Points, badges and the three-week bump

The gamification project often looks quite good in its first report.

Existing actions now award points. There are attractive badges, a leaderboard and perhaps a weekly email telling everybody who moved up. Usage rises. Three weeks later the curious people have collected the easy rewards, the keen people have established an unassailable lead, and everyone else returns to using the product exactly as before.

Eighteen months on, the whole thing disappears during a redesign.

The problem is not that somebody chose the wrong colour for the badges. The product remained unchanged and a scoring system was placed on top of it. Games are rather more involved than scorekeeping.

First ask whether engagement is the problem

Before adding a mechanic, look at the people who already use the product willingly. What are they doing? What job does it complete for them? When do they return, and what tends to happen just before they stop?

If engaged users obtain clear value while new users fail to reach it, progression and feedback may help. If even the most committed users tolerate the product only because an employer requires it, a better badge economy is unlikely to produce affection. That is a product problem wearing an engagement label.

Retention by cohort and behaviour is more useful here than one overall active-user number. The shape tells you whether people fail immediately, lose interest after completing something, or never form a repeatable habit. Those are different design problems.

What games actually do

A good game teaches the player. It introduces a manageable challenge, gives fast and legible feedback, then asks for a slightly more interesting use of the same skill. The player makes meaningful choices and understands enough of the result to improve.

It also packages effort into bounded units. A match, quest, run or level answers “What shall I do now?” and, just as importantly, “When may I stop?” Enterprise products often ask for one long, worthy behaviour with no satisfying edge to it.

Difficulty changes as competence grows. The first task is achievable without being insulting; later ones acknowledge that the user has learned something. A static checklist repeated forever is not progression, even if the fiftieth completion awards a gold icon.

Rewards work best when they recognise behaviour the product genuinely wants. Rewarding raw quantity can make people produce raw quantity. If the useful outcome is careful review, knowledge sharing or completing difficult cases, the measurement must survive contact with a user who would quite like to top the chart.

Mechanics are tools, with sharp edges

Streaks can support a regular habit. They can also make one missed day feel like the destruction of six months’ work. Consider grace periods, recovery and whether daily use is actually healthy for the product.

Levels and visible progress can split an intimidating body of work into manageable stages. They become silly when the levels represent no change in capability, access or challenge.

Social comparison can help where the group is meaningful and the gap feels bridgeable. A company-wide leaderboard usually tells most participants that they cannot win. Smaller peer groups, personal bests or cooperative goals may be much more useful.

Variable rewards are powerful, which is a reason for caution rather than excitement. If the product touches money, children, health, work allocation or other consequential decisions, the ethical and regulatory questions arrive early. “Games do it” is not a defence.

If rewards have value, you have an economy

Once points can buy something, they stop being decorative. They can accumulate, inflate, be exploited and become an accounting or regulatory concern.

Model how the currency enters and leaves the system. What happens when the keenest users have accumulated a year’s supply? Can balances be transferred? Can administrators create them? What happens if the programme ends or a reward is withdrawn? Who carries the liability?

The economy should be designed before launch, not after a spreadsheet reveals that the company owes everybody headphones.

Measure the whole chain

A successful mechanic changes a behaviour, and that behaviour improves an outcome the organisation cares about. Write the chain down:

mechanic → behaviour → business or user outcome

“People earned 80,000 points” confirms only that the counter works. Perhaps the aim was faster onboarding, better data quality, greater course completion or more regular exercise. Measure that outcome, along with defects and unwanted behaviour.

Use a control group where possible. Look at the distribution, not just the mean: a small group of enthusiasts can make the graph rise while the majority is unchanged. Keep measuring after the novelty period and across a few natural cycles of the product. A quarter is far more informative than launch week.

Then try to game the metric before your users do. If a cheap repetitive action produces the same reward as the valuable one, somebody will discover it, usually before the steering committee’s second meeting.

A modest place to begin

I would find the loop already used by the product’s most engaged users and make it clearer, quicker and easier to learn. Improve feedback. Give the work a visible beginning and end. Remove the part everybody has learned to work around.

Then test one small mechanic against a specific behaviour, with a control, and watch it long enough for novelty to fade. If it works, keep it. If it does not, remove it without commissioning a new badge family to rescue it.

There is nothing inherently wrong with points, badges or leaderboards. They are simply the most visible part of games and therefore the easiest part to copy. The useful machinery is usually underneath.