SMART goals
Turn an ambiguous goal into a specific outcome with a measure, a feasibility check, a purpose, and a deadline.
Use SMART when a team agrees on a direction but cannot yet tell what success would look like. Review the goal against five criteria: Specific, Measurable, Achievable, Relevant, and Time-bound. The worksheet helps expose gaps in the plan before people commit to a target.
A proposed outcome, its population and scope, the current baseline, a measurement definition, available resources, and delivery and review dates.
An editable five-part goal worksheet that keeps the baseline, target, feasibility assumptions, and review timing visible.
How to use it
- Name the people or process affected and the observable change you want. Replace broad intentions such as improve activation with a bounded outcome.
- Define the metric, numerator, denominator, observation window, and source. Write the baseline and target separately; keep the definition fixed when comparing them.
- Check whether the team has the time, capacity, and supporting evidence to pursue the target. Mark missing commitments as unconfirmed rather than calling the goal achievable.
- Explain how this outcome serves the wider objective. Add guardrails so improving one metric does not hide harm elsewhere.
- Set a delivery deadline and review points. Allow each cohort its full observation window, then compare actual results with the target before deciding what to change.
Keep its limits in mind
A well-written goal is not evidence that it is feasible or achieved. Check resource assumptions and measure actual results separately. Completing all five sections is not a pass score; keep uncertain assumptions explicit.
Make an activation goal reviewable
Fictional example. Yunqiao has researched onboarding and is preparing a four-week goal. Its current seven-day activation rate is 30%. About 50 new teams join each week, or roughly 200 over four weeks. These are planning volumes, not evidence of statistical power or an achieved improvement.
On a small screen, scroll across the diagram to read it.
A target, not a result: The proposed increase is 10 percentage points. The case supplies neither a post-release result nor proof that a team can deliver that increase.
Delivery and measurement have different clocks: About 200 teams may join over four weeks, but their exposure to the new guidance may differ. Record release exposure and wait for complete seven-day windows before comparing cohorts.
Next step: Confirm ownership, resources, and the metric definition, then build a weekly cohort view that includes release exposure, invitation failures, and complaints.
Three reasons a polished goal is not ready for commitment
Use SMART when a direction is agreed but people cannot tell what progress would look like. Start from the actual work problem and the decision the goal should support. A goal can be specific and measurable while still being irrelevant or beyond available capacity. The five prompts expose missing agreements; they do not award a quality score.
Pause if the baseline has no recoverable definition or source. A target rate is hard to interpret when nobody knows which events counted. Pause if the proposed owner, resources or necessary permissions are unconfirmed. A named role in a sentence is not an accepted assignment. Pause if the deadline ends before the outcome can be observed. Delivery of a feature and completion of its measurement window can require different dates.
Avoid making a weak goal look precise by adding unsupported numbers. Record why a target is proposed, what remains uncertain and what would justify revising it. A stretch target may guide a discussion without being a forecast. Keep relevant safeguards alongside the main metric so a team cannot claim success by improving one number while making the customer experience worse.
Choose the missing piece of the agreement
SMART helps formulate a target; other methods may be needed to define a metric, explore a person’s own choice or review an experiment. Use the smallest next step that resolves the uncertainty.
| Method | Input | Output and boundary |
|---|---|---|
| SMART | A direction, baseline, constraints and proposed horizon | A checkable goal draft; feasibility remains to be confirmed |
| KPI tree | Defined quantities with compatible populations and units | A reproducible metric structure; no target approval |
| GROW | A person’s goal, current reality and options | A chosen next action; no choice on their behalf |
| PDCA | A bounded change and an observation plan | A learning cycle; completion alone does not prove improvement |
Worked continuation: separate the four-week plan from the evidence
Return to the fictional Yunqiao onboarding case above. The supplied baseline is seven-day activation of 30%; the proposed target is 40%. Around 50 new teams may join each week, or roughly 200 in four weeks. These are planning volumes, not an observed baseline denominator. Do not silently rewrite the baseline as “60 of 200” or claim that 200 teams will receive the revised guidance.
The work concerns the first invitation and a collaborative task. However, the exact qualifying activation event, eligible-team definition, source and exclusions still need documenting. Interviews and prototype findings support exploring the proposal, not a guarantee of delivery or a statistically conclusive result. Invitation failures and complaints remain relevant safeguards; no actual post-release result is available.
The draft should therefore preserve known values and missing agreements side by side. The change from 30% to 40% is ten percentage points, or about a 33.3% relative increase in the rate. That calculation describes the proposed target gap only. It is not an achieved improvement or an estimate of how many teams a feature caused to activate.
| Item | Known or proposed | Before commitment or measurement |
|---|---|---|
| Direction | Improve first-invitation and collaborative-task guidance | Confirm which work is in scope |
| Baseline / target | 30% / proposed 40% seven-day activation | Recover event, denominator, source and exclusions |
| Planning volume | About 50 teams/week; roughly 200 in four weeks | Record actual cohorts and release exposure |
| Capacity and ownership | Not confirmed | Obtain actual acceptance and resource agreement |
| Time and safeguards | Four-week delivery; mature seven-day cohorts; failures/complaints | Agree delivery and final measurement checkpoints separately |
Conclude with conditions, then review actual cohorts
A usable request is: “Please confirm the metric definition, owner and capacity for a four-week guidance change. The proposed target is 40% against a 30% baseline under the same definition; measurement follows complete seven-day windows.” Until those conditions are agreed, retain draft status. Do not substitute a confident deadline for missing delivery evidence.
Prepare a cohort ledger with the eligible start period, actual team count, guidance version seen, activation count and observation endpoint. Leave future observations empty. Review cohorts weekly only when their windows are complete; a team entering late in week four needs its full seven days. The delivery deadline does not authorize counting that incomplete outcome as a failure or quietly excluding it.
At review, report the actual numerator and denominator, safeguards and any definition changes. A comparison using different eligibility or exposure needs qualification before it is called progress. If capacity changes, keep the original target and append the revised agreement with its reason and approver. Finishing the guidance change, meeting the target and establishing its causal effect are separate claims requiring different evidence.
Check your work
- The goal names an outcome and scope, not just a list of activities.
- The baseline, target, and actual result are distinct. A change from 30% to 40% is 10 percentage points, not 10%.
- Metric definitions and cohort eligibility are explicit; incomplete observation windows are not counted as final results.
- Resources and responsibilities are confirmed or clearly marked as unresolved.
- The review includes guardrails and limitations. A before-and-after change alone does not prove the intervention caused it.
Framework references
References explain the framework and its scope; they do not validate the fictional example's results.
Put the framework to work
Copy the outline and add your question, evidence, and open questions.
Practice with an example
Switch denominators for the same cohort counts, distinguish the first task from later use, and leave with a bounded problem statement and evidence plan.
New users aren’t coming back. What should you check first? →