A lookback window is the span of time after an ad engagement when a later conversion can be matched to that engagement. The engagement can be a click or a view. If the install or purchase happens inside this window, the credit goes to that ad according to your attribution rules.
Put simply, the lookback window draws a boundary around cause and effect. Inside the boundary, an install can be linked to the ad. Outside the boundary, it is not.
How Lookback Windows Work
- A person clicks or views an ad.
- A timer starts for that device or browser.
- If the person installs or converts while the timer is running, the system attributes the conversion to that touchpoint.
- If another eligible touchpoint happens during the window, your attribution model decides which one wins, usually by last eligible touch.
You can set different windows for clicks and for views. Click windows are often longer than view windows because a click shows stronger intent.
Deterministic vs Probabilistic Attribution
Deterministic
When a device identifier or a reliable login is present, the match is direct. If the conversion arrives during the active window, the credit is assigned with high confidence.
Probabilistic
When no stable identifier is available, the system may use statistical signals such as time, geography, device traits, and referrer patterns. These methods are strongest over short windows. Many teams cap probabilistic windows at about one day to keep accuracy high.
Common Window Types
Seven day click window
If someone clicks an ad and installs within seven days, the ad is eligible for credit unless another qualifying touchpoint occurs later within that window.
Short view window
View through attribution is usually shorter than click based attribution. Common ranges are from a few hours to one day, since a view expresses lower intent than a click.
Custom for in app events
Install windows and in app event windows often differ. You might track installs with a seven day click window while measuring purchases with a three day click window and a same day view window.
Choosing the Right Length
- Match your buying cycle Short consideration products such as casual games or delivery apps usually benefit from shorter windows. Longer consideration products such as education or finance may justify longer windows.
- Respect signal strength Keep view windows shorter than click windows. Retain the shortest windows for probabilistic methods.
- Use your data Study time to install, time to first purchase, and time to key milestones. Select windows that capture most true positives while limiting false matches.
- Test and refine Start with a baseline. Run controlled experiments with alternate windows and compare downstream quality metrics such as retention, repeat purchase, and refund rates.
- Align with platforms and privacy rules Some platforms enforce their own windows or consent prompts. Ensure your setup reflects those requirements and remains consistent with regional privacy laws.
Example Timeline
A person sees a video ad on Monday at 10.00
They click a search ad on Tuesday at 16.00
They install on Wednesday at 09.00
If your view window is 24 hours, the Monday view has expired by the time of install. If your click window is seven days, the Tuesday click is valid and receives credit.
Self Reporting Networks and Independent Attribution
Self reporting networks do not send every click to your attribution partner. Instead, your partner pings the network when it observes a conversion and provides a device or consent based identifier when allowed. The network then responds with any qualifying engagements that occurred within the active window. Your configured rules decide which touchpoint wins.
Why Lookback Windows Matter
Accurate performance measurement
Windows prevent very late conversions from being tied to the wrong campaign.
Budget efficiency
Clear windows let you invest in channels that create timely, measurable impact.
Customer insight
Time based patterns reveal how quickly different audiences move from ad exposure to action.
Pitfalls to Avoid
Windows that are too long
This inflates assisted credit and can pull credit away from organic or later campaigns.
Windows that are too short
This misses real influence and undercounts channels that drive delayed action.
Same window for all channels
High intent channels deserve longer windows than low intent channels.
Ignoring post install quality
Attribute with care, then validate with retention, revenue, and support signals.
Best Practices
- Separate click windows and view windows.
- Keep probabilistic windows short.
- Review time to event distributions each quarter.
- Use holdout tests to validate your chosen windows.
- Document rules so your teams and partners stay aligned.
FAQs
What is the difference between a lookback window and an attribution window
Lookback window usually refers to installs. Attribution window often refers to in app events such as purchases or subscriptions measured after install. Many teams keep separate settings for each.
Do windows reset after each touchpoint
Yes. Each new eligible touchpoint starts its own timer. Your model then decides which touchpoint wins if multiple timers are active when the conversion arrives.
Should view through attribution be enabled
Enable it when you run awareness formats and you can validate incremental impact with tests. Use a short view window so credit reflects lighter intent.
How do privacy changes affect windows
Consent prompts, platform policies, and privacy preserving APIs can restrict identifiers and reporting cadence. Use shorter and more conservative windows when signal quality is limited, and rely on geo or audience level experiments to estimate lift.
What metrics should I watch when changing windows
Track attributed installs, blended installs, cost per outcome, early retention, repeat purchase, and any measure of lifetime value. Avoid decisions based on attributed volume alone.
Can I use different windows per channel
Yes. Many teams set shorter windows for display views, medium windows for social clicks, and longer windows for search or branded clicks. Document these choices to keep reporting consistent.