Before a significant decision, someone usually suggests gathering more information: another quote, a customer survey, a trial, a market study, a few more months of data. It sounds prudent. Sometimes it is. But information is not free. It costs money, time and attention, and waiting can lose a customer, a supplier slot or a market window. More data can also increase confidence without reducing the uncertainty that actually matters.
There is a second, related problem. The ability to change course is highest at the start of any initiative and falls as commitments are made: orders placed, leases signed, equipment installed, staff hired, customers told. Most businesses can work out what it would cost to finish a project and, with some effort, what it would cost to stop it. Few can say when stopping stopped being a realistic option, because nobody recorded which decisions removed which options.
This article explains how to judge when learning more is worth it, why the most valuable time to learn is just before an irreversible step, how the ability to reverse a decision runs out in steps rather than gradually, and a simple way to keep track of what can still be undone.
Information is worth something only if it could change the decision
The value of information depends on whether it could change what you do. A thorough study has little value if every possible result leads to the same action. A small, cheap test can be extremely valuable if it decides whether you spend the next large sum.
Before commissioning any analysis, trial or research, ask five questions:
- Consequence: how much is at stake if the current judgement is wrong?
- Reducibility: can this information actually reduce the uncertainty that matters, within the time available?
- Action sensitivity: would different results lead to a different decision, budget or design?
- Timing: will the information arrive before the next irreversible commitment?
- Cost: is the cost of finding out lower than the value of avoiding a poor decision?
If the answer to the third question is no, the information may be interesting but it is not worth paying for in this decision. A simple habit helps: for every study, trial or analysis requested, write down which decision it will change and how.
Three common misconceptions
- More data is always better. Large amounts of data can raise confidence without reducing the key uncertainty. Evidence must be tied to the decision.
- Research should remove uncertainty. Some uncertainty cannot be removed within the time available. The aim may be to limit exposure rather than to know the future.
- Waiting is neutral. Delay can cost a customer, a window or a price. The value of learning has to be weighed against the value of acting now.
Learn just before the point of no return
Information is most valuable just before the business crosses a threshold that is hard to reverse: signing a long lease, ordering specialised equipment, hiring a team, launching to customers. Testing after that point is often too late to change anything important.
So place the learning ahead of the commitment. That may mean spending a little early on a prototype, a small trial or a sample run that does not itself produce revenue. The best test is not a small version of everything. It is a test aimed at the uncertainty that would most change the decision, designed to show quickly whether the core idea holds.
And know when to stop analysing. Once the remaining uncertainty would not change the decision, or the cost of delay is rising, further study is not prudence. It is delay. Set a stopping rule in advance: enough evidence to make a defensible choice, not perfect knowledge. The plans that detect rather than predict article covers designing tests around the assumptions that carry the most value.
The ability to change course runs out in steps
The idea that influence over a project is highest at the start and falls over time is familiar. What is less often noticed is that it does not fall smoothly. It falls in steps, each tied to a particular decision:
- placing a long-lead order;
- signing a lease or long contract;
- demolishing or modifying a building;
- installing equipment;
- hiring or making people redundant;
- announcing a date or product to customers;
- migrating data and switching off the old system.
Between those steps, little changes. At them, an option can disappear in an afternoon. Because the overall process feels gradual, it is easy to believe there is still room to change course long after the decisive step has passed.
The decision that removes the most options is often not the one that costs the most money, and it is rarely presented as a decision about options at all. A small order placed early to secure a delivery slot can commit the business more firmly than a much larger payment later.
Three kinds of reversibility
It helps to think of reversibility in three forms, which run out on different schedules:
| Kind | Question | Example of it running out |
|---|---|---|
| Physical | Can we put things back as they were? | Old equipment scrapped, old records destroyed, building altered |
| Contractual | Can we exit agreements at a known price? | A lease or supply contract with no affordable exit |
| Relationship | Can we change course without losing trust? | A launch date announced to customers or a commitment made publicly |
The real position is the lowest of the three, not the average. A supplier contract might remain easy to cancel, while the physical option has gone because the old system was dismantled. Watching only the contract creates false confidence.
Keep a simple register of what can still be undone
A short register, kept alongside the risk list for any significant initiative, makes this visible. For each major commitment, record:
- What reversing it would actually involve: the physical, contractual and relationship steps.
- Its current status in each form: available, costly or closed.
- When each is expected to close. A forecast is fine; no date at all is the finding.
- Who is responsible for knowing and for declaring it closed.
- What was gained by giving up the option: time, price or certainty.
Add one line to every significant approval: what does this decision make impossible? If an approval closes an option without saying so, it is not ready to sign.
Watch for decisions made for other reasons
The commitments that remove options fastest are often made for a different purpose. A long-lead order is placed early to protect a delivery date. A contractor is allowed to start early to keep their crew. A launch date is announced to reassure a customer. Each is a reasonable trade, but only one side of it is usually written down: the time or price gained is recorded, the option given up is not. Asking “what does this make impossible?” at the moment of each such decision brings the other side into view.
