Label the status of every input
An observed number has a source and a reporting window. A modelled number has a formula and chosen range. An inference has a reasoning path. An unknown has a next test or an explicit decision to tolerate the gap. Mixing those in one cell makes the final recommendation look more certain than the underlying evidence. Preserve the labels when the values move into a slide or Market record.
- Tag each input as observed, modelled, inferred or unknown.
- For modelled values, state the formula and the range that materially changes the decision.
- Show the decision once with conservative and once with optimistic assumptions.
Evidence classification matrix
| Decision field | Record or rule |
|---|---|
| Fact | directly observed with source and date. |
| Assumption | a chosen value used to explore a scenario. |
| Unknown | missing evidence that could overturn the verdict. |
- 01Fact
- 02Assumption
- 03Unknown
Run a meaningful sensitivity check
Ask which assumption can reverse the decision. Change that input across a plausible range while holding the observations fixed. The public Serbia Finance Launch Case sample treats commercial sensitivity as a planning device and distinguishes selected from rejected scenarios. Use the same discipline for a new entry decision: show why a route survives the range, or state that evidence is insufficient to select it.
Keep assumptions editable without changing history
Suppose a model tests whether a maintained Market could justify its operating cost. Observed profile clicks remain fixed to the reported GSC window. Estimated buyer conversion, service price and maintenance cost are scenario inputs. A reviewer can change one assumption and inspect the decision threshold without rewriting the historical click count. Unknown downstream intent stays unknown, even if the model requires a placeholder range. The separation lets a team discuss a commercial decision honestly while still using imperfect early evidence.
A forecast remains a scenario
A forecast is a planning tool. Do not publish its output as actual demand, traffic, revenue or a guaranteed Market result.
References and source notes
Original 1000&1 methodology reviewed and approved by the author. Public claims and historical measurements are bounded by the linked sources and dates.