Glossary

What is an attribution window?

An attribution window is the period after an ad click or impression during which a subsequent conversion can still be credited to that interaction. Outside the window, the conversion is attributed elsewhere or not at all.

Attribution and measurement

All 220 terms

Click windows and view windows

There are normally two, and they are set separately because they represent different strengths of evidence.

A click-through window applies after a user actively clicked. Because a click is a deliberate act, the window is usually generous — 7 days is a common default, 30 days in considered-purchase categories.

A view-through window applies after an impression the user did not click — see view-through attribution. The evidence is far weaker, so the window is short: 24 hours is typical, and anything longer starts crediting coincidence.

How window length changes the numbers

Lengthening a window never reduces attributed conversions and almost always increases them, because more of the ambient conversion rate falls inside it. That makes window length a lever platforms can pull to look better, and a variable you must control before comparing anything.

Two rules follow. Never compare platforms on windows you have not verified are identical — a 7-day-click platform and a 30-day-click platform are not measuring the same thing. And never change a window mid-analysis without restating history, because the step change will read as performance.

  • Match the window to the real decision cycle. A 30-day window on an impulse purchase mostly captures noise.
  • Keep click and view windows separate, never blended into one number.
  • Record the window alongside every reported figure, so a comparison is possible later.

Windows on AI surfaces

Conversational surfaces stress the concept in two directions at once.

Some outcomes are immediate and inside the window by a wide margin — an agentic conversion completes within the same session, sometimes within seconds. Others are far outside it: a user reads a recommendation, remembers the brand, and acts a fortnight later through a channel with no connection to the exposure. There was no click to start a window, and the eventual conversion looks organic.

Extending the window does not fix the second case, because the problem is a missing interaction rather than an expired one. The instrument that catches it is a holdout, not a longer lookback.

Common questions

What is a sensible default attribution window?

7-day click and 1-day view is the common baseline. Move it only for a reason you can state — a genuinely longer purchase cycle, not because a longer window produces better-looking numbers.

Do attribution windows apply to SKAdNetwork?

Differently. SKAN uses its own fixed measurement periods and timer mechanics rather than a window you configure, which is part of why SKAN numbers cannot be compared directly with conventional attribution.

Where we write about attribution window

More in attribution and measurement

Deciding which touchpoint earned the outcome, and proving it.