Assumptions
What must be true for this to work?
Why it matters
A plan combines things we know with beliefs about customers, operations, costs and timing. A decision can look robust until one belief turns out to be doing most of the work. This lens makes those dependencies visible so the team can spend its limited time testing the ones that could change the decision.
Apply the lens
- Write the proposed commitment in one sentence. Ask each decision-maker to list what must be true for it to work before discussing the list together.
- For each assumption, record the evidence, who owns it, and what observation would weaken it. Mark where the evidence came from a different setting.
- Rank assumptions by uncertainty and cost of being wrong. Decide which evidence would justify moving ahead and which findings would send you back to the options.
Questions to ask
- Which assumption would reverse our recommendation if it were false?
- Are we relying on observed behavior, a vendor claim, or a forecast?
- What evidence would make a reasonable supporter change their mind?
Worked example
In an illustrative purchase-approval pilot, a team assumes routine invoices can be identified from their history. Clean historical data may omit disputed invoices and unusual vendors. Make those exclusions explicit and separate the assumption about classification from the assumption that humans can intervene.
One test to run
Score a shadow batch that includes disputed and first-time vendors. Count how often a case requiring review would be routed automatically, then compare against a limit agreed before looking at the results.
What to capture
An assumption register, evidence review and tests that could change the decision.