How Much Evidence Is Enough for a Strategic Decision?

Before I ask for more research, I want to know what would make me choose differently.

“We need more information” can be a reasonable response to a difficult business decision. But it leaves a lot unexplained. What information? Would a different answer change the investment, the timing, or the decision to proceed? And can we obtain a useful answer before we have to act?

A team can accumulate reports while the assumption carrying most of the proposal remains untouched. A serious gap can also be dismissed as something the business will figure out later.

Neither gives me much confidence. I want the uncertainty described clearly enough that we can decide what to investigate, what to allow for, and what we are prepared to take on.

Which uncertainty could change the choice?

Start by separating an observation from the conclusion attached to it.

Several customers have asked about a new service. That is evidence of interest. The expectation that enough of them will pay the proposed price, stay long enough, and accept the intended service terms still needs support.

The proposal should state those assumptions alongside the revenue forecast built on them.

A request for data should explain how different findings would affect the decision.¹ For the proposed service, enough customers willing to pay the intended price might support a launch. Interest that depends on a substantial discount would require another look at the offer and its economics.

Another broad industry report may add context while doing little to establish whether these customers will buy. Offering them a clearly scoped paid trial would address that question more directly. Their response would still tell us little about renewal until enough time had passed.

What evidence would make the proposal less attractive?

If customers praise the idea but decline a paid trial, that response belongs in the assessment. Treating it as a sales problem by default would protect the proposal from evidence that should challenge it. The confirming-evidence trap affects both the information people seek and how they interpret it.²

The commitment changes how much evidence you need

A long lease and a permanent team require a stronger case than a limited service trial. The cost of being wrong is different, as is the ability to respond.

But “we can always stop” needs examination. A low-cost launch may still involve promises to customers, sensitive knowledge shared with a partner, or work diverted from existing relationships. Stopping the expenditure does not undo all of those effects. Reversibility depends on what it would actually take to change course.

For the proposed service, a narrower initial offer or a staged contract could reduce exposure to uncertain demand. A trial should be designed around what it needs to establish. A few heavily discounted orders, for example, would offer weak support for a full-price service.

Waiting also changes the situation. A customer may choose another supplier. A useful agreement may no longer be available. The team may spend another quarter carrying a problem that already has a credible solution.

The result also depends on what the business does after deciding. Phil Rosenzweig emphasizes this ability to influence outcomes, alongside the need to compete with rivals.³ A launch forecast depends on choices about the offer, staffing, sales, and delivery, as well as the market conditions being estimated.

What work is management prepared to fund and lead? What could competitors do in response? An attractive projection based on capabilities nobody has committed to developing is a weak basis for action.

That leads to a practical threshold: with a credible commercial case, a small step can proceed with less evidence when the downside is affordable and the business can change course without serious harm. A hard-to-reverse commitment needs stronger evidence on the assumptions that could make it unacceptable. If that evidence is still missing, narrow or decline the commitment, or wait for a specific answer. Waiting is justified when an obtainable answer could improve the decision enough to outweigh the delay; otherwise, any decision to proceed should state which uncertainty is being accepted.

Write down the reasoning before the result arrives

A decision can look obvious once everyone knows how it turned out. I want a record of what was actually known when it was made.

This can be a short note. It should capture the commitment being approved, the evidence behind it, the assumptions still unresolved, and when the decision will be reviewed. For example, an illustrative note for a new service might read:

We are approving a six-week paid trial with existing customers. Requests for ongoing support suggest interest; willingness to buy at the proposed price remains untested. The trial will use reserved staff hours. We will review paid uptake, delivery effort, and the effect on existing work before expanding. If paid uptake falls short of what the forecast assumes, we will revisit the offer and its price. If delivery depends on substantial unplanned owner involvement, we will revisit the scope and economics. Six weeks will not establish long-term renewal demand.

The value of that note is its specificity. It gives the team something to examine later and makes it harder to turn a modest test into an open-ended obligation without another decision.

At review, financial results matter. So does the explanation for them. Low uptake might expose weak demand, an unconvincing offer, or a failure to reach suitable buyers. The response depends on which explanation the evidence supports. Strong early sales deserve scrutiny too if exceptional effort or temporary conditions made them possible.

Assessing decision quality means considering what was known at the time, as Hammond and his coauthors recommend.² Keeping the original note allows that comparison. A disappointing outcome should prompt an examination of the assumptions and execution. Early success needs the same examination before it becomes a model for another investment.

When reviewing the decision, return to the unresolved assumptions recorded in the note. Which held up, which failed, and what does that mean for the commitment now?

---

References

1. Menon, Tanya, and Leigh Thompson. “How to Make Better Decisions with Less Data.” Harvard Business Review, November 7, 2016. Adapted as “Making Better Decisions with Less Data” in HBR’s 10 Must Reads on Decision-Making, Updated and Expanded, 2026. Original article.
2. Hammond, John S., Ralph L. Keeney, and Howard Raiffa. “The Hidden Traps in Decision Making.” Harvard Business Review, September–October 1998. Reprinted in HBR’s 10 Must Reads on Decision-Making, Updated and Expanded, 2026. Publication information.
3. Rosenzweig, Phil. “What Makes Strategic Decisions Different.” Harvard Business Review, November 2013. Reprinted in HBR’s 10 Must Reads on Decision-Making, Updated and Expanded, 2026. Original article.