Cognitive Biases · Growth Library
Calibrate Task Estimates With Planned vs Actual Work
Use completed work to update the next comparable estimate.
Try this
Write the deliverable and important assumptions, make an estimate that includes the work you currently know about, complete or reassess the task, record the actual conditions and the main reason for any difference, then consult that record before estimating similar work again.
Check-in: What did I estimate, what actually happened, what caused the largest difference, and what will I change in the next comparable estimate?
Evidence: Practical guidance; no evidence-like claim is detected in the current source record.
Review note: The inherited article contained unsupported authority language, arbitrary milestone ranges, fixed buffer percentages, invented timing rules and claimed calibration improvements. The effective public record is narrowed to a local planned-versus-actual feedback loop for comparable work without presenting a universal estimation formula.
An estimate becomes more useful when it is connected to a clearly defined deliverable and later compared with what completion actually required. The purpose of this protocol is not to find a universal buffer. It is to build a small local record that makes the next comparable estimate easier to explain and revise.
1. Define what the estimate covers
Name the deliverable, the current scope, important inputs and any dependency already known. Replace vague labels such as work on report with an observable state such as draft ready for review. If the scope is unclear, note that uncertainty rather than hiding it inside a precise-looking estimate.
2. Make the estimate from what you know now
Consider the work steps and waiting points that are already visible. When a comparable completed task exists, use it as context. Keep assumptions beside the estimate so a later review can distinguish a poor estimate from a task whose scope or dependencies changed.
3. Record the actual completion conditions
When the task finishes or is materially replanned, capture what happened. Note the completion time or effort when available and the main source of difference: extra review, missing input, rework, interruption, unfamiliar work, waiting, or a scope change.
4. Update the next comparable estimate
Before estimating similar work, inspect the relevant completed records. Adjust the plan for the recurring difference you can actually see, and keep exceptional one-off events separate so they do not become an automatic rule for every future task.
Questions to consider
Should I always add the same buffer?
No. A fixed buffer can hide very different causes of delay. Record the reason for the difference and adjust future planning according to the type of work, known dependencies and your own comparable history.
What if the task changes after I estimate it?
Record the scope change separately. An estimate for one deliverable should not be judged against a larger deliverable created later. The change itself is useful planning information for the next similar task.
What should I record besides time?
Capture the assumption or dependency that mattered most: missing input, review cycles, unfamiliar work, interruptions, waiting for another person, rework, or a changed scope. Keep the record small enough that you will actually reuse it.
Maintenance record
Review history
Current status: Reviewed. An editorial or evidence review is recorded and no later event changes that conclusion.
2026-09-10 · reviewed · Brali claim cleanup agent Source: protocol-content-overrides-claim-cleanup-batch-6.json
Replace unsupported planning-fallacy authority language, universal milestone sizes, fixed correction percentages, invented quick-task timing rules and claimed calibration gains with a bounded planned-versus-actual feedback loop.
Browse the review ledger · Challenge or update this hack
Implementation observations can trigger a review, but they cannot change an evidence conclusion by themselves. Commercial relationships do not control review status or outcomes.
Continue from here
Choose the next useful path.
Article versions
Brali keeps substantive article history visible. Later reviews may refresh wording, sources, examples, or boundaries without silently replacing the record.
- Version 2025-10-06: first recorded publication in the migrated Brali corpus.
- Evidence review 2026-09-10: reviewed by Brali claim cleanup agent; evidence state is
practical.
Canonical source record · Evidence state: practical.