Competitor-response assumptions fail more often than demand assumptions when plans are tested against operator interviews. Plans reliably test their own logic and rarely test whether the competitor will behave as required.
What does a planning process test?
Internal logic, thoroughly. Whether the initiatives sum to the target, whether the resourcing is consistent, whether the phasing works. Most planning processes are very good at this.
What they test lightly is whether the assumptions underneath the arithmetic hold, because those live outside the organization and nobody in the room owns them.
The asymmetry is structural rather than cultural. Internal logic is checkable by the people in the room using material they already have, so it gets checked. An assumption about how a competitor will respond requires evidence from outside the organization, which somebody has to commission, which means somebody has to own it — and in most planning processes nobody does.
Which assumptions fail most often?
Assumptions about competitor response. Plans routinely assume competitors continue as they are while the company changes, which is a convenient simplification that survives because nobody — not even the competitive intelligence function — is accountable for it.
Demand assumptions fail less often, partly because they are more visible and partly because they are the ones the process does interrogate.
The second most common failure is an assumption about how long something takes. Plans routinely assume a regulatory approval, a channel partnership or a hire lands on the schedule the plan needs rather than on the schedule the market runs to. Those assumptions are testable in advance and almost never tested, because a timeline reads as a planning detail rather than as a load-bearing claim.
How do you test a strategic assumption?
Give it to someone who has watched that market behave. Former operators at competitors and in the channel can describe what happened the last time a company attempted the same move, which is the closest available thing to a test.
That is a different exercise from market research. You are not sizing an opportunity; you are asking whether a specific action will provoke a specific reaction.
The question has to be specific enough to be answerable. Asking an operator whether a market is attractive produces an opinion; asking what happened when their former employer cut price by a comparable amount in a comparable segment produces an account. The second is evidence about a mechanism, and it transfers to your situation in a way the first does not.
Who should challenge the plan?
Someone with no stake in it. Internal challenge is bounded by the fact that everyone in the room will be measured against the plan, and that is a structural limit rather than a failure of nerve.
The correction is not more rigorous internal review but a different source of challenge, which is what an outside operator brought into the offsite provides.
Where outside challenge is not available, the next best correction is to make the challenge someone's explicit job. Assigning a named person to argue the case against, with time to prepare and no obligation to be constructive, produces more than a general invitation to raise concerns. It is weaker than genuine outside challenge and considerably better than nothing.
What should reach the board?
The two or three assumptions the plan is most sensitive to, with what was done to test them and what was found. A board pack presenting a plan without naming its load-bearing assumptions is presenting a conclusion, which is also how a scenario plan stops being defensible.
Boards generally accept a named uncertainty and react badly to discovering an unnamed one during questions.
Naming assumptions also changes what happens after the plan is approved. An assumption that has been written down can be reviewed when the year turns out differently, which is how a planning process improves. One that was never named cannot be revisited, so the same optimism reappears in the next cycle wearing different numbers.