handbooksMaking better decisions
Which decisions deserve caution and which deserve speed, plus the seven traps that make product teams confidently wrong.
- Sort decisions by reversibility. Irreversible ones deserve evidence and care. Reversible ones deserve speed, because the cost of being wrong is a week.
- Most discovery decisions are reversible, and teams treat them as though they are not.
- Seven traps account for most bad product decisions. Six come down to working with one option instead of several. The seventh is ignoring what is already known.
- The fix for nearly all of them is the same: work with a set, not a single option.
Two kinds of decision
Not every decision deserves the same care. The useful split is whether you can walk it back.
Choosing a target opportunity is a two-way door: wrong today, cheaply reopened next week.
| Level 1 | Level 2 | |
|---|---|---|
| Reversible? | No, or only at high cost | Yes, cheaply |
| Examples | Pricing model, data architecture, a public commitment, deleting user data | Which opportunity to target, which idea to prototype, how to word a test |
| How to decide | Slowly, with evidence, with more people | Quickly, then learn from the result |
| The real risk | Moving too fast | Moving too slowly |
The mistake is applying Level 1 caution to Level 2 decisions. It looks responsible. It costs you the learning you would have had, and the decision usually gets made anyway, later, with no more information than you had at the start.
Choosing a target opportunity is a two-way door. If next week's interviews suggest a different branch matters more, you switch, and you have lost very little.
If this turns out to be wrong, what does it cost to undo? When the answer is "a week of work," decide now. When the answer is "we cannot," slow down and go get evidence.
Seven traps
These are the recurring ways product teams end up confidently wrong.
1. Falling in love with your first solution
The first idea arrives with the problem and feels obvious. Everything after that becomes a search for reasons it is right. The tell is that your test plan is designed to confirm rather than to discriminate.
The fix: never work with one idea. Carry at least three into evaluation, and require different people to advocate for different ones.
2. Looking too narrow
You frame the problem tightly, generate ideas inside that frame, and never notice the frame was the choice that mattered. Narrow framing feels focused and quietly rules out most of the solution space.
The fix: map sibling opportunities before picking one. Seeing what you are not doing makes the choice a real choice.
3. Seeking to confirm what you already believe
You go looking for support rather than for truth. Interviews turn into gathering quotes that back the plan, and disconfirming evidence gets explained away as an unrepresentative customer.
The fix: write down the result that would change your mind before you collect anything.
4. Letting short-term emotion decide
A frustrating meeting, a competitor launch, one angry customer, and suddenly a priority changes. The feeling is real and the information content is close to zero.
The fix: put distance between the trigger and the decision. If it still looks right in a week, it probably is.
5. Overconfidence
Certainty and accuracy are only loosely related. Teams that have been right before are the most exposed, because past success feels like evidence about the current decision and is not.
The fix: state confidence as a range and name what you are assuming. "We think this holds for enterprise customers, and we have not checked the smaller ones."
6. No benchmark
Without a comparison, any result can be read as good. A 4% conversion rate is excellent or terrible, and nothing in the number itself tells you which.
The fix: decide what counts as good before you look. Use your own history, a competitor, or a stated threshold.
7. Ignoring what is already known
Every team has a moment of deciding that the established practice does not apply here. Occasionally that is right. Usually it is a costly rediscovery of why the practice exists.
The fix: if you are departing from a known approach, say out loud what makes your situation different. If you cannot, you are not departing, you are guessing.
The common thread
Six of the seven come down to working with one option instead of several.
One idea produces confirmation bias, love for the first solution, and no benchmark. One opportunity produces narrow framing. One data point produces emotional decisions and overconfidence.
Sets fix this. A set of sibling opportunities, a set of competing ideas, a set of assumptions ranked against each other. Comparison is what turns a preference into a decision.
Delaying while you gather more information is itself a decision, and usually a worse one than acting on a reversible choice. Ask whether the extra data would actually change what you do. If not, decide.
Take the last significant decision your team made. Was it reversible? Did you treat it as though it was? The gap between those two answers is where your time is going.