SMART Goals: What Survives Contact With Real Work

TL;DR

• The core decision: what the goal should be about. SMART is silent on that, and it's the only question that determines whether the goal is worth having.
• When doing nothing is right: when people already know what they're meant to achieve and could each tell you the same three things.
• What has to be true: somebody can say why this goal rather than another one, before the wording is checked against anything.
• How the options split: by whether the format is used to choose the goal or only to check how it's written.
• Decision rule: choose first, format second. A goal that passes every letter and describes the wrong work is a well-written mistake.
• Outcome to expect: fewer goals, some of them awkward to measure, and a much closer match to what people actually spend their time on.

Five Goals That Pass and Mean Nothing

A manager is asked to set goals for their team. The template says SMART, so they work through it. Specific: name the thing. Measurable: attach a number. Achievable: make sure it's possible. Relevant: connect it to something above. Time-bound: put a date on it.

An hour later there are five goals, and every one passes. They're specific, they have numbers, the dates are sensible. The manager reads them back and has a quiet, hard-to-place feeling that something has gone wrong, because none of these five things is what the team is actually going to spend the year doing. The real work is messier, mostly reactive, and only partly countable, and none of it fitted the format.

So the goals get submitted anyway, because the deadline is Friday and they satisfy the requirement. Nine months later nobody has looked at them. At review time they're either quietly dropped or retrofitted with whatever evidence can be assembled. Everybody involved concludes that goal-setting is a bureaucratic exercise, which, in the form they experienced it, it was.

Recommended Tools

Best tools for Performance Management

B BambooHR HR management system L Lattice Performance management D Deel Global payroll & EOR

View all recommended tools

The real problem isn't that SMART is a bad acronym. It's that it answers the wrong question. SMART is a checklist for how a goal is worded. It has nothing whatsoever to say about whether the goal is the right one, and applying it first means the format has quietly chosen the content before anybody asked what mattered.

When You Genuinely Do Not Need to Act Yet

Your current setup is genuinely fine. Ask three people on your team what they're trying to achieve this quarter and you get three clear, consistent answers that match what they're actually working on. If that's true, whatever you're doing works, and introducing a formal goal process will add paperwork to something already functioning.

Friction is starting to show. Two people have made opposite calls about what to prioritise, or somebody has spent a fortnight on work nobody wanted. That's a clarity problem rather than a goal-format problem. The cheapest check is to ask each person what they'd drop if they had to lose a day a week, and see whether the answers match what you'd have said.

It has become a real cost. Work is being done that nobody values, or you're arbitrating priority disputes weekly, or somebody's review was difficult because neither of you could say what they'd been measured against. At this point a written statement of what matters is worth the effort, and the wording is the least important part of it.

The edge case that forces it. Goals are being tied to pay, promotion or continued employment. That changes what a goal is: it stops being a shared understanding of priorities and becomes a term of an arrangement somebody is judged against. Where goals affect pay or employment decisions, requirements around fairness, documentation and discrimination differ by jurisdiction, so establish what applies where you operate and take local advice before you formalise anything.

Five Questions This Reader Asks at 11pm

What do the letters actually mean? Specific, measurable, achievable, relevant and time-bound, with some variation in the words people use for the last two. The definitions are less interesting than what they're for: the whole acronym is a test applied to a sentence, checking whether the sentence is clear enough to be acted on and assessed. It's a wording test, and it's quite good at that.

Do SMART goals actually work? They work for the job they do, which is preventing a specific failure: the goal so vague that nobody can tell whether it happened. That failure is real and common, so the acronym earns its place. What it can't do is tell you the goal was worth setting, and that's the failure most goal exercises actually suffer from.

How many goals should somebody have? Fewer than you'd think, and the constraint is attention rather than capacity. Three is usually plenty and five is an upper limit. A list of eight isn't a set of priorities, it's a description of a job, and a person facing eight will work on whichever is loudest, which means the list decided nothing.

How do I measure work that isn't countable? By measuring something adjacent and being honest that it's a proxy, or by describing the outcome you'd recognise rather than a number you'd count. Plenty of valuable work has no natural metric, and forcing one on it produces a goal about the measurable edge of the job rather than the job.

