’s dropped, nothing about your advertising broke. The measurement did. That is not a reason to celebrate, but it is a reason to leave the budget alone.
What to actually do about it
- Never compare across January 12, 2026, without flagging it. Any report, dashboard, or automated rule that spans that date is comparing two different measurement systems. Split your reporting into pre-change and post-change periods rather than treating it as one continuous line.
- Run the health ratio check before cutting anything. Compare Meta-reported conversions to your backend numbers for the same window before pausing or reallocating budget based on a reported drop.
- Update every automated rule that was calibrated to the old baseline. Bid caps, CPA targets, and budget rules built on pre-January conversion volume will misfire against the new, lower baseline unless someone resets them deliberately.
- Export or archive anything you need past the new retention limits. Unique-count and hourly data now age out at 13 months, frequency data at 6. If you run true year-over-year analysis, that data needs to live somewhere Meta does not control.
- Build a measurement layer Meta cannot silently redefine. Independent, first-party attribution sitting alongside Meta Ads Manager gives you a number that does not move every time a platform updates its API, which is the actual fix for a problem that is going to happen again.
None of this requires abandoning Meta as a channel.
It requires treating every platform-reported number as something that can change definition without warning, since every time you trust a platform to measure itself, you accept the risk that the rules change without notice, and building a habit of checking your own backend before you act on someone else’s dashboard.
If you want to see what your account’s real numbers look like next to what Meta reported through January and March, you can book a live AdBeacon demo and bring your own account history.
FAQ
Did Meta warn advertisers before removing the 7-day and 28-day view windows?
Yes. Meta published the change on its developer blog on October 13, 2025, roughly three months before it took effect on January 12, 2026. Most advertisers did not see the announcement and only noticed when reported conversions dropped.
Should I have paused campaigns that looked worse after January 12, 2026?
No, in almost every case. Reported conversions dropped industry-wide with no change to targeting, creative, or spend. Pausing a campaign based on that drop alone risks cutting something that was still profitable.
How do I know if a reported drop is real or just the attribution change?
Compare the same period in your own backend, Shopify, WooCommerce, GA4, or CRM, against what Meta reported. If your independently tracked numbers stayed flat while Meta’s dropped, the change is in the measurement, not your campaign.
What is a healthy ratio between Meta-reported and independently tracked conversions?
Roughly 0.9x to 1.4x is normal variation between two measurement systems. A ratio well outside that range usually points to a tracking or event-matching issue worth investigating.
Can I still access the historical data Meta removed from the API?
Rolled-up aggregate totals remain available for up to 37 months, but detailed unique-count and hourly data are now capped at 13 months, and frequency data at 6 months. Anything you need beyond those windows should be exported and archived on your own systems.