Hypothesis Thinking · Issue-Driven Approach
Hypothesis-Driven Approach
What is hypothesis-driven thinking?
- Set the answer first, then verify only the data you need
- Define in advance what evidence would prove you wrong
- Test the easiest-to-check, most critical conditions first
Adopting hypothesis thinking changes the very first question in the weekly meeting. Instead of “How much data have you collected?” it starts with “What do you think the answer is as of today?” Team members hold back, saying the evidence is still thin, but an answer built on thin evidence is exactly the map that shows what to dig into next week. The management metric also shifts from analysis progress to whether a hypothesis survives. A team that produces a plausible draft in week 3 arrives later than a team that gives a wrong answer on day 3 and scraps it.
Issue-driven work and hypothesis thinking go hand in hand, but they work at different layers. Issue-driven work picks out “which question must we answer now,” while hypothesis thinking proposes an answer to that question up front. A usable hypothesis meets three conditions: a clear subject and verb, verifiable with numbers, and room to be proven wrong. “Our sales capability is weak” is a complaint, not a hypothesis; “We visit new accounts half as often as competitors, so deals don’t convert” is a hypothesis. Nor does it mean looking at less data. You simply look at it in the opposite order; cut each test cycle to 2–3 days and run two or three hypotheses at once, so the team keeps moving even when one is rejected.
Practice most often goes wrong when a hypothesis is taken to mean “deciding the answer before you start.” When an executive opens a meeting with “I think it’s a pricing problem,” the team starts collecting only pricing data. That is not hypothesis testing; it is making the case for the approver’s opinion. The test is simple: if not a single hypothesis was rejected by the end of the project, no testing was ever done. In organizations where someone who changes a hypothesis is called “flip-flopping,” nobody changes course midway, and the same failure keeps recurring: conclusions overturned entirely three days before the deadline.
John KimInstructor John Kim, author of B2B Marketing Design and CEO of Nanum Consulting
Ara LimSpecialist in global leadership, AI, and business communicationAdapted from the hypothesis-testing procedure of the scientific method into a consulting problem-solving methodology, it has become the standard approach in strategic consulting practice at firms such as McKinsey.
Consider a hypothetical case. Company A makes industrial valves: a manufacturer with 520 employees and annual revenue of around KRW 90 billion (≈ US$67,335,000). In Q2, the repurchase rate on its flagship line fell from 68% to 51%, and management allotted 12 weeks to find the cause. A task force of five people from corporate planning and sales planning was formed.
- In the first meeting, the Head of Sales proposed pulling all three years of transaction data first, but TF lead Director Kim narrowed the question to a single one: 'Which customers left, and why?'
- The TF set its first hypothesis as 'customers who suffered delivery delays after the line expansion in the second half of last year churned.' If that were true, it specified, churned accounts should have at least twice as many late deliveries as retained ones.
- Instead of a full census, they took only the top 12 by revenue among the 47 churned accounts and cross-checked delivery records. When late deliveries differed by only 1.1x, they dropped the first hypothesis after four days.
- The second hypothesis was a change of sales rep. At 9 of the 12 companies the contact had changed since last September, and one client's purchasing manager said, 'The new rep doesn't know our drawing specs.'
- The task force introduced a handover checklist and two weeks of accompanied visits. The root-cause investigation, budgeted at 12 weeks, finished in 5, and two quarters later the repurchase rate recovered from 51% to 64%.