Should goals be tied to pay? It's a real choice with a real cost, and the cost lands on honesty. Once a goal determines somebody's money, they'll negotiate for an easier one, and you'll get a target they're confident of hitting rather than one worth aiming at. That's not dishonesty, it's the arrangement working as designed. Where you do connect the two, expect goals to become more conservative, and know that fairness and documentation requirements around pay decisions differ by jurisdiction.

Three Honest Categories the Approaches Split Into

Format-led, where the acronym drives the exercise. You open the template, work through the letters, and produce goals that satisfy them. It's right in one narrow case: where the existing goals are so vague that nothing can be assessed, and imposing a wording discipline is a genuine improvement over what came before. It fails as a way of deciding content, because the format has a bias. Measurable is the letter that bites hardest, so the goals that survive the process are the ones with numbers attached, and the parts of the job that resist counting drop out silently. Nobody decides to deprioritise them. The format does it.

Content-led, where the goal is chosen first and worded second. You work out what genuinely matters for this person this period, argue about it, settle it, and only then check the wording against something like SMART. It's right almost always, and it's the sequence most goal advice implies and almost nobody follows. It fails when the choosing step isn't actually done, so it collapses back into the first approach with more ceremony, and it fails when the conversation is skipped in the interest of time, which is what the deadline usually produces.

Conversation-led, with nothing written down. The manager and the person agree what matters and keep adjusting it as the year moves. It's right for small, stable teams doing work that changes fast, and it has a genuine advantage: it can respond to reality, which a goal written in January cannot. It fails on memory and on fairness. Nobody can recall in November what was agreed in February, the account of it shifts to suit whoever is telling it, and when a pay or promotion decision arrives there's nothing to point at except two people's differing recollections.

Five Diagnostic Questions You Can Self-Assess Against

Read last year's goals and ask what you'd have done differently without them. This is the most uncomfortable test available and it takes ten minutes. If the honest answer is nothing, the goals were describing the work rather than directing it, which is a reasonable thing for them to do and a poor use of a quarterly process.

Which parts of the job don't appear anywhere in the goals? List them. In most teams there's a significant amount of work that's genuinely necessary and completely absent from the goal set: maintenance, support, coordination, the things that only get noticed when they stop. That absence is a product of the format, and it tells people what the organisation values whether or not that's intended.

Could two people read this goal and disagree about whether it was met? Try it with a colleague. Disagreement means the goal isn't specific enough to be assessed, which is exactly what the acronym is for and exactly what it catches. This is the one test where the format genuinely earns its place.

Who chose this goal? Trace it. A goal cascaded down from something higher, reworded at each level, often arrives as a sentence nobody in the room believes in and everybody is accountable for. If the person who holds the goal can't say why it's the right goal, they'll comply with it rather than pursue it.

What would have to be true for this to be achievable? Ask before the goal is set rather than at the review. Most goals that fail were never possible with the resources, the dependencies and the competing demands the person actually had, and that was knowable in advance. Nobody asked, because achievable was checked as a feeling rather than as an arithmetic question.

The Five Letters, Reviewed One at a Time

Specific: the letter that genuinely helps

The requirement that a goal name a particular thing rather than a direction. Improve customer satisfaction is not specific; reduce the number of support cases that need a second contact is. It earns its place more than any other letter, because vagueness is the failure that makes everything downstream impossible: you can't plan against a vague goal, can't tell whether it happened, and can't have an honest conversation about it afterwards.

Where it falls short is in tempting people to be specific about the wrong thing. Specificity is easiest to achieve on the narrow, well-defined part of a job, so a goal set written under time pressure will be specific about the small stuff and silent about the large ambiguous thing that actually matters. Somebody whose real contribution is holding a complicated system together will end up with goals about the two tasks in their week that have edges.

The useful discipline is to be specific about the outcome rather than the activity. Activity goals are easy to word and easy to meet without achieving anything, which is why they're so popular with everybody involved.

The quick way to tell them apart is to ask whether the goal could be fully completed and leave the situation unchanged. Run twelve workshops is a goal you can finish while nothing is different afterwards. The outcome version names what should be true once the workshops have happened, and it's harder to write precisely because it's harder to guarantee.

Measurable: the letter that decides what the goal is about

The requirement that you can tell whether it happened. It earns its place because it settles arguments before they start: when the period ends, nobody has to negotiate about whether the thing was achieved, and that negotiation is where a great deal of goodwill gets spent.

