A team writes “Improve customer support this quarter.” One person works on faster replies, another rewrites the help centre, and the manager expects fewer repeat tickets. All three are doing reasonable work against different definitions of success. A SMART goal makes those definitions visible while there is still time to agree on them.
Where SMART came from

George T. Doran introduced SMART in a 1981 management article. His version used Specific, Measurable, Assignable, Realistic, and Time-related. He also treated the objective and its action plan as a pair. This guide uses the familiar modern version, Specific, Measurable, Achievable, Relevant, and Time-bound, and keeps Doran’s question of ownership because it matters when several people share the work. SMART is a writing checklist. The evidence on goal difficulty, commitment, and feedback comes from a separate body of goal-setting research. [1] [2]
S. Specific means defining the scope
Say whose experience should change and which part of it you intend to improve. “Improve customer support” becomes “shorten the wait for a first human reply on weekday billing tickets.” This narrows the work to a queue the team can examine and gives the reader a way to tell which cases belong in the calculation. State who owns the result and who can change the workflow. An owner can coordinate a goal even when other people carry out parts of it.
M. Measurable means deciding what evidence counts
Start with a baseline. Suppose the median first human reply took 18 business hours last month. Define what counts as a human reply, how weekends and holidays are handled, and what period will be used to measure the result. If automated acknowledgements count as replies, the score can improve while customers wait just as long. When sample sizes are small, show how many tickets were included before interpreting a change.
A single measure can also hide poor outcomes. Suppose quick but incomplete replies cause customers to reopen tickets. Pair the response-time target with a quality check, such as keeping the seven-day reopen rate at or below its 8% baseline. For a skill or creative project, evidence can be a rubric or an observable demonstration. The test is whether people reviewing the same work can explain why it meets the standard.
A. Achievable means checking the constraints
Estimate what the proposed change requires and compare it with actual capacity. The support team might test two daily queue reviews using its current rota, then look at the first week of response data before committing to the full target. If a software change requires approval outside the team, put the approval date in the plan. A hard goal can focus effort when people have the skills, commitment, and feedback to pursue it. An unfamiliar task may first need a goal about finding an effective strategy. [2] [3]
R. Relevant means connecting the target to a problem
Talk to the people affected or examine a sample of recent cases. If customers are mainly frustrated by having to repeat their billing details, faster first replies may be a secondary improvement. If most complaints are about waiting without help, the response-time target has a clearer reason to exist. Write that reason in one line. It gives the team something to refer to when this work competes with other priorities.
T. Time-bound means scheduling a decision
Give the target a deadline and schedule an earlier review while you can still act. In this example the team checks after two weeks whether the queue reviews are happening, whether reply times are moving, and whether reopens are rising. A final date tells people when the result will be assessed. An earlier date creates space to revise the method or stop a change that is making service worse.
A worked example for a team
Over the next six weeks, reduce the median time to first human reply on weekday billing tickets from 18 to 12 business hours, while keeping the seven-day reopen rate at or below its 8% baseline. The support lead will pilot two queue reviews a day and review the results with the team after week two.
The team tracks queue reviews as an action it can control, reply time as the outcome it wants, and reopens as a quality check. Keep the definitions beside the target. If volume changes sharply, examine both the baseline and the current period before deciding whether the workflow caused the change.
A worked example for an individual
Over the next four weeks, draft one portfolio case study and ask two readers to review it. By the final Friday, revise it so a reader can identify the problem, my contribution, and the evidence behind the result without a separate explanation.
Here the important measure is reader comprehension. Two reviews create an opportunity to get feedback, but simply collecting both comments does not meet the quality criterion. Book the first reader before writing the final draft, so a scheduling delay leaves room to adjust.
When to try a learning goal first
When you have a workable approach, SMART helps define how far to take it. When you are still finding out how the work is done, make the first deliverable evidence about the method. Before promising a response-time improvement, for example, sample recent tickets and identify the two longest sources of delay. For a portfolio, show one rough case study to a reader before committing to a full site. Seijts and Latham distinguish learning goals that direct attention to finding strategies from performance goals that direct attention to achieving a result. [3]
To compare a measurable finish line with an ongoing practice, see how to choose between SMART and PACT.
References
- Doran (1981). There’s a S.M.A.R.T. way to write management’s goals and objectives.
- Locke & Latham (2002). Building a practically useful theory of goal setting and task motivation.
- Seijts & Latham (2005). Learning versus performance goals.
Further reading
Read the original guide from Atlassian.