Buy reversibility on purpose
Sometimes it is worth paying to keep options open: leasing rather than buying, a smaller first order, a staged commitment, a short contract at a higher rate, a deposit to hold a slot rather than a full order. These premiums can look inefficient, but when uncertainty is high they buy something valuable. Treat them as deliberate purchases, recorded as such, rather than costs to squeeze out. The risk is also the opportunity you miss article discusses separating the decision to learn from the decision to scale.
Equally, sometimes giving up an option is the right price for speed, certainty or a better deal. The point is to make that trade knowingly.
A worked example
This is an illustration. A specialty food manufacturer plans to launch a new range of chilled sauces into a supermarket chain. The plan involves several commitments over the next three months:
- ordering printed packaging film, with a 12-week lead time and a minimum order of about $25,000;
- buying a second filling machine, about $85,000;
- hiring two production staff;
- confirming the launch date with the supermarket buyer;
- leasing an additional cool room, on a two-year lease.
The owner maps when each option closes. The packaging order is the first irreversible step, needed in two weeks to meet the launch date, and it commits the design and quantities. The filling machine order closes next. The launch date, once confirmed to the buyer, closes the relationship option. Together with the cool room lease deposit and recruitment costs, around $120,000 of commitments become hard to reverse within six weeks.
The main uncertainty is whether consumers will buy the product at the planned price. The owner considers a small trial: producing a short run in existing equipment with plain labels and selling it through two independent grocers and a farmers’ market for four weeks, at a cost of about $9,000 including the delay.
Applying the five questions: the consequence is large; the trial would reduce the key uncertainty; a weak result would change the plan by reducing the range and delaying the machine; and the trial can finish before the machine order, though not before the packaging order. The owner estimates roughly a 30% chance the trial would show weak demand. In that case, the business would avoid most of the $120,000 commitment, so the trial’s value is in the order of 0.3 × $120,000 = $36,000, well above its cost.
The owner reorders the steps. The supermarket launch date is moved four weeks later, which the buyer accepts. Packaging is ordered in a smaller first run at a higher unit price, keeping the design flexible. The trial runs first. Its results show strong demand for two of the four sauces and weak demand for the other two. The range is cut to two products, the filling machine is replaced by a cheaper upgrade to the existing machine, and the cool room lease is deferred. The owner records each decision, what it closed and what it bought.
How this applies to a small Australian business
Small businesses often make commitments in quick succession, especially when launching something new. Practical steps:
- Ask which decision any proposed study or trial will change.
- Use the five questions before paying for information.
- Place tests before irreversible steps, not after.
- Set a stopping rule for analysis.
- Map when each option closes: physical, contractual and relationship.
- Add “what this makes impossible” to significant approvals.
- Pay for flexibility deliberately when uncertainty is high.
- Check contract exit terms before signing, and take advice on significant ones.
The where strategy is really decided article covers matching scrutiny to how reversible a decision is.
Signals worth watching
- Reports growing with no decision linked to the new information.
- Critical tests scheduled after contracts or equipment are committed.
- Long-lead orders placed early with no record of what they commit the business to.
- Public announcements made before key uncertainties are resolved.
- Nobody able to say when the last chance to change course was.
- Analysis continuing long after it stopped changing anyone’s view.
Common mistakes
- Gathering information that could not change the decision.
- Testing too late to influence the key commitments.
- Treating delay as free.
- Assuming reversibility declines smoothly.
- Watching only contractual exit while physical or relationship options close.
- Treating flexibility premiums as waste rather than deliberate purchases.
Frequently asked questions
How do we estimate the value of a trial without complex maths? A rough calculation is often enough: the chance the trial changes the decision, multiplied by the cost it would help avoid or the gain it would create. If that is clearly larger than the trial’s cost, it is probably worth doing.
What if the market will not wait for us to test? Then consider a faster, narrower test aimed at the single most important uncertainty, or a commitment structured to stay partly reversible, such as a smaller first order.
Isn’t keeping a register of reversibility too much work? For small decisions, yes. For significant initiatives, a one-page list of the main commitments and when each closes takes an hour or two and often changes the order of steps.
Should we always delay irreversible steps? No. Sometimes committing early secures a better price, a scarce slot or a customer relationship. Make the trade deliberately and record what was gained.
Who should own the register? The person responsible for the initiative, with the owner reviewing it before each significant approval.
What if the test results are mixed? Decide in advance what result would lead to which action, including a middle outcome such as a smaller launch or a second, narrower test. Mixed results are common, and a pre-agreed response stops them being read as whatever people hoped for.
Questions to ask
- Which decision will this study or trial change?
- Will the information arrive before the next irreversible step?
- Which of our upcoming commitments closes the most options?
- When does each option close: physically, contractually and with customers?
- What does this approval make impossible?
- Where would paying for flexibility be worth it?
Bringing it together
Information is worth gathering only if it could change a significant decision in time. Ask which decision each test will change, place learning before the point of no return and stop analysing once more evidence would not change your mind. Recognise that the ability to change course runs out in steps, tied to particular orders, contracts and announcements, and keep a simple record of what each commitment makes impossible. Pay for flexibility where uncertainty is high, and give it up knowingly where speed or certainty is worth more.
Source: KEVOS notes, drawing on teaching material on decision analysis, the value of information and the declining ability to influence projects over their life. Examples and figures in this article are illustrations.
