# Review effort estimates against their assumptions

Illustrative planning brief; no automatic product import.

Illustrative case: an AI proposal calls a reporting export change “small,” but the team has not checked the data boundary or reviewer availability.

## Decision

Investigate the unresolved data boundary before accepting the effort category for planning.

## Owned work

- Check the scoped outcome
  - Owner role: Task owner
  - Acceptance evidence: The estimate refers to the agreed export behavior and exclusions.
- Expose unknown work
  - Owner role: Engineer
  - Acceptance evidence: The brief lists the data-boundary question and any implementation path that depends on it.
- Review the planning assumption
  - Owner role: Sprint lead
  - Acceptance evidence: The owner confirms whether the estimate is usable or needs a bounded investigation.

## Workflow

1. Compare the estimate with the actual acceptance criteria and owner context.
2. List unknowns that could materially change the amount of work.
3. Agree whether to estimate a range, investigate first or reduce the scope.

## Judgment

AI estimates are proposals. Do not derive individual performance ratings or guaranteed delivery dates from them.


## Filled illustrative decision

The invented export brief first appears small, but the current dataset omits timezone boundaries. The team records that uncertainty and schedules a bounded inspection before assigning an effort category. The proposal stays uncommitted until the data boundary and reviewer availability are understood. The published estimate uses a broad range justified by the available evidence; extra calculated digits remain detail, not a more precise commitment.


## Workflow questions

### Is a range better than one number?

A range can express uncertainty if the team explains its assumptions. It still needs scope and owner review.

### What should we do with missing information?

Identify the smallest investigation that can change the planning decision, then review the estimate again with that evidence.

### How precise should the published estimate be?

Match the reported precision to the input evidence. Hour-level observations do not support a promise stated to the second. Keep raw calculations available for inspection and present an appropriately rounded range with its assumptions.

## Product connection

Use the scope and uncertainty notes to review an AI-assisted planning proposal before accepting an effort estimate.

Confirm account availability before adopting this manual outline.