Many tasks stay open because nobody defined what complete means. One person thinks the analysis needs a recommendation; another expects raw data and three options. The work expands through revisions, while the task title continues to say prepare analysis. A useful definition of done closes that interpretation gap before effort becomes expensive.
Completion criteria do not require a formal specification. For everyday work, a short finish-line statement can identify the artifact, required evidence, decision owner, and acceptable quality. That statement gives the maker permission to stop, gives the reviewer a fair basis for feedback, and makes handoffs considerably faster.
Describe the observable result
Replace activity language with an object or condition that another person can inspect. Research vendors becomes a comparison of three qualified vendors with price, lead time, warranty, and a recommendation. Follow up with client becomes a recorded answer or a dated next-contact commitment. Observable results remove the need to guess whether motion equals completion.
Include the intended audience and use. A one-page briefing for an executive decision differs from a technical record for future maintenance, even when the topic is identical. Naming the reader forces useful choices about depth, format, and terminology. It also prevents a reviewer from requesting an entirely different product at the end.
Separate required quality from optional polish
Write two lists: must be true and nice if time allows. Required criteria may include accurate totals, source links, accessibility, or legal approval. Optional polish may include custom illustrations or extra scenarios. This distinction protects quality while stopping perfectionism from consuming time that belongs to other commitments.
Quality should match the decision risk. A reversible internal experiment may need clear assumptions and a working test. A customer contract or safety procedure needs stronger review and traceability. Do not use one universal standard called excellent. Define the checks appropriate to the consequence of being wrong.
Name who accepts the work
A finish line is incomplete without an acceptance path. Identify the reviewer, the decision they are making, and when feedback is due. If several people comment, name one person who resolves conflicts. Otherwise, each reviewer can keep the task open by adding preferences that were never part of the original need.
Agree on the review method. A document may require comments by a deadline; a design may need a fifteen-minute walkthrough; a data change may need a test result. The method should make acceptance visible. Silence is not reliable approval unless everyone explicitly agrees that no response by a stated time means accepted.
Change the finish line openly
New information can make the original criteria obsolete. When that happens, update the finish line rather than quietly adding work. State what changed, why it matters, and what it does to time or scope. This keeps the task trustworthy and gives stakeholders a real choice instead of presenting extra work as unavoidable.
After completion, compare the result with the criteria. If a requirement was misunderstood, improve the wording for similar work. If optional polish repeatedly proves necessary, promote it to a required check. Definitions of done become valuable organizational memory when teams refine them instead of recreating expectations for every task.
Create a four-line finish contract
- Choose one active task and write the observable artifact or changed condition that will prove it is complete.
- List three must-be-true checks and no more than two optional improvements, matching the standard to the risk of error.
- Name the accepting reviewer, the decision they will make, the feedback method, and the review deadline.
- Send the finish contract before doing more work and ask the reviewer to correct any missing expectation now.
Common questions
Is a definition of done excessive for a small task?
It can be one sentence. The smaller the task, the lighter the statement should be, but even a brief observable finish prevents unnecessary checking and repeated clarification.
What if the reviewer refuses to define expectations?
Propose a reasonable finish line yourself and ask them to amend it. A concrete draft is easier to react to than an open question, and their response creates a usable record.
