Against the 'whale' metaphor
Top-spending users aren't a species. They're a stage. The taxonomy is costing you product decisions.
The mobile games industry inherited the word whale from the casino floor, and we should give it back. It’s a metaphor that’s mostly fine when you’re operating a roulette table and mostly destructive when you’re shipping a free-to-play game.
The objection isn’t moral. It’s analytical. The whale framing implies that top spenders are a type of user — a species you can identify, target, retain, and forecast. They aren’t. They’re a stage that some users pass through, sometimes for a long time, sometimes for one weekend. The taxonomy is wrong, and wrong taxonomies produce wrong product decisions.
What the framing costs you
A team that thinks in whales builds:
-
Segmentation models that classify users by historical spend, not by current behavior. A user who hasn’t spent in six weeks is still in the “whale” segment, getting the whale treatment. Most of them aren’t coming back.
-
Retention features built for whale psychology — exclusivity, VIP tiers, status — when the actual question is whether they’re entering or exiting the high-spend stage, and the right intervention is completely different in each direction.
-
Pricing experiments that lock in the top tier of the SKU ladder because “whales love the $99 bundle,” when in fact the cohort that buys the $99 bundle is mostly users who passed through that stage two years ago and rebuilt their session habits around the SKU shape — not because $99 is the right number, but because that’s the SKU that was there when they were ready to buy.
Each of these is a real mistake I’ve seen. The whale framing made them feel correct in the moment.
A better model
Think of high spending as a state, not an identity. Users enter the state because of some combination of: time in product, social context, what their progression bottleneck looks like that week, and a half-dozen things outside your control. They leave the state for symmetric reasons.
The questions worth asking, in order:
- What’s the entry rate? Of users in week N of their lifetime, what fraction enter the high-spend state in any given week?
- What’s the duration? Once entered, how long does the state last, in weeks of continued high spend?
- What’s the exit rate, and where do exits go? Do they downgrade gradually? Churn outright? Re-enter later?
If you can answer those three questions, you have a model that tells you something useful. The whale framework — who spends — doesn’t, because who turns out to be the wrong question.
The substitution
Stop saying whales. Say users currently in the high-spend stage. It’s clunky on purpose. The clunkiness is doing work — it forces you to think about the cohort as a flow rather than a population, which is what they are.
Your monetization team will protest at first. Then, two months in, they’ll quietly drop the word and replace their slides with cohort transition diagrams, and your ARPDAU forecasts will get better.
There’s a deeper version of this argument — that almost every “type of user” framing in product is doing the same kind of damage — but that’s a longer post.