Challenge

With a sophisticated app marketing team, the brand was investing in both last-touch attribution (LTA) and marketing mix modeling (MMM) for a more holistic perspective of channel efficiency. These distinct measurement disciplines surfaced conflicting signals, forcing the team to evolve their approach to performance analysis and media mix optimization.

In one instance, the brand’s LTA data pointed to Google as a dominant driver of their paramount key performance indicator (KPI): 7-day first-time conversions on Android. Google appeared to shoulder an outsized share of conversions, justifying a significant slice of the media budget. However, when the team cross-referenced this view with their MMM data, the story looked quite different. MMM showed that Google’s incremental contribution was far more modest than LTA reporting alone suggested—and that upper-funnel channels were playing a pivotal, albeit uncredited, role in driving users to that final touchpoint prior to conversion.

While last-touch attribution was correctly answering What was the last touchpoint before conversion?, it was never designed to account for full-funnel, cross-channel dynamics shaping a consumer’s path to purchase.

The team faced a strategic dilemma, as the marketers managing distinct lower-funnel vs. upper-funnel budgets debated over which data set should guide future spend decisions. The stakes were high: Continuing potential overinvestment in a channel based solely on LTA reporting could mean leaving real incremental growth on the table, underfunding channels that actually moved the needle.

Solution

Rather than choosing between LTA and MMM, the app marketing team—working closely with Kochava—reframed the objective. The goal wasn’t to pick “the right” measurement methodology. Each approach is designed to answer a different question and serve a unique purpose:

When MMM and LTA produce differing outputs, a third methodology can be engaged as referee and validator:

Kochava assisted the app marketing team in activating an incremental pulse test.

Pulse Test Results

With MMM signaling that Google’s true impact was significantly lower than what LTA reported, the brand worked with Kochava to design and activate a pulse test. Google campaigns were cycled on and off at predefined periods while the team closely monitored 7-day first-time conversion events across the full Android user base.

• LTA-predicted conversion drop: ~2,500 first-time conversions
• Actual conversion drop observed: ~500 first-time conversions
• LTA overcredit factor: 5X vs. actual impact
• MMM prediction: Validated—MMM aligned closely with what actually happened

The results of the incremental pulse test were definitive: LTA was indeed overcrediting Google. Much of the demand Google captured at the bottom of the funnel was already created by upper-funnel media, specifically TV and other brand-building channels.

Armed with empirical validation, the team restructured their channel mix. Upper-funnel media received the investment recognition it deserved. Google’s budget was right-sized to reflect its true incremental role. Going forward, the team orchestrated investment decisions based on their MMM Insights dashboard in Kochava, while leaning on LTA data for rapid, campaign-level feedback and in-flight creative iteration.