A game that remembers a return begins, sooner or later, to value the reason for it.
A return can mean delight. It can mean curiosity.
It can also mean a timer finished.
Or a streak is about to break, or the last available action has become a purchasable absence.
The same tap produces all of those explanations with one identical piece of evidence.
The app opened again.
That evidence matters because mobile games are rarely paid for at the moment they’re found.
Discovery becomes an install. An install enters the first session. The first session becomes a return. A return may become a conversion.
The funnel isn’t a journey.
It’s a sequence of questions asked of the same person.
A studio buying discovery isn’t purchasing an audience in the abstract.
It’s purchasing the possibility that later behaviour will cover an earlier invoice.
The invoice arrived first.
The opening minutes carry that pressure quietly.
A first session compresses tutorial, early win and core loop into a few minutes.
The alternative is finding out how quickly an icon can be deleted.
Keep teaching. Keep rewarding. Keep making the game legible before the player becomes absent.
The retention curve then does something that ought to be discussed more often.
A healthy title keeps something like a quarter of its players the day after install.
A week later that figure has fallen to low single digits.
A month later most games sit under a few per cent.
A return is not a unit of health.
It’s a measurement whose meaning depends entirely on what follows it.
Day one can describe a successful introduction.
Day seven can describe a habit.
Day thirty can describe a game that still has a place in somebody’s routine.
Those are adjacent facts, not interchangeable ones, and the same word covers all three.
Retention design has an honest case, and it’s worth stating before the objections.
An ongoing game needs some way to tell a pleasant first encounter from a relationship.
A relationship that can support updates, servers, the next piece of content.
Daily calendars give consecutive logins escalating value.
Time-limited events make one particular return window matter.
Without a record of return, an open-ended game can’t know whether anyone is inhabiting the continuity it promised.
Energy systems are the most compact version of that bargain.
They cap available actions and refill on a timer.
The next session is ready before it’s wanted.
They can regulate pace.
They can stop one session becoming an exhausting blur.
They’re a real design tool with a real defence.
They also present an empty meter beside a refill priced in real money.
The timer doesn’t have to become a villain to acquire a price tag.
Commerce has always been alert to the emotional condition known as nearly done.
The business has its own case, and it’s arithmetic rather than ideology.
Lifetime value has to exceed the cost of acquiring a player within some payback window.
Otherwise buying discovery stops making sense.
Every purchase then meets a store commission before the studio sees anything.
None of that makes an individual prompt defensible.
It does explain why a return gets made measurable, segmentable and valuable enough to follow.
Title-level revenue can’t say which return paid for itself.
Two numbers separate the readings.
Average revenue per daily active user divides all revenue across everyone who showed up.
Payers and non-payers alike.
Average revenue per paying user divides it across payers only.
The gap between them can describe a disquieting kind of success.
The first figure holds steady while the second climbs and conversion falls.
Fewer people are paying more.
A stable line conceals a narrowing base, and the dashboard reports it as no change.
That narrowing isn’t imaginary.
In a typical free game most active players never purchase at all.
Among those who do, a small fraction of payers supplies the majority of the revenue.
Those are directional shapes rather than a universal character test.
A gacha role-playing game and an ad-supported puzzle game don’t distribute money the same way.
Two revenue tracks run alongside each other regardless.
Non-payers watch rewarded video. Payers buy.
A rewarded-video prompt hands out currency.
In the same gesture it introduces the economy a later purchase will accelerate.
The non-payer isn’t outside the business model.
The payer isn’t the only person whose attention is being counted.
This is where player safety enters.
Not as an accusation against every free game.
The term dark pattern was coined for deceptive commercial interfaces.
Manufactured urgency. Manufactured scarcity. Hidden costs. Nagging. Confirmshaming. Forced action.
It was built around transactions a person could complete and walk away from.
Games moved the question somewhere harder.
A dark game design pattern is a deliberate choice.
Made to work against a player’s interest, for somebody else’s benefit.
Publisher revenue, another player’s advantage, or an engagement number.
That definition doesn’t condemn a return loop for existing.
It asks whether the loop has turned a relationship into a series of priced disadvantages.
Free access changes the incentive more than it changes the interface.
A boxed sale has a fixed ceiling per copy.
An open-ended game earns through spend intensity across an unbounded period.
That makes the frequency and size of in-session purchase decisions consequential.
A boxed game never had to care about either.
Regulators have occasionally seen the machinery clearly.
Japan’s ban on complete gacha targeted a very specific compound structure.
Collect a full set of random items, and only then does the desirable reward unlock.
It didn’t ban randomized draws generally.
It identified an arrangement where scarcity and sunk cost grow together.
That precision matters.
A category broad enough to blame every randomized reward can no longer name what it fears.
Other boundaries have gone less well.
Probability disclosure has been mandated in more than one jurisdiction.
Studies of compliance keep finding disclosures that are technically present and practically unreadable.
A probability has learned to hide in plain sight.
Which is a more sophisticated failure than not publishing one.
Enforcement actions have followed, some of them very large.
Over children’s data, and over interfaces that led players into purchases they didn’t intend.
Those are dated examples rather than a settled map.
What they show is a line still unresolved.
Between a priced return and an engineered misstep, and calling the interface a game settles nothing.
Retention work gains a way to fund an ongoing service.
To tell passing interest from durable attachment, and to avoid designing blind.
It gives up the innocence of treating every return as uncomplicated affection.
The safety critique gains language for urgency, obscured price, sunk cost and pressure.
It gives up the comfort of treating every timer as identical evidence of harm.
A timer can pace an experience.
It can make a return meaningful, and sell relief from its own constraint.
All with the same code.
The measurement will not tell you which of the three it was doing.
What evidence of care can a game collect without making care itself the product?