Reward the process, not the result

Here is the arrangement almost everyone builds by default. Lose the ten pounds, buy the jacket. Land the deal, open the good bottle. Finish the book, take the weekend.

It feels obviously correct. It is also, for a habit you are still building, close to the worst structure available, and it took me a long time and a lot of unclaimed jackets to work out why.

A result is not a behaviour

Think about what actually stands between you and the outcome.

You wanted to start more conversations this week. You started eleven. Four went nowhere because the other person was in a hurry, one because you caught someone on a genuinely terrible day, and two turned into something. Did you earn the reward?

Under an outcome rule, you earned it if the two conversions were the bar and you got there. Under the same rule, you earned nothing if the week had handed you slightly worse luck with identical behaviour.

Reward the result and you are paying yourself, partly, for the weather.

That would be tolerable if the reward were only a treat. It is not. A reward is a signal about what matters, and you are the one receiving it. Pay out on results and you teach yourself to want results, which you already did. What you have not learned is the thing that produces them.

There is a second problem, quieter and worse. An outcome you do not control is an outcome you can fail at while doing everything right. Do that a few times and the sensible response is to stop trying, because the effort is not tracking the payoff. People are good at noticing when a system is not paying them fairly, including when the system is their own.

What the evidence says about rewarding the how

The distinction between rewarding what someone did and rewarding how it turned out is not a self-help idea. It has been tested for decades, mostly in classrooms and on sports fields. The evidence is stronger on how people behave and how they feel than on what they eventually score, which is worth knowing before you lean on it.

Mueller & Dweck (1998), Praise for intelligence can undermine children’s motivation and performance, Journal of Personality and Social Psychology 75(1), 33–52. Across six studies, children were praised after succeeding either for being clever (“you must be smart at these problems”) or for working hard. Asked to choose what to do next, 88% of the effort-praised children picked problems they would learn from, while 69% of the ability-praised children picked problems that would make them look good. After a later failure the effort-praised children solved more problems and enjoyed the task more; the ability-praised children did worse than they had at the start.

The praise is the reward in that study, and the finding is about where you point it. Point it at the person and it becomes something to protect. Point it at what the person did and it becomes something to repeat.

Zimmerman & Kitsantas (1997), Developmental phases in self-regulation: Shifting from process goals to outcome goals, Journal of Educational Psychology 89(1), 29–36. Ninety high-school girls learned to throw darts. Those given goals about the method beat those given goals about the score, on skill, on self-efficacy and on interest in the game. The group that did best of all began on process goals and moved to outcome goals later in practice.

That last sentence is the honest version of this whole argument, and I want to be straight about it. Process first is not a rule for ever. It is a rule for while you are still learning the thing.

Kingston & Hardy (1997), Effects of different types of goals on processes that support performance, The Sport Psychologist 11(3), 277–293. Thirty-seven club golfers trained for 54 weeks on process goals, on performance goals, or on no goals. The process-goal group came out higher on self-efficacy, on control of cognitive anxiety and on concentration. Their playing performance did not separate reliably from the other groups, which is worth saying plainly: the gain here is in how the players held themselves, not on the scorecard.

The awkward part: rewards can backfire

I am not going to skip this, because any honest page about rewards has to deal with it.

Deci, Koestner & Ryan (1999), A meta-analytic review of experiments examining the effects of extrinsic rewards on intrinsic motivation, Psychological Bulletin 125(6), 627–668. Across 128 experiments, tangible rewards reliably reduced intrinsic motivation for interesting tasks. Expected rewards did the damage; unexpected ones had no measurable effect (d = 0.01). Rewards paid for simply engaging with a task, or for finishing it, showed the largest undermining (d = −0.40 and −0.36). Rewards paid for meeting a standard of performance showed less (d = −0.28), and did not significantly reduce self-reported interest — though the differences between contingency types were not statistically reliable on the behavioural measure.

Read that quickly and you would conclude that the reward feature in this app is a mistake. Read it carefully and the conditions matter enormously.

The undermining effect is strongest when the task was already interesting, when someone else controls the reward, and when the payment is for turning up rather than for meeting a standard you set. It is weakest when the reward recognises a level of performance the person was aiming at, and on the self-report measure that kind of reward did no damage at all. That distinction has been argued over for thirty years, and the argument itself is the useful part: what the reward is contingent on is doing the work.

A prize you set yourself, on a bar you chose, for a habit you do not yet enjoy, is not the arrangement those experiments broke.

And the habits people bring to this app are almost never the interesting ones. Nobody needs a reward to keep doing something they already love. They need one for going first, for making the call, for the extra half hour — the things that are worth doing and do not feel good yet. Intrinsic motivation cannot be undermined where it has not turned up.

Why the bar is a rate

So the reward should attach to a behaviour. Which one?

Not a count. “Ten conversations this week” sounds like a behaviour, but it is hostage to how many chances the week produced. A quiet week makes it impossible and a busy week makes it trivial, and neither version says anything about you. That is the same problem as a habit goal you picked in advance.

