What does it mean to decide with incomplete information?

It means choosing an action while relevant facts or future outcomes remain uncertain, using the evidence available and a plan for managing errors.

When should you stop gathering information and decide?

Decide when another check is unlikely to change your action enough to justify its cost and delay, provided the remaining risk is acceptable.

How to Make Better Choices With Incomplete Information

You have two proposals, a deadline, and a question neither proposal answers. One supplier looks faster. The other has stronger references. You could spend another week comparing them, but the project would start late.

To make better choices with incomplete information, define the decision, identify which uncertainty could change your action, and investigate that uncertainty before committing. When the remaining risk is acceptable and another check is unlikely to justify its cost, choose and set a review point.

The goal is a decision you can defend using what was available at the time. A good process can still produce an unlucky outcome; waiting for certainty can also produce a costly one.

What Makes Choices With Incomplete Information Difficult?

Incomplete information means you lack relevant facts about your options, their consequences, or the conditions in which they will play out. Sometimes the missing fact already exists: a supplier's available capacity. Sometimes it does not: how customers will respond to a service you have not launched.

That difference matters. You can ask the supplier about capacity. Customer response may require a small experiment, and even that experiment will leave uncertainty.

A useful question is: What would I do differently if I knew the answer?

If the answer would change your choice, commitment size, or safeguards, the information may be worth pursuing. If it would only make you feel more informed, it may deserve less attention.

This is the practical idea behind the value of information: information is valuable when it can improve an action enough to justify the effort, expense, and delay of obtaining it.

Step 1: Define the Choice and Its Constraints

Write the decision in a sentence that names an action and a deadline.

"Research suppliers" is an activity. "Choose a supplier by Friday for a four-week pilot" is a decision.

Then list the requirements an acceptable choice must meet. Separate requirements from preferences. A supplier who cannot meet a necessary delivery date should not win because their presentation looks more polished.

For a small business commissioning a booking system, the requirements might be:

Deliver a usable version before the seasonal rush.
Stay within the agreed spending limit.
Allow the business to export its customer records.
Provide a way to stop the project if the first milestone fails.

Preferences might include visual style or a familiar project management tool. Those still matter, but they should not obscure the conditions that make the project viable.

This step also exposes a neglected option: changing the scope. You may not need to choose between two large proposals if a smaller version would solve the immediate problem.

Step 2: Separate Evidence From Assumptions

For each option, record what you know and what you are inferring.

"The supplier demonstrated an existing booking system" is evidence. "They can adapt it to our unusual scheduling rules in four weeks" is an assumption until you have checked the relevant work.

Keep the list short. Identify the two or three assumptions that could make an otherwise attractive choice fail.

Ask where each belief came from:

A direct observation?
A comparable completed project?
A claim by someone who benefits from your agreement?
Your own expectation about how things usually work?

None of these sources automatically settles the question. A sales claim can be accurate, and an impressive reference can describe a much simpler project. The purpose is to judge how much weight the evidence deserves.

Probabilistic thinking helps here: treat a conclusion as provisional and express confidence in proportion to the evidence. You can say "plausible but untested" without inventing a precise percentage.

Step 3: Find the Question That Could Change Your Action

Imagine receiving an answer that contradicts your current preference. Would you change suppliers, reduce the scope, delay the project, or add a safeguard?

If yes, you have found a potentially useful question. If no, ask whether you are researching a detail that has little bearing on the decision.

For the booking system, "Can the supplier handle overlapping appointments with different staff availability?" could change the choice. "Which slide template did they use?" probably would not.

Choose the cheapest credible way to resolve the important uncertainty. That might be a reference call, a demonstration using your own example, or a paid sample of the difficult work.

A cheap check is useful only if it tests the claim that matters. Watching a generic product tour tells you little about whether your unusual scheduling requirement will work.

A Worked Example: Choosing a Supplier Before Friday

Consider a hypothetical business deciding between two suppliers for its booking system.

Supplier A quotes $6,000 and promises delivery in four weeks. Supplier B quotes $7,000 and has delivered a similar system for another business. Both fit the spending limit. Missing the seasonal launch would require extra manual administration.

