Your top-spender group shrinks before revenue does

Revenue is the number everyone watches, and it is one of the last things to move when a game gets into trouble.

The reason is arithmetic. Revenue is the number of paying players multiplied by what they spend. In most games those two move in opposite directions during a decline — the count falls while the average rises — and the product stays flat for months.

Why the average rises while the count falls

The players who leave first are not the biggest spenders. They are the marginal ones: players who bought occasionally, whose engagement was moderate, who were closest to the edge of caring.

When they go, they are removed from the denominator. The remaining paying population is more committed on average than it was, so the average spend per paying player goes up. Nothing about anyone's behaviour changed — the composition did.

Meanwhile the top spenders, who are the most invested, are the last to leave. They keep spending at their usual rate for a long time after the marginal players have gone.

So the reported picture is stable or improving revenue and rising average spend, which reads as a healthy game, while the paying population is quietly draining. This continues until the top spenders start leaving too, at which point revenue falls sharply and appears to have collapsed without warning.

The number that moves first

The count of players who spent anything in a period is the leading indicator, and it is rarely on the main dashboard because it is unglamorous.

It moves early because it captures the marginal population directly. A decline in the number of paying players, with revenue flat, is the signature of the composition shift described above, and it usually precedes the revenue drop by a considerable margin.

The related number is the count of new paying players — people making their first purchase. This one is even earlier, because it reflects whether the game is still converting anyone. A game can maintain revenue for a long time on existing spenders while converting nobody new, and by the time that shows up in revenue the conversion problem is old.

Both of these are counts rather than averages, and counts are much harder to flatter with composition effects.

Reading concentration deliberately

The other measurement worth having is what share of revenue comes from the top slice of spenders.

Rising concentration is the same story from a different angle. If a larger share of revenue is coming from fewer people, the game is becoming dependent on a shrinking group. That is fragile in a way flat revenue does not communicate — the variance goes up, and a small number of departures can move the total considerably.

Concentration also changes what the right decisions are. A game with broad-based revenue can afford to experiment with pricing and offers. A game whose revenue depends on a small group has to be much more careful, because the population that can be alienated is the population that matters.

Tracking concentration over time is more useful than its absolute level, since what counts as high varies by genre. The direction is the signal.

See how we track paying population alongside revenue →

What to do when the count is falling

The response depends on which end is failing, and the two counts distinguish it.

If new paying players are falling while existing ones hold, the conversion path is broken. Something changed in what new players experience, or acquisition is bringing in a different kind of player, or the early game stopped producing the moment where people decide to buy. This is fixable and the fix is upstream of the store.

If existing paying players are dropping out while new ones convert normally, the problem is later in the lifecycle. Players convert and then stop — usually because they ran out of things worth buying, or the game stopped rewarding continued investment. This is a content and progression problem wearing a monetisation costume.

If both are falling, the game is declining generally and the monetisation numbers are a symptom rather than the issue.

The reason to look at these before revenue moves is simply that there is more time to act. A conversion problem caught while revenue is still flat is a normal piece of product work. The same problem caught when revenue drops is an emergency, and it has already been running for months.

The reason these counts are absent from most dashboards is that they look uninteresting when the game is healthy — a flat line next to a revenue chart that moves. That is exactly the property that makes them useful. A metric that only moves when something is wrong is a better alarm than one that moves constantly for reasons that need explaining, and the counts have that character. Adding them costs one query and they earn their space the first time they diverge from revenue.

Talk to us about leading indicators for a live game →