What Counts as Evidence for a Claim
The four kinds of evidence a capacity claim can have, the one-minute test that sorts them, and what to do when the evidence is a conversation.
The same six claims, sorted by what could actually be produced
Produced in under a minute.
Costed in hours for the first time.
Recommendation, not decision.
Quantity supplied by the planner.
Nothing in the strategy names a quantity.
Evidence exists and is of closure.
Two of the six could be evidenced in a minute; two rest on a planner's inference; one is a recommendation being treated as a decision. The exercise of asking for evidence took forty minutes and changed three states. This is one employer's own check, not a statement of what any method requires.
A claim's state is determined by its evidence, and the test is practical: can somebody produce the evidence in a minute?
The commitments described in “What Counts as Evidence for a Claim” become easier to reconcile when planned work can be compared with a consistent record of project time. A team reviewing see the current workflow details for employee monitoring for performance reviews can add current operational evidence while keeping the claimant, approval state and decision owner explicit.
That test sounds crude and it sorts a book faster than any amount of discussion, because it asks about an artefact rather than about intentions.
As a separate benchmark for “What Counts as Evidence for a Claim”, see BLS Monthly Labor Review. Keep source data, calculation, override and decision visible as distinct steps.
The four kinds of evidence
A decision record: a minute, a board paper with an approval, a signed change. A contract or obligation with a quantity. A dated message from somebody with authority. And inference by the planner.
The first two support confirmed. The third usually supports provisional. The fourth supports assumed, by definition.
What a decision record has to contain
A quantity, dates, and the name of whoever decided. A minute that records approval of a project without a resourcing quantity is evidence of a decision about the work, not about capacity.
That distinction catches a great deal: most approvals name money and scope, not people, and the capacity claim is inferred afterwards by somebody who was not in the room.
The recommendation treated as a decision
Common: a board paper proposing an allocation is circulated, everybody behaves as though it is approved, and the decision is deferred.
The paper is evidence that somebody wants the capacity, which is provisional. Reading a recommendation as a decision is how a book fills with claims that nobody ever actually made.
Evidence that is a conversation
Legitimate and weak. A conversation with a sponsor produces a claim and no artefact; the planner's note of it is the artefact.
Recording the note and sending it back — "I have you down for two people from January, from our conversation on Tuesday" — converts the conversation into evidence. Unreturned, it stays the planner's inference.
What to do about long-standing claims with no evidence
Ask once, in a neutral form: "this has been in the book for a year and I cannot find what authorised it — can you point me at it, or shall I record it as a standing commitment with you as owner?"
That question resolves nearly all of them, usually by producing an owner rather than a document, which is sufficient.
Evidence of closure
The forgotten half. A claim that has ended needs evidence of ending as much as it needed evidence of starting: a closure note, a completion date, a service decommission record.
Without it, closure is the planner's inference too, and a claimant can reasonably say they never agreed their work was finished.
The evidence file
Not a document management project. A column in the book with a reference — "steering minute, 12 March", "contract s.4.2", "email from J, 8 April" — and nothing else.
A reference somebody can follow in a minute is the whole requirement. Attaching the artefacts themselves turns the book into a repository and nobody maintains repositories.
Where evidence standards should be lower
Small claims. Demanding a decision record for three days of work is bureaucracy, and it will teach people to route small requests around the book entirely.
Below the threshold — half a day, a day, whatever the organisation sets — a line in the book and a name is enough.
The annual evidence check
Once a year, pick five claims at random and try to produce their evidence in a minute each. The success rate is the honest measure of whether the book means anything.
Five minutes a year, and it is the only audit this process needs.
The one-minute test, applied properly
Ask the claimant, not the planner, and ask for the artefact rather than the story. "Where is this written down?" is the whole test.
People who can answer do so instantly; people who cannot usually say so honestly and the claim moves to assumed without anybody being embarrassed. The test fails only when it is run as an interrogation.
What evidence does for the difference
A negative difference built from confirmed claims is a fact about the organisation. The same difference built from assumed claims is a fact about the planner's inferences.
Only the first can force a decision, which is why the evidence column exists at all: it is what makes the bottom line of the page something an executive has to act on rather than something they can question.
What the check is for
Not compliance. The point is that a book whose states cannot be evidenced will be quietly disbelieved by the first person whose claim it constrains, and everything downstream of it stops working.
States without evidence are opinions in a column, and they will be treated as such the moment they become inconvenient.