Turn incrementality into better budget decisions | Adjust

Summary

Incrementality testing complements traditional mobile attribution by measuring the causal impact of marketing spend. While attribution shows where conversions are credited, incrementality estimates how many conversions would not have happened without the campaign. Adjust InSight uses a counterfactual model: comparing observed performance against a baseline of what would likely have occurred absent the marketing action.

The article's core argument is that such tests should be designed around the decision they inform. If you need to know whether a budget increase generated additional installs, target the install event; if you're measuring revenue impact, run a revenue-based experiment. The author warns against combining metrics from different test types—for example, a campaign reporting 125% attributed ROAS and a 30% incremental lift on installs cannot be used to infer incremental ROAS, because that 30% applies only to installs.

To calculate true incremental ROAS, divide incremental revenue by incremental ad spend, remembering that this figure excludes app store and payment processing fees, so it is not profitability. When reviewing InSight results, start with the incremental effect, then check statistical significance; a large lift that is not significant lacks enough evidence to justify budget changes. Next, examine the Organic result to identify organic cannibalization, where paid activity replaces organic demand—such as when increased spend boosts paid installs but organic installs decline.

Based on these signals, the article recommends increasing spend when incrementality is strong, statistically significant, delivers meaningful volume, and exceeds performance targets; maintaining spend when returns are stable; and reducing or reallocating spend when lift is low or cannibalization is present. However, these signals do not determine decisions—objectives and targets still define what makes a result actionable. The ultimate takeaway is that incrementality adds causal context to existing metrics, not a replacement, giving marketers stronger evidence for budget allocation.

Analyst Note

The timing of this article reflects a broader shift: as attribution limitations grow under privacy regulations, incrementality has moved from a specialized research technique to a core feature in marketing measurement platforms. What's notable here is the article's insistence on matching the test's target metric to the decision at hand—a discipline that prevents a common analytical error. The distinction between an install-level incremental lift and a revenue-level incremental ROAS matters because the two answer fundamentally different questions, and conflating them can distort budget recommendations.

The treatment of organic cannibalization is also significant: by estimating whether paid campaigns merely displace organic conversions, the analysis gives media buyers a more honest view of true incremental value. The key implication for UA teams is that statistical significance and effect size should be read together before adjusting spend, while counterfactual estimates become actionable only when interpreted through campaign targets. Worth watching is how such causal inference becomes a standard layer in attribution platforms, signaling a maturation of the AdTech measurement stack.

This is less about replacing traditional attribution and more about providing decision-ready evidence, which is increasingly essential as budgets tighten and accountability demands rise.