OKRs at Work: A Goal-Setting Guide for HR Teams

Most HR teams in India already run some form of goal setting. There are KRAs in the appraisal form, a rating scale, and a review cycle that fires up once or twice a year. And yet the same complaint surfaces every cycle: nobody quite remembers what they committed to in April, half the goals were written to be safely achievable, and the conversation in December is really about increments rather than about progress. OKRs — Objectives and Key Results — are one of the few goal-setting frameworks that consistently fixes this, but only when they are implemented as an operating rhythm rather than as a new form to fill.

This guide is a practical, end-to-end treatment of OKR goal setting for HR teams. Not a philosophy lecture. We will cover what OKRs actually are, how they differ from KRAs, KPIs and SMART goals, how to write objectives and key results that hold up under scrutiny, how many to set, how to run the quarterly cadence week by week, how to score them, whether they should touch appraisals and pay, worked OKR sets for six different functions, the anti-patterns that quietly kill adoption, and a 90-day rollout plan you can start on Monday.

The audience here is the HR leader or HR business partner in a 20-to-500 person Indian company — services, SaaS, manufacturing, D2C, it does not matter much. You are usually the person who has to make goal setting work across functions that all think their work is uniquely unmeasurable. The good news: OKRs are simple enough to explain in one workshop. The hard part is the discipline afterwards, and that is mostly what this guide is about.

What OKRs Are

An OKR has exactly two parts.

The Objective is a short, qualitative statement of what you want to achieve in a defined period. It should be memorable, directional and slightly uncomfortable. "Make onboarding something new joiners talk about" is an objective. It does not contain a number.

The Key Results are two to five measurable statements that tell you whether the objective was achieved. They are quantitative, time-bound and verifiable by someone other than the person who wrote them. "New joiner 30-day satisfaction score of 4.4 out of 5" is a key result.

Put together: the objective tells you where you are going; the key results tell you how you will know you have arrived. The framework originated in Silicon Valley management practice in the 1970s and spread widely through the tech industry from the late 1990s onward. That history is interesting but not important. What matters is the mechanic: pair an ambition with evidence, review it frequently, and score it honestly.

Three properties make OKRs different from most goal frameworks:

1. They are time-boxed and repeating. Almost always quarterly, sometimes with an annual layer above. The short cycle is deliberate — it forces re-negotiation with reality four times a year instead of once.
2. They are transparent. Everyone can see everyone else's OKRs. This is not a nice-to-have; it is the mechanism by which alignment happens without a chain of command doing the aligning.
3. They are deliberately incomplete. OKRs capture what needs to change, not everything you do. A payroll executive still processes payroll every month whether or not it appears in an OKR.

What OKRs Are Not

This section prevents more failed rollouts than any other.

OKRs are not a task list. If your key result is "conduct 12 training sessions," you have written a to-do item with a number attached. You will complete all 12 and learn nothing about whether training worked. The outcome version is "raise post-training assessment pass rate from 61% to 80%."

OKRs are not a job description. Business-as-usual work — running payroll, closing books, answering tickets — belongs in health metrics or KPIs, not in quarterly OKRs. If everything you do becomes an OKR, the framework loses all signal.

OKRs are not a performance appraisal system. More on this later, but the short version: the moment an OKR score determines someone's rating or increment, people write easy OKRs. The framework's entire value comes from ambition, and ambition dies the instant it becomes risky.

OKRs are not a project plan. A project plan tells you what will be done and by when. An OKR tells you what will be different afterwards. You need both; do not confuse them.

OKRs are not a monitoring tool for managers. If OKR check-ins become status-reporting theatre where people defend themselves, you have built a surveillance system wearing a goal-setting costume.

OKRs vs KRAs vs KPIs vs SMART Goals

Indian HR practice is layered with legacy frameworks. Most companies already have KRAs baked into the appraisal template, KPIs on a dashboard somewhere, and a memory of a SMART goals workshop. These are not competitors — they answer different questions and they can coexist quite comfortably if you are clear about what each is for.

