Learn
How to test the assumptions a plan depends on
Every plan rests on beliefs that have not yet been proven: that customers will pay, that a cost will hold, that a key hire can be made by spring. Assumption testing names those beliefs, ranks them by how much the plan depends on each, and tests the most dangerous ones before money is committed.
Discovery-driven planning
In 1995 Rita McGrath and Ian MacMillan described an approach in the Harvard Business Review called discovery-driven planning. New ventures, they observed, are planned as though their assumptions were facts. They proposed starting instead from the profit required and working back to what must be true to earn it, with each requirement tested at a checkpoint before the next round of spending.
Ranking what you believe
List every assumption, then rate each on two questions: how much would the plan suffer if this were false, and how sure are we that it is true? Those rated high on harm and low on certainty are the load-bearing unknowns. Test those first, and leave the rest until they matter.
A plan typically carries a dozen assumptions, of which two or three decide its fate.
Designing a cheap test
A good test is the smallest action that could prove the assumption false. To learn whether customers will pay, ask a few to pay in advance. To check a cost, request a written quote. To check a timeline, talk to someone who has done the same work recently.
Before running the test, write down the result that would count as failure, so the bar cannot move once the answer arrives.
The common mistake
Testing the easy assumptions and leaving the sacred ones alone. Teams check what is simple to check, report progress, and never touch the belief the whole plan was built around, often because it belongs to someone senior. The second mistake is setting the pass mark after the result is in.
Assumptions inside a Decision Room
Argumont holds an assumption as a claim that has yet to be checked, never as a fact. The Scout looks for public evidence that bears on it, the Inspector marks it supported, partly supported, unverified, or contradicted, and the Challenger presses on the one the leading option depends on most. The Verdict names the assumptions that could change the result, so the Human Decision is made knowing which beliefs still carry the weight.
The cost of serving a customer
A subscription service plans to charge $20 a month, assuming it costs $6 a month to serve each customer. At that cost, 2,000 customers yield $28,000 a month in gross profit. The team ranks this assumption first, because if the true cost reached $14, gross profit would fall to $12,000 and no longer cover $15,000 of monthly overheads. A month's pilot with 80 customers shows an average cost of $9, with the heaviest tenth of users costing $31 each. The plan survives at $22,000 a month, and the test also shows that a usage limit for the heaviest users is worth designing now.
Questions
- What is the difference between an assumption and a fact?
- A fact has been checked against a source that could have shown it to be wrong. An assumption has not, however confidently it is stated.
- How many assumptions should I test?
- Test the two or three that combine high harm if false with low certainty. Testing everything is slow, and it spreads attention across beliefs that do not decide the outcome.
Further reading
Bring Argumont a Decision worth challenging.
Start with a paid 60 day pilot on your own real Decisions.