The owner initially favors A because of the lower price. The biggest uncertainty is whether A's existing software can handle the scheduling rules without extensive custom work.

Here are three possible next steps:

Next step	What it could reveal	Main limitation
Read more general software reviews	Common strengths and complaints	May say little about this scheduling problem
Ask A to demonstrate the exact scheduling case	Whether the current system supports the difficult requirement	A demonstration does not prove the whole project will arrive on time
Commission a limited first milestone	Whether A can implement and deliver relevant work	Costs money and uses some of the available time

The owner requests the demonstration first because it is quick and directly relevant. A shows that the existing software cannot handle the requirement without a substantial change.

That evidence changes the action. The owner can now ask B to confirm the same requirement, negotiate a smaller initial release, or keep the current manual process while testing a solution. The initial price comparison no longer captures the real tradeoff.

Suppose A instead demonstrates the feature successfully. That would strengthen the case for A, but it would not eliminate delivery risk. The owner might still use a first milestone with clear acceptance criteria and retain access to all project files.

The lesson is specific: one targeted check can be more useful than a large amount of loosely related research. Its value comes from the decision it changes.

Step 4: Match Your Commitment to the Remaining Uncertainty

After checking the crucial assumption, ask how much you need to commit now.

Some decisions can be tested in stages. You can pilot a service with a few customers, commission one deliverable, or try a new routine for two weeks before reorganizing your calendar around it.

A staged commitment works when the early stage produces useful evidence and leaves later options open. Define in advance what you expect to learn and what would justify continuing.

For the booking project, that might mean:

Implement the difficult scheduling case using test records.
Have the staff member who manages appointments try it.
Continue only if it handles the agreed cases and the records can be exported.

Check the actual exit costs. A small first payment does not make a project easy to reverse if your records become inaccessible or the contract commits you to the full fee.

When a decision is difficult to undo, demand stronger evidence or a larger buffer. If the downside would be unacceptable, a deadline alone does not make the choice sensible. Reduce the commitment, find another option, or decline it.

Step 5: Set a Stopping Rule and a Review Point

Before opening another tab, name the check you intend to make and the result that could alter your action.

A practical stopping rule is: I will decide once I have checked the main failure risk, confirmed the essential constraints, and found no further check worth its cost and delay.

This requires judgment. You rarely know the exact value of another conversation. But you can ask whether it is likely to reveal anything different from the last three conversations, and whether waiting consumes time you need for implementation.

Then record a review point. For the supplier, review progress at the first deliverable, while there is still time to respond. Specify what evidence would make you continue, change scope, or stop.

Bayesian thinking offers a useful habit for that review: update your view when new evidence arrives. A provisional choice should remain open to revision when its supporting assumptions weaken.

Common Mistakes to Avoid

Collecting more of the same evidence. Five similar testimonials may add less than one demonstration of the requirement you are worried about. Seek evidence that tests a different failure mode.

Treating delay as free. Waiting can protect you from a bad commitment, but it can also consume a launch window or leave a problem unresolved. Include that cost in the comparison.

Giving guesses numerical authority. A spreadsheet with unsupported probabilities can look more reliable than it is. State assumptions clearly and check whether changing them would reverse the choice.

Judging the process only by the outcome. A successful project does not prove every shortcut was sensible. A failed project does not prove that you ignored available evidence. Review what you could reasonably have known and which checks would have helped.

A Short Decision Note You Can Reuse

Before making the choice, write six lines:

Decision and deadline: What must I choose, and by when?
Essential constraints: What must an acceptable option satisfy?
Main uncertainty: Which missing fact could change my action?
Useful check: How can I investigate it credibly and cheaply?
Commitment limit: What can I afford to lose if I am wrong?
Review trigger: What new evidence would make me revise the choice?

You will still have unanswered questions. The note helps you identify whether those questions justify more work or whether you have enough to take a manageable next step.

For a broader set of tools to support decisions like these, explore 100 Mental Models.