Here is how they nest in practice for a single person — say, a Talent Acquisition Specialist at a 120-person company.

• Her KRA is "Technology hiring for the product org." That is the territory she owns. It will still be her KRA next year.
• Her KPIs are time-to-fill, offer acceptance rate, cost per hire, and 90-day new-hire retention. These are the vital signs of her territory. They are always being measured, quarter or no quarter.
• Her SMART goal might be "Complete the structured interview training for all 14 engineering interviewers by 30 September." Specific, dated, achievable. It is a commitment, not a stretch.
• Her OKR for the quarter is the thing that needs to change: "Objective — Make engineering hiring fast enough that we stop losing candidates to slower competitors. KR1: Reduce median time-to-offer from 34 days to 18. KR2: Raise offer acceptance from 62% to 80%. KR3: 90% of engineering interviews conducted by trained interviewers."

Notice the SMART goal became a supporting activity for KR3. That is the correct relationship. SMART goals and project plans are how you deliver OKRs; they are not substitutes.

Where each belongs in an Indian SMB's performance process

A workable division of labour looks like this:

• KRAs stay in the job description and the annual appraisal form. They define scope and are the basis for "is this person doing their job?"
• KPIs live on a dashboard in your HRMS or BI tool. They are reviewed monthly by function heads. They answer "is the engine healthy?"
• OKRs live in a separate, visible, quarterly document or module. They answer "what are we changing this quarter?" They are reviewed weekly by teams.
• SMART goals and projects are the execution layer underneath OKRs.

Companies get into trouble when they try to merge all four into one appraisal form. The form becomes 3 pages long, nobody reads it, and the quarterly ambition gets diluted into annual bureaucracy. Keep OKRs separate. They can inform the appraisal conversation without being the appraisal.

Anatomy of a Good Objective

An objective is a sentence. It should pass four tests.

1. It is qualitative, not numeric. The number belongs in the key result. If your objective already contains "increase retention to 92%," you have written a key result and left the "why" unstated.

Weak: Achieve 92% retention. Strong: Make our best engineers want to stay another year.

2. It has a direction and an edge. A good objective implies that today's situation is not acceptable. "Continue our excellent hiring practices" has no edge. "Stop losing senior candidates at the offer stage" does.

3. It is memorable. If a team member cannot recite the objective roughly from memory in week 7, it is too long. Aim for under 12 words. Plain language beats corporate language. "Payroll that never surprises anyone" beats "Enhance payroll operational excellence through process optimisation."

4. It is owned by one team or one person. Shared ownership sounds collaborative and produces orphans. If two teams must both deliver it, one owns the objective and the other owns a supporting objective with a KR that links to it.

A useful sanity test: read the objective aloud and ask "so what?" If the answer is obvious and motivating, it is a good objective. If the answer is "well, it is what we do anyway," it is a KRA in disguise.

Anatomy of a Good Key Result

Key results are where most of the craft lives. Three distinctions matter.

Output vs outcome

An output is something you produced. An outcome is something that changed in the world because of it.

Outputs are not forbidden — sometimes a launch genuinely is the goal, especially in the first quarter of a new initiative. But if every KR in your set is an output, you are running a project tracker.

Leading vs lagging

A lagging indicator tells you what already happened: attrition rate, revenue, engagement score. A leading indicator predicts it: number of stay conversations held, pipeline coverage, 1:1 completion rate.

A well-balanced OKR set contains both. Lagging KRs keep you honest about the goal; leading KRs give you something to act on in week 3 when the lagging number has not moved yet.

Example for an HR objective around attrition:

• KR1 (lagging): Reduce voluntary attrition in the engineering team from 24% to 15% annualised.
• KR2 (leading): Complete stay conversations with all 22 engineers rated "high impact."
• KR3 (leading): Close 100% of compensation-band outliers identified in the July benchmarking review.

