On this pageJump to a section
If a Meta dashboard you rely on went strangely quiet this week, you may not have a traffic problem. You may have a window problem.
On January 12, 2026, Meta stopped returning two attribution windows from its Ads Insights API. Nothing crashed. Reports that depended on those windows did not show an error. They just showed less. (I laid out the warning in Meta Is Removing View Attribution Windows.) This is exactly the kind of change that produces a nervous message from a founder before anyone has opened the actual account.
What exactly did Meta remove?
Two view-based windows, in one place.
Meta's developer blog says the 7-day view (7d_view) and 28-day view (28d_view) windows in the action_attribution_windows parameter "will no longer be returned" from January 12, 2026. What stays is every click window (1, 7 and 28 days), along with 1-day engaged view and 1-day view.1 Meta frames it as a review of the metrics it offers, and says the change matches Ads Manager behavior.1
It is an API change first
That last line matters. The change applies to the Ads Insights API, which is what connectors, spreadsheets and dashboard tools pull from. If you only read results inside Ads Manager, the change may be less visible to you than to the person who built your reporting stack.
Retention changed too
Alongside the windows, Meta set new limits on history. Unique-count fields and hourly breakdowns are limited to 13 months, frequency breakdowns to six months, and marketing mix modeling breakdowns run only through asynchronous jobs.1 Total values are unaffected and remain available for up to 37 months.1
13 months
Why did dashboards go quiet instead of breaking?
Because an empty answer is still an answer.
PPC Land reported that requests using the removed windows return empty data sets rather than error messages, so failures in a reporting system can go unnoticed.2 A tool that asks for 7d_view gets back nothing, treats nothing as zero and draws a flat line. Nobody sees an error. They see a smaller number.
Where to look
Open the tools you actually use and ask three questions. Which attribution windows does each report request? Which of those windows no longer exist? Is the number I see on screen the sum of the windows I think it is? If you cannot answer the third question, do not trust the chart until you can.
What does 1-day view tell you, and what does it not?
Less than 7-day view did, and more than people give it credit for.
A view-based window credits a conversion to an ad that someone saw but did not click, if they converted within the window. A 1-day view only counts conversions that happen within one day of the view. That is a tighter claim: it asks whether the ad plausibly caused a quick decision, not whether a person who saw it a week ago might have remembered it.
Why the tighter window is healthier
A long view window can claim a lot of credit for people who would have bought anyway. Seer Interactive's test of Meta's incremental attribution setting is a useful reminder. Across six accounts and about $1.05 million in spend, Meta reported an average of 87% of conversions as incremental, versus 67% when measured with GA4, and Seer advised treating Meta's figure as directional and cross-checking it.3 Shorter windows give platform reports less room to flatter themselves. The measurement logic is in Incremental Attribution.
Barry Hott, who says he has advertised on Meta since 2008, put the incentive problem plainly in a video posted on May 2, 2025:
The system is just trying to get you as many conversions as it can.
In his view, a system rewarded for conversions will happily claim ones that were coming anyway. That is one more reason to lean on cash, not on any single window.
How do you re-baseline your reports?
Compare like with like, and write down the seam.
- Choose one attribution setup that exists before and after January 12, such as 7-day click plus 1-day view.
- Pull the previous 30 and 90 days on that setup and save them as your new baseline.
- Mark January 12 on every chart, so no one reads a step change as a performance change.
- Keep old exports labeled "legacy windows" and never add them to new data.
- Update any custom API call or connector that requests
7d_viewor28d_view. - Re-check cost per result on the new baseline for two weeks before changing budgets.
Decide on money, not on windows
For acquisition decisions, line up what Meta reports with your own records: orders, booked calls, cash collected. When the platform's story and your bank balance diverge, the bank balance wins. Windows are accounting conventions. Revenue is not.
How do you explain this to a founder?
Short, calm and in three sentences.
| What changed | What did not |
|---|---|
| Meta no longer returns 7-day and 28-day view-based conversions through its API | Your ads, your spend and your customers |
| Some reports now show fewer reported conversions | The cash you collected and the leads you booked |
| We are re-baselining on windows that exist on both sides | Our judgment: we decide on cost per sale, not on a window |
Then say it out loud: "A reporting definition changed on January 12. Our reports now use windows that still exist. We are measuring against bank deposits and booked sales so we are not guessing." That is the whole speech. Anything longer sounds like an excuse.
What I would not say
Do not say performance is unchanged until you have checked. Do not say performance dropped until you have compared like for like. Say you are checking, and by when.
When a number changes, the first question is not "why did it fall." It is "did the ruler change."
What to do this week
Audit every dashboard, sheet and connector that pulls Meta data (the year-end account audit has the checklist), and fix the ones that request the removed windows. Re-baseline on windows that remain (the Q1 2026 prep covers January budgets), mark the seam and give your founder or client the three-sentence version before they ask for it.
This is a measurement change, not a verdict on your advertising. The brands that handle it well will be the ones who keep one honest source of truth and stop asking a single platform to grade its own homework. Keeping that source of truth is built into the systems I build for clients.
Keep building.Ekai
Questions people ask
What did Meta remove from the Ads Insights API on January 12, 2026?
Meta stopped returning the 7-day view (7d_view) and 28-day view (28d_view) attribution windows. The three click windows remain, plus 1-day engaged view and 1-day view.
Did my ads get worse because reported conversions dropped?
Not necessarily. If a report relied on the removed view windows, those conversions simply stopped being returned through the API. Compare results on the same windows before and after, not totals across the boundary.
Will broken reports show an error?
Trade coverage says asking for a removed window returns nothing at all, not an error, so a dashboard can look quiet instead of broken. Check your connectors and exports directly.
How far back does the Insights API keep data now?
Meta says unique-count fields and hourly breakdowns are limited to 13 months of history, frequency breakdowns to 6 months, while total values remain available for up to 37 months.
Sources
- 1
- 2
- 3
- 4
