MeasureAttribution Basics
Know how an install gets its label before you trust the label.
Every install in your dashboard carries a campaign name, but a rule chose it. Attribution hands out credit; it is not a camera recording what happened. Treat it like a referee: useful, consistent, occasionally wrong.
How to use it
- Write down your rules: click and view lookback windows, which touch wins (usually the last eligible click), and how ties are broken.
- Draw one typical player's journey: ad seen, ad clicked, store visit, install. Mark which touch gets credit and which vanish.
- List the installs your setup cannot see: players who opted out of tracking, platforms that report only their own conversions, organic installs nudged by ads (see Organic vs Paid).
- Compare totals weekly: installs claimed by all sources added together versus installs the stores report. A gap above zero means overlapping claims.
- Agree which number decides, and label every report with its source.
Example
A hypothetical puzzle game sees two paid social platforms each claim 1,000 installs in a week, while your attribution provider credits them 600 and 550. Each platform counts its own touches; the provider de-duplicates. Neither is lying, they just use different rules, so you pick one for decisions and note the other.
Watch out
- Last-click credit undervalues the ads that created the interest. Pair it with an Incrementality Test before cutting upper-funnel channels.
- Rules shift as platforms change privacy mechanisms; read Privacy-Era Measurement and check current platform rules.