One swap at a time: how Swappsy suggests a better option

Swappsy is still being built. This is a note on the thinking behind it rather than a description of a shipped product.

The decisions people most want to improve are the ones they make fastest. The coffee, the cleaning spray, the subscription renewing on Friday. Each is small enough not to justify research, and repeated often enough that the aggregate matters.

The design constraint follows from that: whatever the app does has to fit in the few seconds someone will actually give it. Anything that turns a small decision into a research task has already failed.

Why one alternative and not a list

A list is the obvious answer and it is the wrong one. Returning ten options converts a small decision into a comparison exercise — which is precisely the work the person was avoiding by not thinking about it in the first place. You have not saved them anything; you have added a step.

One alternative can be accepted or dismissed in a moment. It is a smaller promise and a much larger obligation: with only one suggestion there is nowhere to hide a bad one.

The reasoning is not optional

A recommendation without a reason asks for trust the app has not earned, and trust is the entire product. So every suggestion carries the reason with it, in a line short enough to read and specific enough to check.

Making the reason visible also disciplines the suggestion. If the justification cannot be written in one checkable line, the swap is usually not as good as it looked.

Whose definition of better

Cheaper, healthier and greener frequently point at different products, and a tool that silently picks one of those and calls it better is making a values judgement on the user behalf. Which weighting applies is the user setting, not ours. The same query returns a different swap depending on what that person has said matters, and the reason line changes to match.

The honest failure state is saying there is no better option. A tool that always finds a swap is not evaluating anything.

What we are watching for

The open question is whether one confident suggestion feels helpful or presumptuous, and we expect that to vary by category. Household goods and subscriptions are probably comfortable. Food is more personal, and a swap that ignores what someone actually likes eating will be rejected regardless of how well reasoned it is.