Attribution & Measurement

Incrementality

Also known asIncrementality TestingCausal LiftHoldout Testing

The causal lift a campaign produces — installs / conversions that happened BECAUSE of the campaign, not just installs the campaign happened to be present for. Measured via holdout experiments.

Key takeaways

  1. 01Attribution credits observed installs. Incrementality measures CAUSAL lift — how many installs happened because of the campaign that wouldn't have otherwise.
  2. 02Standard mechanism: holdout test. Control group sees no ads, treatment group sees the campaign, lift in conversions = incremental impact.
  3. 03Typical incrementality test reveals that 30-60% of "attributed" installs were actually organic — they'd have happened anyway.
  4. 04Post-ATT, incrementality is the most rigorous way to evaluate UA — cross-network MTA degraded, but holdout testing still works.

Attribution and incrementality answer different questions. Attribution asks "which campaign should get credit for this observed install?" — a credit-assignment problem. Incrementality asks "how many of these installs happened BECAUSE of the campaign, that wouldn't have happened otherwise?" — a causal-impact problem. Attribution can give credit to a campaign that contributed nothing to the install (the user would have installed anyway from organic). Incrementality measures only the actual causal lift.

The standard mechanism is a holdout test

  1. Define a target audience (e.g., users in a specific country and demographic).
  2. Randomly split into control group (sees no ads from this campaign) and treatment group (sees the campaign).
  3. Run for a meaningful period (typically 2-6 weeks).
  4. Measure install / conversion rates in both groups.
  5. Incremental conversions = treatment conversions − control conversions.

The key constraint: the control group must genuinely not see the ads. Easy in walled gardens (Meta, TikTok, Google) where the platform can guarantee exclusion. Hard for cross-network programmatic where you can't easily exclude users.

The brutal truth most incrementality tests reveal: a substantial fraction of "attributed" installs are not actually incremental. The user would have installed anyway — from organic search, App Store featuring, word-of-mouth, brand awareness — and the campaign just happened to be present in the journey. Typical incrementality tests find 30-60% of attributed installs are organic. For mature brand-driven apps the fraction can be even higher.

Types of incrementality tests

When incrementality matters most

When incrementality is overkill: small-budget tactical decisions, individual creative tests, daily bid adjustments. Stick with attribution for those.

Incrementality test types compared

Test typeHow it worksPrecisionWhere it works
Geo holdoutHold out entire markets; compare to live onesLower — geo-level confoundsAny channel
User-level (PSA / ghost bids)Control sees a public-service ad; treatment sees the real campaignHighest — same exposure, randomizedMeta, Google
Conversion-lift studyPlatform randomizes users internally and reports liftHighWalled gardens
Cross-channelUser-level holdout spanning networksHard to achieve post-ATTLimited on iOS

User-level holdouts give the cleanest causal estimate; geo tests are easiest to run but least precise. Post-ATT, cross-channel incrementality is the hardest to execute because user-level matching across networks broke.

Quick answers

What is incrementality in mobile UA?

**Incrementality** measures the causal lift a campaign produces — installs and conversions that happened BECAUSE of the campaign, not just installs the campaign happened to be present for. Different from attribution, which credits observed installs based on touch rules. Incrementality is measured via holdout experiments: a control group sees no campaign ads, treatment sees them, the lift between groups is the incremental impact.

How is incrementality different from attribution?

**Attribution** credits observed installs — "this ad was the last touch, so it gets the install." Can give credit to campaigns that contributed nothing causally (user would have installed anyway). **Incrementality** measures actual causal impact — how many installs happened BECAUSE of the campaign. More rigorous, requires holdout-group infrastructure, slower / more expensive to run. Use attribution for tactical day-to-day; use incrementality for big strategic decisions.

How do I run an incrementality test?

(1) Define a target audience. (2) Randomly split into control (no ads from this campaign) and treatment (sees the campaign). (3) Run for 2-6 weeks. (4) Measure install / conversion rates in both groups. (5) Incremental conversions = treatment − control. Cleanest implementation: Meta's Conversion Lift, Google's Brand Lift, or platform-supported PSA / ghost-bid mechanisms that handle user-level randomization for you.

What percentage of "attributed" installs are typically incremental?

40-70%. Most incrementality tests find 30-60% of attributed installs would have happened anyway — they're organic conversions the campaign just happened to be present for. Retargeting and re-engagement campaigns often have the lowest incrementality (high organic-claim rate). Cold-acquisition campaigns to non-aware audiences have the highest. The exact fraction varies wildly by campaign, brand maturity, and channel.

Back to glossary