How to Test Strategic Assumptions Before Committing: A Practical Guide

Most strategic decisions fail not because the logic was wrong, but because one or two load-bearing assumptions were never properly tested. This guide sets out how to find those assumptions, pressure-test them in the time you actually have, and decide whether to commit, adjust, or walk away.

Start by separating assumptions from facts

In most strategy papers, assumptions and facts are written in the same tone of voice. The first job is to pull them apart. Go through the business case and mark every claim as one of three things: evidenced (with a source you would defend to the board), inferred (reasonable but untested), or assumed (believed because it has to be true for the strategy to work).

The assumed category is where the risk lives. Typical examples in financial services: that a target customer segment will tolerate a repricing, that distribution partners will prioritise your product, that a regulator will treat a novel structure as analogous to an existing one, that integration synergies survive first contact with two different operating models, that key talent will stay through the transition.

Rank assumptions by consequence, not by comfort

Not every assumption deserves a test. Rank them on two axes: how material the assumption is to the decision, and how confident you actually are (not how confident the sponsor sounds). The assumptions that are both highly material and weakly evidenced are your testing agenda. Everything else is noise or can be monitored post-commitment.

Where teams go wrong: they test the assumptions that are easy to test, or the ones the sponsor is already confident about. That produces a thick evidence pack and no new information.

Design the cheapest credible test

For each priority assumption, ask: what is the smallest piece of evidence that would genuinely change my mind? The answer dictates the method.

• Customer behaviour assumptions: structured interviews with buyers who have made the exact decision you are modelling, not survey panels. Ten real conversations beat a thousand data points on stated preference.
• Competitor response assumptions: war-game the move with people who have worked at the competitor, or run a pre-mortem assuming they respond faster and harder than your base case.
• Regulatory assumptions: engage early and transparently with the relevant regulator on the substance of the question, and read recent speeches, Dear CEO letters and enforcement actions for the direction of travel. The goal is to understand the genuine supervisory expectation, not to secure a favourable reading.
• Internal capability assumptions: ask the people who would actually do the work, not their managers, how long it would take and what would break.
• Financial assumptions: identify the two or three variables that drive 80% of the NPV and sensitivity-test them against adverse but plausible values, not just the base case plus or minus 10%.

Look for disconfirming evidence on purpose

Assign someone, ideally outside the deal team, to argue the opposite case using the same data. If they cannot build a credible counter-case, either the strategy is unusually robust or the data is too thin. Both are useful to know.

The sponsor should not run this exercise. The incentive to find reassurance is too strong.

Decide what would make you stop

Before testing, write down the specific findings that would cause you to pause or kill the project. Doing this in advance prevents the common pattern where disconfirming evidence arrives and gets reinterpreted as a manageable risk. If no finding would stop you, you are not testing, you are seeking approval.

The commit decision

When you reconvene, the question is not "are we confident?" It is: of the assumptions we identified as load-bearing, which have been evidenced, which remain assumed, and are we willing to commit on the ones still assumed? If yes, document why. If no, either test further or reshape the decision so it does not depend on them.

The next action is simple: pull out your current strategy paper and mark every claim as evidenced, inferred or assumed. The gap between what you believe and what you have tested is the real agenda.

Frequently Asked Questions

How long should assumption testing take?

For most board-level decisions, two to six weeks of focused work is enough to test the three to five assumptions that actually matter. Longer than that usually means the team is retesting things they already know, or the decision itself is poorly framed.

Who should run the testing?

A mix. The deal team runs the mechanics, but disconfirming analysis and external interviews should sit with someone whose career does not depend on the deal proceeding. That independence is what gives the output credibility with the board.

What if a key assumption cannot be tested in time?

Name it explicitly, size the downside if it is wrong, and decide whether the strategy can be structured to limit exposure, through staged commitment, optionality, or exit rights. Untested assumptions are acceptable; unacknowledged ones are not.

How do we handle assumptions about regulatory treatment?

Engage the regulator directly on the substantive question, present your analysis fully including the weaknesses, and take the response at face value. Build the strategy to meet the supervisory expectation properly, not to the minimum you think you can defend.

What is the single most common mistake?

Treating the business case as the thing to validate rather than the thing to challenge. The purpose of testing is to find out whether the strategy is wrong while it is still cheap to change course.