How to Run an Assumption Audit on Any Business Decision

Every business decision rests on assumptions. Most of them are invisible.

A product roadmap assumes the team knows what customers want. A hiring plan assumes the current org structure is right. A pricing change assumes customers will pay more for the same thing. These assumptions aren’t wrong by default — but when they go unexamined, they become the architecture of expensive mistakes.

An assumption audit is a structured way to make those hidden beliefs visible, sort them by how much evidence supports them, and decide which ones to test before committing resources. It takes about 30 minutes for a single decision, and it can prevent weeks or months of misdirected work.

This article walks through the process step by step, with a worked example.

Why assumptions matter more than analysis

Most decision-making effort goes into analyzing options: comparing costs, modeling outcomes, debating trade-offs. That analysis is valuable — but it all happens inside a frame. And the frame itself is built on assumptions that rarely get examined.

Consider a team debating whether to build Feature A or Feature B. They model the engineering cost, estimate user impact, project revenue. What they don’t question is whether either feature addresses the actual reason customers are leaving. The analysis is rigorous. The frame is wrong.

As one Hacker News commenter observed: “I meet a lot of people who don’t think to question the assumptions of the problem statement.” The assumption audit targets exactly this gap — the space between “the problem as given” and “the problem as it actually is.”

The assumption audit: 5 steps

Step 1: State the decision clearly

Write out the decision you’re about to make in one sentence. Be specific.

Weak: “We need to improve our product.” Strong: “We’re deciding whether to invest Q4 engineering capacity in an AI analytics dashboard.”

The clearer the decision statement, the easier it is to surface the assumptions hiding inside it. Notice how the “strong” version already reveals embedded choices — Q4 timing, engineering allocation, the specific feature.

Step 2: List every assumption

This is the core of the audit. Write down everything that must be true for this decision to be the right one. Don’t filter. Don’t judge. Just list.

For the AI dashboard decision, the list might look like this:

1. Customers want AI-powered analytics
2. An AI dashboard will reduce churn
3. Churn is primarily a product/feature problem
4. Competitors’ AI features are driving their growth
5. Our engineering team can build this in one quarter
6. Customers compare features when deciding whether to renew
7. Building new features is a better use of Q4 than improving existing ones
8. The AI dashboard will be used regularly, not just tried and abandoned

Tip: Ask each person involved to write their list independently before sharing. You’ll be surprised how different the lists are — and how many assumptions only one person holds while everyone else takes them for granted.

Step 3: Label each assumption

For each assumption, assign one of four labels:

• Fact — backed by direct evidence (data, research, documented observation)
• Inference — seems reasonable but isn’t verified
• Convention — believed because it’s standard in the industry or organization
• Unknown — no one on the team has information either way

Here’s what the AI dashboard list looks like after labeling:

The pattern is usually striking: most assumptions are inferences or conventions. Very few are facts. This table alone often changes the conversation.

Step 4: Rank by impact and uncertainty

Not every assumption is equally dangerous. Rank them by asking two questions:

1. If this assumption is wrong, how bad is the damage? (Impact)
2. How confident are we that it’s right? (Certainty)

The assumptions that are both high-impact and low-certainty are your audit priorities — the ones you need to test before proceeding.

For the dashboard example:

Step 5: Design a test for each priority assumption

For each high-priority assumption, ask: What’s the cheapest way to find out if this is true?

Total time to test all four: roughly 2-3 weeks. Compare that to a full quarter of building the wrong thing.

A worked example: the bakery that didn’t cut prices

A small-town bakery was losing customers to a national chain that opened nearby. The owner’s first instinct — and the conventional response — was to cut prices.

But instead of reacting, she ran a quick assumption audit:

Decision: Lower prices to compete with the chain.

Assumptions:

1. Customers are leaving because of price
2. The chain’s prices are significantly lower
3. Matching their prices is financially sustainable
4. Customers choose bakeries primarily on price
5. The bakery’s current offerings are interchangeable with the chain’s

Labels:

• #1: Inference (no customer feedback collected)
• #2: Fact (verified by checking)
• #3: Unknown (hadn’t modeled the margin impact)
• #4: Convention (assumed because “everyone competes on price”)
• #5: Inference (never asked customers what they valued)

Priority assumptions: #1 and #4 — both high-impact, both unverified.

Test: She spent a week talking to regulars and lapsed customers. What she heard surprised her. Customers didn’t leave because of price. They left because the chain had longer hours and a drive-through — convenience, not cost. And the customers who stayed valued freshness, personalization, and the experience of watching bread being made.

Instead of cutting prices (which would have damaged margins without solving the problem), the bakery leaned into what made it different: made-to-order cakes, a viewing window into the kitchen, and extended morning hours. Within months, the bakery had grown its customer base — not by competing on the chain’s terms, but by understanding what customers actually valued.

The assumption audit took a week. The wrong decision would have cost the business.

When to use an assumption audit

An assumption audit is most valuable when:

• The stakes are high — significant resources, time, or strategic direction at risk
• The team is confident — paradoxically, high confidence without evidence is exactly when assumptions are most dangerous
• The decision is inherited — someone else framed the problem, and the team is executing without questioning the frame
• You’re copying a competitor — adopting someone else’s strategy means adopting their assumptions too
• The problem feels stuck — repeated attempts at the same type of solution suggest the frame, not the execution, is wrong

You don’t need to audit every decision. Routine, reversible, low-stakes choices don’t justify the time. (The 60-second decision test can help you decide which problems are worth the deeper analysis.)

The assumption audit template

Here’s the format. Copy it into a doc, spreadsheet, or whiteboard:

DECISION: [One sentence describing the decision]

ASSUMPTIONS:
#  | Assumption | Label (Fact/Inference/Convention/Unknown) | Evidence

PRIORITY RANKING:
#  | Assumption | Impact if wrong | Confidence level

TESTS:
#  | Assumption | Cheapest test | Timeline | Success criteria

RESULTS:
#  | What we learned | Decision: proceed / pivot / investigate further

Use it before your next product decision, strategic plan, or resource allocation. It takes 30 minutes to fill out and could save your team months.

The deeper skill

Running an assumption audit once is useful. Running it habitually changes how you think about decisions.

Teams that practice this regularly develop a shared language: “Is that a fact or an inference?” becomes a normal question in meetings, not an accusation. The audit makes disagreements productive — instead of arguing about conclusions, people argue about evidence. And when a decision does go wrong, the audit provides a clear record of what was believed and why, making it easier to learn from the mistake.

In The First Principles Playbook, this process of separating facts from inherited assumptions is the foundation of the entire method. The book goes deeper — with 20+ case studies, exercises for inversion and pre-mortem analysis, and frameworks for knowing when to challenge conventions versus when to trust them.

Want the full 8-step process? The Problem Deconstruction Worksheet walks you through it — from stating the real problem to selecting your cheapest test. Free, printable, no signup required.