Creative measurement beyond CTR and IPM

Click-through rate and installs per mille are attention metrics. They tell a creative team whether a hook stopped a thumb and whether the ad's premise translated into an install. Neither says anything about what happens after the install, and for most mobile games the money is made or lost in the first week, not the first three seconds. A creative that posts installs per mille well above account average can still be a net negative once it is paired with cost per install and a cohort that churns before it pays back.

The fix most measurement teams reach for is a scorecard rather than a single number: installs per mille to confirm the hook works, cost per install to confirm the auction likes it, day-one retention to confirm the promise made in the ad matches the game, day-seven payback to confirm the cohort is worth what it cost, and a fatigue slope to say how long any of that holds. Each metric answers a different question, and none of them substitutes for the others.

Building the scorecard properly

Installs per mille and click-through rate arrive first and fastest, often within hours of a creative going live, which is exactly why they get overweighted. A concept can post a strong installs-per-mille figure because the hook promises something the build does not deliver, curiosity rather than fit. That shows up a day later as weak day-one retention: users who install because a scene looked dramatic then bounce once the actual mechanic appears. Reading installs per mille and cost per install without waiting for day-one retention is the single most common way a portfolio ends up scaling a concept that should have been killed.

Day-seven payback is the metric that actually matters to a studio's finance function, and it is also the slowest to arrive and the noisiest at low volume. A handful of high-value payers in a small cohort can make a mediocre creative look excellent, and a handful of early churners can make a good one look bad. Industry benchmark reports from measurement providers routinely note that day-seven revenue reads swing hardest on the smallest cohorts, which is the practical argument for not trusting a payback read until a creative has accumulated a meaningful number of converting users, not just impressions.

Fatigue slope is the metric creative teams treat as an afterthought and buyers treat as an emergency. Every creative decays: cost per install climbs as frequency rises and the same audience sees the same hook repeatedly. The useful version of this metric is not whether a creative is fatigued yet but the rate of climb once a threshold frequency is crossed, because a slow fatigue slope justifies a longer refresh cycle and a steep one tells a team to accelerate variant production before performance visibly breaks.

The minimum installs problem

The scorecard only works if a team is honest about when a read becomes reliable. Reading day-one retention off a few dozen installs, or day-seven payback off a cohort in the low hundreds, produces numbers that move more from noise than from the creative itself. The practical discipline, described in an earlier piece on this site, Cohort quality versus volume: a scorecard for weekly UA reviews, is to separate a directional early read from a decision-grade one and to say explicitly, in the same dashboard, which threshold a given number has cleared.

Most teams settle on a rule of thumb rather than a formal power calculation: a concept needs a few hundred installs before installs per mille and cost per install mean anything, and a converting cohort large enough to include a representative share of payers before a day-seven payback figure gets treated as a verdict rather than a hint. The exact threshold varies by genre and average revenue per user, which is precisely why it needs to be a stated rule inside a team rather than a judgement call made fresh for every concept.

A creative team that only ever looks at the fast metrics will keep discovering, a week too late, that the concept it scaled on Monday's numbers was actually a curiosity click dressed up as a winner. A creative team that waits for every metric to mature before making any decision will lose the speed advantage that makes rapid iteration valuable in the first place. The scorecard exists to let a team make a fast call on the metrics that are fast to trust, and hold its nerve on the ones that are not.

What the gate actually changes

The operational payoff of running installs per mille through cost per install, retention, payback and fatigue slope in sequence, with a minimum sample gate at each stage, is fewer creatives killed on a bad first day and fewer creatives scaled on a good first hour. It also changes how a weekly creative review reads: instead of a single ranked list by cost per install, a scorecard produces a shortlist of concepts that have cleared each gate in turn, with the ones still waiting on a slower metric clearly marked as provisional rather than silently ranked alongside the concepts that have already proven themselves on every measure that matters.

Related archive reading

These articles provide related context and remain subject to their stated review status.

• UA metrics explained: CPI, ROAS, LTV and payback
• Creative fatigue detection metrics for your dashboard