If KR2 and KR3 are green in week 6 and KR1 is still flat, you have not failed — you have a lag. If KR2 and KR3 are red, you know exactly why KR1 will not move.

Milestone vs metric

A metric KR moves along a number line: from X to Y. A milestone KR is binary: done or not done.

Metric KRs are better. They give partial credit, they show trend, and they resist gaming. But milestone KRs are legitimate when the thing genuinely has no meaningful intermediate state — "POSH committee constituted and all members trained by 15 September" is done or it isn't.

Rule of thumb: at most one milestone KR per objective. If you have three milestones and no metrics, you have written a project plan.

Writing the number properly

Every metric KR should have three components: baseline, target, and source.

Weak: Improve employee engagement. Better: Improve engagement score to 4.2. Strong: Improve engagement score from 3.6 (June pulse) to 4.2 (September pulse), measured on the same 12-question instrument.

The "same instrument" clause matters more than people expect. Half of all disputed OKR scores come from someone changing how the number was measured mid-quarter.

How Many OKRs Should You Set?

The single most common failure mode is volume. Here is a defensible set of limits.

Some deliberate advice on individual OKRs: do not start with them. In companies under about 60 people, team-level OKRs plus good 1:1s cover the ground. Individual OKRs multiply the administrative load by the headcount and are the fastest route to OKR fatigue. Introduce them in year two, if at all, and only for roles where individual contribution is genuinely separable from team output.

A quick arithmetic check for a 120-person company: 4 company objectives, 6 functions with 2-3 objectives each, and a handful of team objectives gives you roughly 25 objectives and 70 key results in total. That is already a lot to review. If your first draft has 60 objectives, cut it in half, then cut it again.

Cascading vs Aligning

The classic mental model is a waterfall: the CEO sets objectives, each function takes one of the CEO's key results and makes it their objective, each team does the same, down to individuals. It is tidy on a slide and it fails in practice for three reasons.

It is slow. Each layer must wait for the layer above to finish. In a 200-person company with four layers, pure cascading eats three to four weeks of the quarter before anyone starts work.

It produces mechanical goals. When your objective is forced to be a restatement of your manager's key result, you write what fits rather than what matters. Support teams suffer most — a payroll team asked to derive its objective from a revenue KR produces something meaningless.

It kills the bottom-up signal. The people closest to the work usually know the highest-leverage improvement available. Pure cascading gives them no channel to say so.

The alignment model

Alignment keeps the connection but drops the rigid derivation. It works like this:

1. Leadership publishes draft company objectives early — ideally three weeks before the quarter starts, deliberately marked as a draft.
2. Functions draft their own objectives in parallel, not in sequence. Each function asks: "given where the company is going, what is the most valuable thing we could change in the next 90 days?"
3. Each function OKR names the company objective it supports. That is the link — a reference, not a derivation.
4. Teams hold cross-functional alignment sessions where dependencies are surfaced and negotiated before publishing.
5. Leadership finalises company OKRs after hearing the function drafts, adjusting where the bottom-up signal reveals something they missed.

The practical difference: in a cascade, HR asks "what did the CEO give us?" In an alignment model, HR asks "what does the company need from us, and what do we know that they don't?"

Alignment also handles the awkward reality that some functions serve the company rather than a single company objective. A payroll and compliance team may support "scale to 300 people without operational breakage" while also owning statutory obligations that map to no company objective at all. That is fine. Not every function OKR needs a parent.

Handling dependencies

The alignment session is where you catch the sentence "our KR depends on the product team shipping X." Two rules:

• If your KR depends on another team, that other team must have a matching KR or an explicit written commitment. An unacknowledged dependency is a broken OKR waiting to happen.
• If they will not commit, change your KR. Do not carry a KR whose success is entirely outside your control — you will spend the quarter building an excuse.

