How window settings change your results

What is an attribution window setting?

An attribution window setting is the click-through or view-through period, chosen in an ad platform or MMP, within which an ad interaction can still be credited with an install or event. Chapter 3, attribution windows, explained, covers what the windows are. This chapter is about what happens to your numbers when you pick one length rather than another.

Default attribution windows at a glance

Defaults are the starting point for most teams, so they shape most reports. These are the defaults confirmed on the platforms' own pages, plus TikTok's as reported:

• Google Ads app campaigns: installs and first opens within 30 days of the ad interaction; view-through fixed at 1 day; in-app conversions within 90 days.
• TikTok: 7 days after a click and 1 day after a view by default, as reported from TikTok's help centre.
• Apple SKAdNetwork: 30 days from a StoreKit-rendered ad to install, and 24 hours after a view-through ad.
• Apple AdAttributionKit: 30 days for clicks and 1 day for views by default, configurable from 1 to 30 days and 1 to 7 days.
• GA4: 30 days for first_open by default, with 7 days as the other option.
• Meta app installs: 1-day, 7-day or 28-day click settings exist, but the default for a new app install ad set is not confirmed on a live Meta page, so check each ad set.

MMP defaults vary by vendor and are usually set per network, so read them from your own MMP's settings rather than assuming.

What happens when you lengthen a click window?

A longer click window gives paid campaigns credit for installs that happened further from the click, so paid installs rise, organic falls, and cost per install drops in the report even though nobody new installed.

Take one install that happens 10 days after a tap on a Google ad:

• Under a 30-day click window, the install is credited to Google.
• Under a 7-day click window, Google gets nothing. If no other touchpoint qualifies, the install is organic.

Across a month of installs, that one rule moves a block of users between columns. The knock-on effects:

• CPI and CPA fall for the network, because the same spend is divided by more installs.
• ROAS rises, because the revenue of those extra users now sits in the campaign's cohort.
• Organic shrinks, and the organic share looks worse than it is. Organic vs paid installs explains why organic is the residual.
• Older clicks can win. With last-click rules, a long window lets an old click on one network compete with a recent click on another, so cross-network splits shift too.
• Click spam pays more. Fraud that fires large numbers of fake clicks benefits from long windows, because more of its clicks are still eligible when a real install happens. See mobile ad fraud and attribution.

Shortening the window reverses every one of these effects.

View-through windows move numbers fastest

View windows are short, so each extra hour or day is a large relative change. Lengthening a view window from 1 day to 7 days, which AdAttributionKit allows, means a week of impressions can now claim an install.

• Video-heavy and upper-funnel campaigns gain the most, because they produce many views and few taps.
• When a click and a view both qualify, the click usually wins, so view-through mostly takes credit from organic, not from other paid clicks.
• A view window set to off removes that credit entirely, which can make awareness campaigns look like they deliver nothing.

Attribution windows versus event and cohort windows

Three different time limits often get mixed up in one report:

Changing the attribution window changes who is in the cohort. Changing the cohort day changes how much of their revenue you read. A ROAS figure only means something with both stated. In-app events, postbacks and ROAS covers cohorts in full.

Re-attribution windows: an MMP worked example

Re-attribution settings decide whether a returning user's reinstall counts as a new attributed install. They have the same effect on totals as click windows: looser settings give campaigns more credit.

Linkrunner, for example, lists a configurable inactivity window with a default of 0 days and a re-attribution eligibility window with a default of 90 days. Other MMPs use different names and defaults for the same idea, and Google Ads counts Android reinstalls as installs, so totals differ before any click window is considered.

How to change a window without breaking your trends

1. Write down the current settings for every network and for the MMP, including view-through.
2. Change one network at a time, at the start of a reporting period, and annotate the date in every dashboard.
3. Never compare across the change without saying so. A drop in CPI the week a window grew is the window, not the campaign.
4. Know what locks. TikTok attribution settings reportedly cannot change after an ad group is published, and AdAttributionKit windows are set in the app's Info.plist, so they ship with an app release.
5. Match both sides when reconciling. Filter the network and the MMP to the same window, time zone and date basis before reading a gap.

For a longer method on picking lengths that fit how your users actually behave, Linkrunner has a guide to setting click and view window lengths. To see how window differences play out in a real Meta discrepancy, this Meta vs MMP install discrepancy explainer walks through the usual causes.