There is one number in your ad account that changes your reported performance without changing a single sale. It is the lookback window, and most teams have never touched it.
PPC Land has published a deep explainer on the mechanic, and it is worth every marketer’s attention – because in 2026 the platforms are moving in opposite directions.
What a lookback window actually is
A lookback window is the stretch of time before a conversion during which an earlier click, view or impression stays eligible for credit. Set it to seven days and an ad seen eight days ago earns nothing. Set it to ninety and that same ad takes the sale.
The buyer does not change. Only the accounting does.
Google, unusually, keeps two terms separate. A conversion window runs forwards from the ad interaction and decides whether the conversion is recorded at all. A lookback window runs backwards from the conversion and decides which prior interactions compete for credit. Google’s own example: a 30-day lookback for a conversion logged on January 30 considers interactions from December 31 onward.
Where the numbers live
According to PPC Land, the defaults differ everywhere:
- Google Ads: click-through conversion windows run 1 to 90 days, with 30 days applied automatically to new Search and Display conversion actions. View-through defaults to a single day. The setting sits on the conversion action, so newsletter sign-ups can run 7 days while a financed purchase runs 90.
- Google Analytics: acquisition key events (first_open, first_visit) default to 30 days; everything else defaults to 90. That 90-day default also drives session attribution – which is why analysts keep “discovering” sessions credited to campaigns from weeks earlier and filing it as a bug.
- Campaign Manager 360: the API exposes clickDuration and postImpressionActivitiesDuration, each 0 to 90. Note the trap: 0 means a 24-hour window, not same-day, and setting impressions to 0 switches off view-through credit entirely.
- Meta: named combinations only – 1-day click, 7-day click, 28-day click, 1-day view, 1-day engaged view.
Why 2026 broke the comparability
The two biggest platforms are diverging. On August 11, 2026, Google Analytics scrapped its fixed three-day engaged-view window and made both video and click windows fully configurable integers (1-30 and 1-90 respectively).
Meta went the other way. On January 12, 2026 it removed 7-day and 28-day view-through windows from the Ads Insights API – gutting the measurement upper-funnel campaigns leaned on. Then on March 3, 2026 it limited click-through attribution to link clicks only for website and in-store conversions. That follows the earlier discovery that Meta had been counting likes, shares and saves as clicks, crediting conversions from users who never left the app.
Amazon tightened view attribution in January 2026. OpenAI shipped view-through conversions in ChatGPT Ads on a fixed one-day window, excluded from bidding and billing.
The part that actually costs you money
The window is not just a report filter. It is a training signal.
An account set to 7-day click teaches automated bidding to hunt for people who convert within seven days. It shapes delivery, it does not merely describe it. Change the window and you change who Google and Meta go looking for.
Two more truths worth pinning above your desk:
Lengthening a window increases reported conversions without increasing sales. Credit is redistributed, not created. That is precisely why incrementality testing exists.
Configured is not the same as achievable. Apple’s Intelligent Tracking Prevention capped script-written persistent cookies at seven days back in February 2019, and applied it across all iOS browsers by iOS 14. A 90-day window on that browser is an intention, not a measurement.
What to do this week
Audit, then standardise. Open every conversion action and check the window matches the real sales cycle – short for lead magnets, long for considered purchases. Document what each platform is set to, because comparing a 7-day Meta number to a 90-day GA4 number is comparing two different questions.
And be careful when you edit: most changes are not retroactive, so a June change splits your year into two definitions.
The confidence gap is real. Research published in October 2025 found 54.1% of marketers reported no year-on-year change in measurement confidence, and IAB research in February 2026 recorded up to 75% of buy-side decision-makers rating attribution, incrementality and mix modelling as underperforming.
Fixing your windows will not solve that alone. But it is the cheapest place to start.
Source: PPC Land