It falls short by being the most demanding letter, and its demands have consequences. Because it's easier to satisfy with a number than with a description, it exerts steady pressure towards goals about countable things, which means the goal set drifts towards whatever your systems happen to record. That's a poor basis for deciding what matters. The other failure is subtler: a measurable goal tells people precisely what to optimise, and they will, including in ways you didn't intend.

Read as a test of clarity, this letter is excellent. Read as a requirement for a number, it quietly writes your priorities for you, and it's the letter to hold most lightly for work that doesn't naturally count.

Achievable: where the negotiation actually happens

The requirement that the goal is possible. It earns its place by forcing a conversation that would otherwise happen in November, when it's too late: what resources, what dependencies, what else this person is carrying, and whether the thing can be done alongside all of it.

It falls short because the word invites a judgement rather than a calculation, and both parties have incentives. A manager under pressure wants an ambitious goal. Somebody whose pay depends on it wants a comfortable one. Neither has a reason to state the honest number, so achievable becomes whatever both can live with, which is usually a target neither believes. The other problem is that it's assessed at the start and never revisited, so a goal that stopped being achievable in March stays on the list until December.

The version that works replaces the judgement with a question: what would have to be true for this to happen? If the answer includes something outside the person's control, that dependency is part of the goal and should be written into it.

Relevant: the only letter that asks whether it's worth having

The requirement that the goal matters, usually interpreted as connecting to something above it in the organisation. It earns its place as the only letter pointed at content rather than wording, and if it were taken seriously it would prevent most bad goals.

It falls short because it's the easiest letter to satisfy dishonestly. Almost any goal can be connected to a higher objective with one sentence of reasoning, so the test passes trivially and provides no filter. It's also frequently interpreted as alignment, meaning the goal traces to something senior, which is a different and weaker claim than the goal being worth pursuing. Cascaded goals pass relevance by construction and are often the least useful goals in the organisation.

Applied properly, this is the letter that should be used first and hardest. The question isn't whether the goal connects to something, it's what would be lost if this goal didn't exist, and that question kills a lot of goals.

Time-bound: useful, and doing less work than people think

The requirement that there's a date. It earns its place by creating a point at which the thing gets looked at, and by preventing the goal that recedes indefinitely. Dates also help with sequencing, since knowing when something is due tells you what has to happen before it.

It falls short by being satisfied almost automatically. Nearly every goal in a quarterly or annual process inherits the period's end date, so the letter is met without anybody thinking about timing at all. That's a missed opportunity: the genuinely useful version is a date that's earlier than the period end, chosen because that's when the thing needs to exist for something else to happen.

It also creates an artefact worth watching. Goals with a single end date produce work concentrated at the end of the period, and where that pattern is visible in your teams it's usually the dates rather than the people.

A second date fixes most of that at no cost. Naming a point partway through at which some visible part of the thing should exist turns a distant deadline into something with a near edge, and it gives the mid-period conversation a specific question rather than a general enquiry about how it's going.

The Decision Table

The fourth row is where most goal exercises actually sit, and it's the hardest to see from inside, because the resulting goals look correct. Everything passes, the numbers are there, and the only evidence of the problem is that the goals don't describe what anybody spends their week on.

What the Acronym Cannot Tell You

SMART is a test applied to a sentence that already exists. Every question about what the sentence should say sits outside it, and those are the questions that determine whether the goal is any good.

The first row is the whole argument. An organisation can produce a hundred goals that pass every letter and describe the wrong hundred things, and nothing in the process will detect it, because the process was never looking at that.

The fourth row is the one that causes quiet damage over time. Work that never appears in anybody's goals is work the organisation has implicitly decided doesn't count, and the people doing it notice long before anybody says it out loud. In most teams that's the maintenance, the support, the coordination and the covering for other people, which is to say a large part of what keeps the place running.

Goals for Work That Does Not Have Numbers

A great deal of valuable work produces nothing countable. Somebody who keeps a complicated system stable, who supports colleagues, who exercises judgement in ambiguous situations, or whose main contribution is that certain problems don't happen, is badly served by a format that treats measurability as a requirement.

The first honest option is to describe the outcome you'd recognise rather than a number you'd count. A goal can say what the situation should look like at the end of the period in terms specific enough that two people would agree whether it had happened. That satisfies the thing measurable is actually for, which is preventing an unresolvable argument, without pretending the work is numeric.

