Opinionated Simplicity
Remove every choice you can; each user decision is an abdication of your judgement.
- Difficulty
- Advanced
- Time to result
- ~weeks to results
- Steps
- 6
- Confidence
- 75%
The mechanism inverts the usual product instinct. Rather than resolving disagreement by shipping an option, you resolve it by forming an opinion and deleting the option. Every click, extra button, and settings entry is treated as evidence that the team declined to decide something and handed the decision to the user instead. Choices for the user are an abdication of your responsibility, tolerable only for legal or genuinely important reasons. The reference arc is search: Google looked like the simplest possible product, a single box, yet it still trained users to enter careful keywords rather than talk naturally, and it returned a page of results to sift and sort. Conversational AI removed even that, letting people just ask and get one straight answer that is good enough. The output is a product where the surface area shrinks, the remaining details are made perfect, and the burden of deciding moves from the user back to the builder.
Origin
Extracted from the Naval podcast episode Curate People, framed around Jack Dorsey's rule to limit the number of details and make every detail perfect, and illustrated with the arc from Google to conversational AI products.
Core principles
- 01To be simple you have to be extremely opinionated.
- 02A strong vision of what a product should not do is as important as what it should do.
- 03Every setting is a decision you refused to make on the user's behalf.
- 04Users do not want choices; they want the cognitive load taken off them.
- 05Without an opinion you get a giant mess of competing features.
- 06Limit the number of details, then make every remaining detail perfect.
How to run it
- 1
Form the opinion first
Decide explicitly what the product should and should not do before designing anything. Without that, feature requests have no filter to fail.
Watch out If you do not have a strong vision of what it should not do, you end up with a giant mess of competing features.
- 2
Inventory every decision you pushed onto the user
Enumerate all clicks, buttons, toggles, and settings. Treat each as an open question you failed to answer.
Pro tip The settings menu is the fastest place to find abdicated responsibility.
- 3
Choose the right default
For each item, work out what the user should be doing or looking at, and make that the behaviour rather than an option.
- 4
Delete the option
Once the default is set, remove the control entirely instead of leaving it as an escape hatch.
Pro tip Struggle and resist against every single choice; the burden of proof sits with keeping the option, not removing it.
Watch out Keep only the exceptions that legal or genuinely important constraints demand.
- 5
Perfect what remains
With the surface area reduced, spend the recovered effort making each surviving detail exact rather than adding new ones.
- 6
Retest against a simpler baseline
Compare the product against the simplest thing a user could plausibly do instead, and ask what else can be removed.
Pro tip You cannot make a product that is simple enough; treat the ceiling as unreachable.
In the wild
Google looked like the simplest product anyone could build: one box. But that box still had rules. Users were trained not to speak conversationally, to pick keywords carefully, and to accept that synonym matching was imperfect. It then returned a page of results to sift through, judging which were ads, which were real, and whether the ordering could be trusted. Chat-based AI removed all of that. You talk to it in your own words, by voice or text, and get one straight answer back. It is not always right, but it is good enough, and it made the previously simplest product on the internet look complicated.
→ A product widely considered maximally simple was displaced by removing the remaining decisions it still imposed.
A team reviews its preferences screen and finds a dozen toggles, most added to end internal disagreements. Under this framework each toggle is reclassified as an unresolved product decision. The team picks the correct default for each, deletes the control, and keeps only the two required for legal reasons.
→ The settings surface collapses to two entries and the team is forced to hold an actual opinion about each behaviour.
Common mistakes
Shipping an option to settle an argument
Adding a setting because the team cannot agree converts an internal disagreement into permanent user cognitive load.
Confusing minimal visuals with simplicity
A clean interface that still trains users in its own rules is not simple. The measure is how many decisions the user has to make, not how sparse the screen looks.
Defining only what the product should do
Without an explicit statement of what it should not do, every plausible feature request passes and the product becomes a pile of competing features.
Is it for you?
Best for
Consumer product teams and founders holding product vision who need to decide what to cut rather than what to add.
Not ideal for
Deep professional or enterprise tooling where expert users legitimately require fine-grained configuration.
From the transcript
“To be simple, you have to be extremely opinionated.”
“In fact, things in the settings menu are an indication that you've abdicated your responsibility to the user. Choices for the user are an abdication…”
“Jack Dorsey has a great phrase limit the number of details and make every detail perfect.”
From the episode
Curate People