A new breakdown every week — read the newsletter →
PeopleBusinessesTools
Strategies Mental ModelsDecision Tools Business ModelsFrameworksMoats
Learn Book SummariesReading Lists GuidesQuote CollectionsLearning Paths
Artificial IntelligenceNewsletter AboutContact
Decision Tools

Confidence determines speed vs. quality

Let how sure you are decide whether to ship fast or build carefully.

Overview

This model resolves the endless speed-versus-quality argument by adding a third variable: confidence. When you are not confident an idea will work, optimise for speed — ship the cheapest version that tests the assumption, because polishing something that might be wrong is waste. When you are highly confident, optimise for quality — the idea is proven, so craftsmanship compounds.

The trap it names is investing quality effort into unproven ideas (slow and risky) or shipping sloppy versions of things you already know people want (fast but self-sabotaging). Match the effort to the evidence.

When to use it

Deciding how much to invest in an idea before building — features, bets, experiments.

How to use it

Step 1

Assess confidence

How sure are you, honestly, that this will work? Base it on evidence, not enthusiasm.

Step 2

Low confidence → speed

Build the smallest thing that tests the core assumption. Learn before you invest.

Step 3

High confidence → quality

The idea is validated; now invest in doing it well, because quality now compounds.

Step 4

Re-assess as evidence arrives

Confidence is not fixed — as tests return data, move the same idea from speed mode to quality mode.

Worked example

A team debates whether to build a new feature beautifully or quickly. They are only 30% sure users want it, so they ship a rough version to test demand (speed). Data shows strong uptake, confidence jumps to 85%, and only then do they rebuild it properly (quality). They avoided polishing a feature that might have flopped.

Common pitfalls

  • Perfecting unproven ideas, burning time on things that may not matter.
  • Shipping permanently sloppy versions of proven, high-confidence work.
  • Mistaking personal excitement for genuine confidence.

Frequently asked questions

How do I measure confidence?

By evidence — prior data, user signals, comparable cases — not by how much you like the idea.

Doesn’t ‘speed’ mean shipping bad work?

It means shipping the smallest honest test, not shipping carelessly. The goal is learning, cheaply.

When do I switch to quality?

When evidence has raised your confidence enough that further investment is clearly justified.

Related tools