The Quarterly Cadence

OKRs live or die on rhythm. The framework itself takes an hour to explain; the cadence takes a quarter to build and a year to make automatic. Here is a workable calendar for a company on an April-June quarter. Adapt the dates to your own fiscal calendar — many Indian companies run April-March, so Q1 is April-June.

Some notes on the individual ceremonies.

Drafting

Give teams a blank template and two prompts: "What is the most important thing that must be different at the end of this quarter?" and "How would a sceptical outsider verify it changed?" Do not let people draft in a spreadsheet alone at their desk. The good version is a 90-minute working session where the team argues about priority. Most of the value of OKRs is generated in that argument, not in the document that results.

Draft more objectives than you will keep. Six candidate objectives cut to two is a healthier process than two objectives written to fill two slots.

Alignment sessions

One rule makes these sessions work: every team presents its draft OKRs to peers, not to their manager. Peer review surfaces duplication ("both of us have a KR about the referral programme"), catches vagueness, and finds dependencies. Manager review tends to become approval, which turns into negotiation, which turns into sandbagging.

Timebox each team to five minutes plus five minutes of questions. Questions to seed the room with:

• "If you hit all your key results, is the objective actually achieved?"
• "What would have to be true for this to fail?"
• "Whose help do you need, and have you asked them?"
• "Which of these is a task rather than an outcome?"

Publishing

Publish everything in one place, visible to everyone. This is non-negotiable. Hidden OKRs cannot align anything and immediately breed suspicion that some teams have easier goals. An HRMS with a goals module is the natural home; a shared document works at 20 people but breaks around 60.

Run a 45-minute all-hands where each function head reads out their objectives — not the key results, just the objectives. Ten minutes of context beats a 40-slide deck nobody re-opens.

Weekly check-ins

Twenty minutes, same slot every week, run by the team itself. The agenda is fixed:

1. For each KR: current value and confidence (a simple 1-10 or red/amber/green).
2. What changed since last week?
3. What is blocked, and who can unblock it?
4. What are the top three things we will do this week to move these numbers?

The check-in is not a status report to a manager. It is the team looking at its own numbers. The moment it becomes performance theatre for a senior observer, people start reporting green until week 11.

Confidence tracking is underrated. A KR that has been at "confidence 4" for three consecutive weeks is a louder signal than any percentage-complete bar. Make confidence the first thing you look at.

Mid-quarter recalibration

Around week 6, hold a session that explicitly permits change. Three legitimate outcomes:

• Keep — the KR is still the right one; push harder.
• Adjust — the world changed (a client left, a regulation shifted, a key person resigned). Adjust the target and record why. An adjusted KR is not a failed KR.
• Drop — the objective is no longer the priority. Kill it publicly rather than letting it rot on the page.

The recorded reason matters. Without it, adjustment becomes a quiet escape hatch and the whole system loses credibility. With it, adjustment is a sign that the company is paying attention.

Do not allow adjustment after week 8. Late-quarter target changes are indistinguishable from moving the goalposts.

Scoring and the 0.0-1.0 Convention

At the end of the quarter, each key result gets a score between 0.0 and 1.0. The objective's score is the average of its key results — or a weighted average if some KRs clearly matter more, though weighting adds complexity you may not need in year one.

The arithmetic is straightforward for metric KRs:

Score = (final value − baseline) ÷ (target − baseline)

If you set out to move attrition from 24% to 15% and you landed at 19%, the score is (24 − 19) ÷ (24 − 15) = 5 ÷ 9 = 0.56.

Cap scores at 1.0 unless your culture explicitly allows over-achievement credit. Allowing 1.4 scores tends to reward under-ambitious targets more than genuine outperformance.

Milestone KRs score 0 or 1, though many teams allow 0.5 for "substantially done, one piece remaining." Be careful — 0.5 becomes a default hiding place.

