insights

InsightHub

How Top PMs Make Decisions With Incomplete Information

October 6, 2025
schedule 4 min read

The Perfect Information Trap

Junior PMs often delay decisions waiting for more data. Senior PMs know that "more data" is a mirage—there's always more to learn. The skill isn't having complete information; it's knowing when you have enough.

Jeff Bezos calls this the "70% rule": If you have 70% of the information you wish you had, make the decision. Waiting for 90% means you're probably moving too slowly.

But how do you know when you're at 70%?

Framework 1: Reversibility Assessment

Ask: How reversible is this decision?

Type 1 decisions (irreversible):

  • Major architectural changes
  • Public commitments
  • Pricing model changes
  • Market positioning shifts

For Type 1: Seek more information. These are worth getting right.

Type 2 decisions (reversible):

  • Feature details
  • UI variations
  • Experiment designs
  • Process changes

For Type 2: Decide faster. You can course-correct.

Most decisions are Type 2, but organizations often treat them like Type 1.

Framework 2: Minimum Viable Confidence

Ask: What would I need to believe to move forward?

Instead of asking "Is this the right decision?" ask "What assumptions would need to be true?"

Then assess each assumption:

  • How confident am I? (High/Medium/Low)
  • What evidence supports this?
  • What would change my mind?
  • Can I test this quickly?

If key assumptions have low confidence AND are testable, test before deciding. If they're not testable, decide based on your best judgment.

Framework 3: Regret Minimization

Ask: Which mistake would I regret more?

Two types of errors:

  • False positive: Doing something that shouldn't have been done
  • False negative: Not doing something that should have been done

In product, false negatives are often more costly:

  • The feature you didn't build that competitors did
  • The market you didn't enter that exploded
  • The problem you didn't solve that caused churn

Bias toward action when false negatives are more costly than false positives.

Framework 4: Disagree and Commit

Ask: Do I need consensus?

Some decisions require alignment. Most don't.

Require consensus:

  • Decisions that need cross-team execution
  • Decisions that affect others' work
  • Decisions with significant downside risk

Don't require consensus:

  • Decisions within your authority
  • Reversible experiments
  • Low-stakes optimizations

When you disagree with a decision but can't achieve consensus, "disagree and commit"—state your concerns, then execute fully if the decision goes against you.

Practical Tactics for Faster Decisions

Tactic 1: Time-box deliberation

  • Set a deadline for the decision
  • Gather information until the deadline
  • Decide at the deadline regardless

Tactic 2: Pre-mortem analysis

  • Imagine the decision failed
  • List reasons why it failed
  • Address the most likely failure modes
  • Proceed if failure modes are acceptable

Tactic 3: Pilot before commit

  • Can you test with a subset of users?
  • Can you build a smaller version first?
  • Can you validate assumptions before full investment?

Tactic 4: Seek disconfirming evidence

  • Don't just look for data supporting your hunch
  • Actively seek evidence that you're wrong
  • Decision is stronger if it survives disconfirmation

Tactic 5: Separate data gathering from deciding

  • Research phase: Open-ended exploration
  • Decision phase: Closed evaluation of options
  • Don't let research become procrastination

The Meta-Skill: Calibration

Over time, track your decisions:

  • What did you decide?
  • What was your confidence level?
  • What was the outcome?

This calibrates your intuition. You learn which types of decisions you make well with little information, and which require more.

Great PMs aren't right more often—they're right enough to learn, wrong in recoverable ways, and calibrated about their own uncertainty.