The Spectrum of Thinking Interventions
Kosh wants users to do three things: move money on payday, glance at their spending every morning, and decide whether to pay off a credit card or save. A single approach — say, "send a reminder" — will not suit all three. The first design decision is how much thinking each behavior should ask of the user.
The rule
Ask for as little thinking as the behavior and the user's consent allow.
Every intervention can be placed on a line according to how much conscious effort it asks of the user:
| Little thought | Some thought | Full thought |
|---|---|---|
| Cheating — the product does the action for the user | Habits — the user repeats a routine triggered by a cue, almost on autopilot | Conscious action — the user weighs and decides each time |
| Automatic payday transfer | Checking spending after morning tea | Choosing between paying debt and saving |
"Cheating" sounds sneaky but is not: it means the user agrees once, and after that the behavior no longer depends on their attention or willpower.
Why start at the low-thinking end? Attention and willpower are scarce and unreliable, as the fast and slow modes showed. A behavior that needs fresh deliberation every time will be skipped whenever the user is tired or busy.
But two things push you towards the conscious end:
- Consent. You may only automate what the user has knowingly agreed to.
- Variety and stakes. If the right choice changes from case to case, or the stakes are high and personal, the user must think. Automating a decision that needs judgement makes things worse.
The shortcut
Ask three questions, in order: Can the product do it for them, with their consent? If not, can it become a routine tied to a cue? If not, how do we help them decide well? Stop at the first yes.
Payday transfer
Same action every month, and the user asked for it. The product can do it: cheating, with an auto-transfer set up once.
The mistake at one extreme is making people think about everything, which fails when they are tired. At the other, it is automating choices that need a person's judgement, or were never agreed to.
Drills
Answer fast — trust the rule.
Which strategy asks the user for the least conscious thought?
A user must choose the tenure of a home loan. Which strategy fits?
A small action repeats daily, cannot be done by the product, and has a natural cue. Which strategy fits best?
Why prefer the low-thinking end when it is ethical and appropriate?
Remember
- Ask for as little thinking as the behavior and the user's consent allow.
- Shortcut: can we do it for them? Can it become a habit? Otherwise, help them decide.
- Trap: automating choices that need judgement or were never agreed to.