An action rate does not have that flaw. It is actions divided by the opportunities that actually arose, so the denominator moves with the week. Sixty per cent means you took six of every ten chances you got, whether the week gave you five chances or fifty.

That is about as close to a pure description of behaviour as a single number gets. It is not a result. It is the share of the moments that were available to you in which you did the thing.

So what number should the bar be?

This is the question everybody asks second, and there is a real answer.

Start from what you already do, not from what you wish you did. Work out your action rate over the last two or three closed periods. Set the bar a little above it. The app helps here: the reward editor shows the last eight periods against whatever bar you are dialling, and tells you how many of them would have earned it. If the answer is eight of eight, the bar is decoration. If it is nought of eight, you have written yourself a rule you will never collect on.

Aim to miss sometimes. A bar worth having is one you clear on a good period and miss on a poor one. Something you would have earned four or five times out of the last eight is about right. That is not a rule from anywhere; it is just what a bar has to do to carry information.

In practice that lands most people somewhere between 60% and 80%. The app starts you at 60. Below about half and you are rewarding yourself for letting most chances go; above 80 and you are one bad Tuesday from a period that was good and did not count. Move it in fives, which is the step the editor uses, and move it up only after you have beaten it twice.

A bar you always clear is a decoration. A bar you never clear is a rule you have written to disappoint yourself with.

One more setting worth knowing: you can scope the reward to one habit or to all of them. One habit is the sharper instrument while you are building something specific. All habits is the gentler one, because a bad week on one can be carried by a good week on another.

Decide the prize before you have earned it

This is the part the app is strict about, and the strictness is the point.

You name the reward, the period and the bar before the period runs. When it ends, the app works out whether you cleared it. If you did, the reward is there to claim. If you did not, it expires and the next period begins.

There is no adjusting the bar on Sunday evening because you came in at 54%. That sounds severe, and it is the single most valuable thing in the feature. A bar you can move afterwards is not a bar — it is a conversation you are going to win, because you are the one having it with yourself, and you are extremely persuasive.

Committing in advance also does something useful to the week itself. When the condition is already written down, a chance arriving on Thursday is no longer a neutral event. You know exactly what it is worth.

Make it soon, and make it real

Two practical findings are worth more here than any amount of theory.

Woolley & Fishbach (2017), Immediate rewards predict adherence to long-term goals, Personality and Social Psychology Bulletin 43(2), 151–162. The presence of an immediate reward predicted persistence at a goal-related activity better than the presence of a delayed one. In the New Year’s resolution study, immediate rewards such as enjoyment predicted current persistence and delayed rewards did not. These are predictive measures, not an experiment that manipulated enjoyment.

Which argues for short periods. A reward at the end of the year is a reward you will not feel on a wet Tuesday in February. A reward at the end of this week is close enough to touch, which is why the app offers a day, a week, a month, a quarter and a year, and why most people should start at a week.

Second: make it something you actually want and would not otherwise have. A reward you would have taken anyway is not a reward, it is a Tuesday. The good coffee, the long walk, the album, the thing you keep not getting round to. Small and certain beats large and theoretical.

The other bar, and when to move to it

I would rather not oversell one idea at the expense of a true one.

Opportunity Habit has a second bar. Instead of your action rate it measures goals met — the share of your habits that hit their goal for the period. That is an outcome bar, and outcome bars are not wrong. They are the right instrument at a later stage, and the obviously correct one where the result is the whole point and is genuinely in your hands.

The recommendation here is only about ordering, and it is the one Zimmerman and Kitsantas found. Reward the rate while the behaviour is still being built. Move to the goal once the behaviour is reliable enough that the goal is mostly a consequence of it. Starting at the outcome skips the stage where you are learning what to repeat.

That second bar has its own quirks — it behaves completely differently on one habit than on five, and there is a case for setting it to zero — so it has its own guide: when to reward the goal, not the rate.

The mindset this is really for

Under the mechanics, this is an argument about attention.

If the thing being rewarded is a result, your attention goes to the scoreboard, and the scoreboard only updates occasionally and partly at random. If the thing being rewarded is the share of chances you took, your attention has to go to the chances — which means noticing them, which means looking for them.

That is the shift the whole app is built around. Not a person who wants a better outcome; nearly everybody wants that already. A person who spots the moment while it is still open, and acts in it, and has a number that says how often.

You cannot take more chances without first seeing more chances.

The reward is the part that makes the looking worth something in the short run, while the habit is still new and the outcome is still a rumour.

What to do with this

Pick one habit. Work out your action rate over the last fortnight — honestly, which usually means lower than you would guess. Set the bar a little above it. Name a reward you genuinely want and can have within a day of the period closing. Write all three down before the period starts.

Then leave it alone until the period ends, and let the arithmetic decide.

You will find out two things. Whether you clear the bar, which is mildly interesting. And what you do on the days you notice a chance and know precisely what it is worth, which is the part that changes how you behave.