Give each assumption a data type, source, date, owner and affected decision. Preserve earlier values when updating so the team can distinguish new evidence from a new management choice.
A forecast may calculate correctly while nobody knows where the rates or sold nights came from. An assumptions register checks input provenance before calculation. It supplies no universal input values or investment assurance.
Separate three kinds of input
An observation is something checked or recorded within a stated scope. An estimate represents something not yet known. A management choice is an intended course of action, such as a service model or planned start.
All three can inform a plan, but should not be relabelled. A target rate is not an achieved selling rate. A competitor's advertised rate is not evidence of the revenue it received.
Use a traceable register
| Field | Record |
|---|---|
| Assumption reference | Link to the calculation or decision |
| Item and unit | Per night, room, month or another explicit unit |
| Type | Observation, estimate or choice |
| Value and coverage | Consistent period and basis |
| Source and date | Evidence that can be reopened |
| Uncertainty | What the evidence does not cover |
| Owner | Person gathering or reviewing information |
| Decision impact | What changes if the assumption is wrong |
| Review trigger | New evidence or changed conditions |
Use unknown or pending where appropriate. Do not insert zero simply to keep a formula running without alerting the reader.
A fictional rate example
An owner chooses a desired rate while the team captures nearby advertised prices for selected dates. Keep two records: a management choice and a dated observation. Note room and condition differences. They do not combine into a verified market rate.
When actual bookings become available, add their period and scope and assess whether they justify changing the assumption. Retain the earlier target so the original decision remains understandable.
Prioritise consequential uncertainty
Investigate inputs that would materially alter investment or operating decisions if wrong, such as sellable inventory or omitted costs. Assign evidence collection rather than relying only on confidence scores that different people interpret differently.
Explain confidence through evidence coverage. A plausible-looking number is not verified. If older information must be used, retain its original date and limitations.
Update with a history
Preserve previous value, new value, reason, date and the person checking downstream effects. A strategy change is a new choice, not newly discovered market evidence.
Before a decision meeting, surface important unresolved assumptions and their owners. A change in units or date coverage also requires checking the calculations that use them.
Before using the plan
- Every figure has a unit and period.
- Targets remain distinct from actuals.
- Evidence can be reopened.
- Missing data is not silently zero.
- Consequential uncertainties have owners.
- Earlier versions remain available.
Frequently asked questions
Can I plan without actual data?
Yes, as an explicitly hypothetical plan. State which evidence must be obtained before commitments rather than calling the plan verified.
Must every assumption be updated daily?
Use a rhythm suited to the decision and the information's volatility. Stable and changing inputs need not share one schedule.
Check inputs before feasibility calculations
Use the small-hotel feasibility and break-even guide after checking input provenance.
Ask LINE OA @kines about current Hotelpreneur course details, dates and format if you want to organise the assessment process.
Tool provenance
This Kines editorial register adapts forecasting and actual-versus-plan review concepts. It is not an audit or certification of user-entered data.


