Objectives you can verify: from “better” to a threshold
Echipa HR 365 · reviewed 2026-09-07 · 5 min read
The test of an objective is simple and unforgiving: if at the end you and the person could reach different conclusions about whether it was met, it is not an objective. Almost every formulation of the type “improving”, “raising quality”, “better collaboration” fails the test.
Why the usual formulations fail
Not out of laziness, but because they are written so as not to upset anyone. A precise objective can be visibly missed, and that is uncomfortable for both sides — so a formulation that can be read either way is preferred. The price is that at the end the discussion is about interpretation rather than result.
The second cause is more practical: nobody knows what can be measured. For many roles there is no ready-made indicator, and in its absence an intention gets written. The solution is not to invent a number, but to describe a piece of evidence.
The method, in four steps
Applied to any vague formulation
- 1 — Ask: if this happened, what would you see differently? The answer is usually the real objective.
- 2 — Choose the evidence: a number if one exists, otherwise an artefact — a document, a list, a piece of work completed.
- 3 — Set a threshold and a date. Without a threshold, any progress can be declared sufficient or insufficient.
- 4 — Check whether the person can influence it alone. If it depends on somebody else, it is somebody else’s objective.
Eight examples, before and after
| Before | After |
|---|---|
| Improve communication with the team | Announces any delivery delay in writing before the deadline, on every piece of work this quarter |
| Raise the quality of the work | Fewer than three pieces of work returned for corrections, in the quarter |
| Develop competencies | Runs the monthly close alone, without review, from month three |
| More involvement | Proposes and completes one improvement to a process they use daily |
| Support new colleagues | Trains one new colleague to the point where they work unsupervised on X |
| Meet deadlines | All work delivered on time or renegotiated at least 2 days in advance |
| Know the product | Answers client questions about X alone, without escalating |
| Better organisation | Documents the procedure for Y, used by somebody else without further explanation |
The observation repeated across all eight: the right-hand column mostly contains no number. It contains verifiable evidence — somebody was trained, a procedure was used, a piece of work was delivered. Good objectives are not necessarily numeric; they are necessarily verifiable.
How many, and over what period
Two per quarter, three per half-year. Beyond that it becomes a list nobody tracks, and at the end whatever happened anyway gets ticked off. The constraint of two forces the conversation about priorities, which is the real value of the exercise.
The period matters less than the mid-point check. A six-month objective with no conversation at three months is, in practice, addressed in the last fortnight — or not addressed at all.
Do I set objectives for repetitive operational roles too?
Yes, but differently: not on output, which is set by the volume of work, but on something the person can change — a documented procedure, a colleague trained, a recurring error eliminated.
What about objectives that depend on external context?
You reformulate them around what the person controls: not “sign five contracts”, but “take five conversations through to an offer”. The result depends on the market too; the effort and the quality do not.
Do I link objectives to a bonus?
If you do, objectives will be negotiated downwards, systematically — that is a rational response, not bad faith. Many companies that work well keep the link, but with a separate conversation and a margin of judgement.
Where to start
Take one person’s current objectives and apply the test: could you reach different conclusions? The ones that fail get rewritten with the question “what would you see differently”.
Then apply step four to each. Usually at least one depends on somebody else — and that is the one that will produce the unpleasant conversation at the end.
Who writes them
The version that works most often: the person proposes, the manager adjusts. Not out of democracy, but because whoever does the work knows best what can change in it — and an objective they proposed has a better chance of being taken seriously.
What does not work: objectives handed down from above, identical for everyone in a department. In practice they become a formality each person interprets as they like, and at review time nobody refers to them.
What you do when an objective is missed
| What happened | What it means | What you do |
|---|---|---|
| It was not a priority for either of you | The objective was not, in fact, important | You drop it rather than rolling it over |
| The context changed | Normal, it happens often | You rewrite it at the mid-point check, not at the end |
| It depended on somebody else | Step four was not applied | You reformulate it around what they control |
| It was too large | Ambition, not planning | You cut it to a piece that fits the period |
The first row is the most instructive and the most avoided. An objective nobody tracked for six months was not important; rolling it over automatically only moves the problem, and by the third rollover it becomes an internal joke.
Objectives with a mid-point check, tracked across the year
Objectives and their evidence sit alongside the review, with the mid-point check scheduled — so the final conversation starts from facts rather than interpretations.
Free account, every module for 7 days, no card required.