Explainer · Problem Solving & Consulting · Intermediate

Start with the answer: hypothesis-driven problem solving

Consultants guess the answer on day one, then try to prove themselves wrong. Done honestly it is faster than gathering everything first, and it is not bias.

Metanoia Editorial · 12 Oct 2026 · 6 min

Start with the answer: hypothesis-driven problem solving

What is happening

Most people solve problems by collecting information until an answer appears. It feels thorough, but on a real deadline it rarely finishes: there is always more data, and nothing tells you which data matters. The hypothesis-driven approach reverses the order. You state your best current guess at the answer early, list what would have to be true for it to hold, and then gather only the evidence that would confirm or kill it.

The method is described in Ethan Rasiel’s “The McKinsey Way” and developed in Conn and McLean’s “Bulletproof Problem Solving”. Its safeguard is the willingness to be wrong. A hypothesis is a tool for focus, not a conclusion to defend. If the first test fails, you drop the hypothesis and form the next one, and the failed test still taught you where the answer is not.

Why it matters

Time is the scarcest resource on any project. A hypothesis tells you which analysis to run first, and it lets your manager react to a concrete idea instead of waiting for a pile of facts.

The framework

The hypothesis loop

State a falsifiable answer“Sales fell because a competitor undercut our price in two key cities” can be tested. “Market conditions changed” cannot.
List what must be trueUsually two to four conditions: the competitor cut prices; our decline is concentrated in those cities; customers cite price.
Design the cheapest test for eachA price check, a sales split by city, five customer calls. Hours, not weeks.
Run tests in order of powerStart with the test most likely to kill the hypothesis.
Keep, refine or replaceUpdate the hypothesis and say so aloud. Changing your mind on evidence is the method working.

Worked example

High downloads, low use

A bank’s new mobile app has strong downloads but few monthly active users. Day-one hypothesis: users download it for one task, such as checking a balance after a branch visit, and find no reason to return because the app lacks the features they use weekly.

What must be true: most users open the app only in their first week; their first session is a single simple task; the features they use weekly elsewhere, such as transfers or e-wallet top-ups, are missing or hard to find. Tests: a cohort retention curve, a first-session event log, and a feature comparison with the two most-used alternatives.

If the retention curve shows users stay for a month and then leave, the hypothesis is wrong in an informative way: the problem is not first-week value but something that breaks later, such as login friction after an update. The next hypothesis is already half-written.

Illustrative case; no real bank is described.

What this means for you

Fresh graduateIn a case interview, offering a hypothesis early, and dropping it gracefully when the data disagrees, signals maturity.

Early professionalOpen your next update with “my current view is…” rather than a list of what you have done. Managers can steer a view; they cannot steer a list.

Experienced professionalAsk your team for their hypothesis before their analysis plan. It shows you how they think and saves days of misdirected work.

What to do

For your current biggest problem, write a one-sentence hypothesis you could be proven wrong about.
List three things that must be true and the cheapest test for each.
Run the most powerful test first and write down what you would conclude from each possible result before you see it.
Share the hypothesis with one colleague and invite them to attack it.

Practice drill

Hypothesis and three tests6 min, timed

A regional minimarket chain’s sales per store have been flat for a year while its new stores grow quickly. Write one falsifiable hypothesis, three conditions that must be true, and the cheapest test for each.

Ask whether new stores could be taking sales from old ones.
Make the hypothesis specific enough to fail.

Show a model answer

Hypothesis: new stores opened near existing ones are drawing customers away, so the chain’s growth is partly a reshuffle of the same customers rather than new demand.

Must be true: (1) old stores within a short distance of a new opening show sales drops after the opening date, while distant old stores do not; (2) transaction counts fall more than basket size at affected stores; (3) loyalty-card data shows the same customers shifting stores. Tests: a before-and-after comparison by distance band, a split of sales into transactions and basket, and a match of loyalty IDs across stores.

If distant stores are also flat, the hypothesis fails and the next one points to a chain-wide cause, such as range, pricing or a competitor.

Reflect

Which belief about your own career have you never tested? What would the cheapest test be?
How do you react when your first idea is proven wrong in front of others? What would make that easier?

Do it with the pillars

The Rope
The Rope’s Module 3 runs a “hypothesis hour” on a real role before the interview; Module 6 applies the method inside a case.The Route
The Route’s Lesson 9.4, Thinking Like a Leader as an Individual Contributor, uses the same loop with a pre-mortem at work.

Sources & provenance

Ethan M. Rasiel, “The McKinsey Way” (McGraw-Hill, 1999); Charles Conn and Robert McLean, “Bulletproof Problem Solving” (Wiley, 2018). Original Metanoia writing; both cases are fictional.

Written for

Fresh graduateEarly professionalExperienced professional