The second is to measure something adjacent and label it honestly as a proxy. Proxies are fine when everybody knows that's what they are, and dangerous when they get mistaken for the thing. The test is whether you'd still be pleased if the proxy improved and the underlying thing didn't, and if the answer is no, write that down next to the goal so the person knows they're not being asked to move the number by itself.

The third is to set a goal about the hardest judgement call in the role rather than about output. For work that's mostly decisions, a reasonable goal describes the decisions you want made better, and the assessment is a conversation about specific instances. That's less tidy than a number and considerably more useful, and it's the format most senior work needs.

What's worth avoiding is the fourth option, which is what usually happens: finding the one countable thing at the edge of the job and making the goal about that. It produces a technically compliant goal set, and it tells somebody whose real contribution is judgement that the organisation values their ticket count.

One more thing, for anybody in this position. If your work genuinely resists measurement and the process demands numbers, say so in writing at the time rather than at the review. A note attached to the goal explaining what it can't capture is the single most useful piece of documentation in this whole area, and it's the thing that makes an honest conversation possible nine months later.

What to Put in Writing

Most of what determines whether a goal works is decided in conversation and never recorded, which is why the same problems recur every cycle.

The third row is the one that prevents the most unfairness. Goals that depend on another team, a decision that hasn't been made, or a piece of work somebody else has to deliver first are extremely common, and when that dependency isn't written down the failure gets attributed to the person holding the goal.

Questions to Ask Before You Commit

On merit. What's lost if this goal doesn't exist? A bad answer explains how it connects to a higher objective.

On displacement. What's this more important than? A bad answer is that everything matters.

On capacity. What else is this person carrying, and does the arithmetic work? A bad answer is that they'll manage.

On the measure. Would you be pleased if this number moved and nothing else did? A bad answer hasn't considered it.

On omission. What part of this job appears in no goal? A bad answer is nothing.

On control. What does this depend on that they can't affect? A bad answer is that they'll need to influence it.

What Getting This Wrong Costs

The first cost is the one everybody recognises and nobody quite prices: the process becomes theatre. Goals are written to satisfy a deadline, filed, ignored for most of the period, and reconstructed at review time with whatever evidence can be assembled. That consumes real manager hours twice a year and produces nothing, and worse, it teaches the organisation that this category of exercise is something to be complied with rather than used. The next attempt, even a good one, inherits that.

The second cost lands on the work that doesn't count. A format that rewards measurability will, over several cycles, quietly reallocate effort towards whatever can be counted, because people are responsive to what they're assessed on. The maintenance, the support, the unglamorous coordination all continue for a while on goodwill and then start to erode, and the erosion shows up somewhere else entirely, usually as a reliability problem nobody connects to a goal-setting decision made two years earlier.

The third cost is to the conversation at the end. A goal that was never achievable, or that depended on something outside the person's control, or that measured a proxy nobody labelled as one, produces a review conversation where both people know the assessment is unfair and neither can say so cleanly. That's corrosive in a specific way: it teaches somebody that the formal account of their work doesn't match the real one, and after two of those, nobody engages honestly with the process again.

So before you write anything, work out which of three problems you have. A clarity problem means people don't know what's expected, and almost any written statement helps. A priority problem means people know what's expected and there's too much of it, and goals only help if something is explicitly dropped. A measurement problem means you can tell what matters and can't count it, and the answer is an honest description rather than a number. Only the first is solved by a better format.

When You Are Ready to Go Further

None of this needs a system or a template. It needs the goal chosen before the wording is checked, three of them rather than eight, and a written note of what the measure can't capture.

The step beyond your own team is the sequencing question at an organisational level. Most goal processes run format-first because the template arrives before the conversation does, and the calendar makes the wording deadline the visible obligation. Changing the order, so that the conversation about what matters happens first and the format check happens last, costs nothing and is mostly a question of when things are asked for.

HROpsLab publishes independent comparison work across HR tooling, applicant tracking and payroll. We sell nothing, we take no vendor money, and we publish no paid placements. If the next step is looking at what your current tooling does to how goals get written, our comparison work is one place to start.

Frequently Asked Questions

SMART tells you whether a goal is well written. It has never told anybody whether it was worth writing.