The counter-intuitive part, and the part that most Indian companies find hardest culturally: an average score around 0.7 is the target, not 1.0. If your teams routinely score 0.95, the targets are too soft. If they routinely score 0.3, either targets are fantasy or the team is under-resourced — both are worth investigating, and both are usually a leadership problem rather than a team problem.

This is genuinely hard to socialise in an environment where "target achieved" has always meant 100%. Two things help. First, say it explicitly and repeatedly: "0.7 is a good score here." Second, and more importantly, do not attach money to the score. Which brings us to the most consequential design decision in the whole system.

Should OKRs Drive Appraisals and Pay?

Short answer: not directly. Here is the reasoning, and the nuance.

Why coupling breaks OKRs

The mechanism is simple and predictable. If OKR score determines rating, and rating determines increment, then every employee faces a rational incentive to negotiate the easiest possible target in week 1. This is not a character flaw; it is what any sensible person does when their child's school fees depend on the number.

What follows:

• Targets get set at or below the expected baseline (sandbagging).
• Ambitious teams that swing big and score 0.5 look worse than cautious teams that score 1.0, so ambition gets punished.
• Mid-quarter honesty disappears — nobody flags a red KR in week 6 if red means a lower increment.
• Retrospectives become defensive rather than diagnostic.

Within two cycles, you have a system that generates safe numbers and no learning. The framework is intact on paper and dead in practice.

What decoupling actually means

Decoupling does not mean OKRs are irrelevant to performance. It means the score is not an input to a formula. Concretely:

• Do use OKRs as evidence in the performance conversation. "You owned the hiring OKR, hit 0.8 on two KRs and 0.3 on the third, and here is what you learned" is rich material for a review.
• Do assess how someone pursued their OKRs — the judgement, collaboration and resilience they showed. That is behaviour, and behaviour belongs in an appraisal.
• Do not write "OKR achievement = 40% of the rating" in the appraisal policy.
• Do not auto-populate a rating from an average OKR score in your HRMS, even if the software offers to do it.

A useful framing for managers: OKR scores tell you what happened; the appraisal is about the person. A team can score 0.4 in a quarter where the market collapsed and still contain the best performer in the company.

The pragmatic middle ground

Some companies cannot decouple entirely — particularly where sales incentive schemes or statutory bonus structures already tie payment to numeric targets. In those cases:

• Keep commit OKRs (things you are confident of delivering, targeted at 1.0) separate from aspirational OKRs (stretch, targeted at 0.7). Only commit OKRs can touch variable pay.
• For sales, keep the quota where it is — in the incentive plan — and let sales OKRs cover the things quota does not, like pipeline quality, discounting discipline or new-segment penetration.
• Review the split annually. Most companies that start coupled find they can decouple within two years once trust in the process is established.

Worked Example OKR Sets

Below are complete, realistic OKR sets for six functions. They are written for a hypothetical 150-person B2B software company in its third year. Use them as patterns, not as templates to copy — your context will differ.

Human Resources / People Team

Objective 1: Make the first 90 days the best 90 days of a new joiner's year. - KR1: Raise 30-day new-joiner satisfaction from 3.7 to 4.4 out of 5 (same onboarding survey). - KR2: Reduce time-to-first-meaningful-contribution, self-reported by manager, from 42 days to 25 days. - KR3: 100% of new joiners have a documented 30-60-90 plan on day one, up from 35%. - KR4: Reduce 90-day voluntary exits from 6 to 1.

Objective 2: Stop losing our strongest people to preventable causes. - KR1: Reduce regretted attrition from 18% to 11% annualised. - KR2: Complete stay conversations with all 30 employees identified as high-impact; log outcomes. - KR3: Close 100% of compensation outliers more than 15% below band midpoint. - KR4: Raise "I can see a path to grow here" pulse score from 3.1 to 3.9.

Objective 3: Give managers the tools to actually manage. - KR1: 1:1 completion rate rises from 54% to 90% across all people managers. - KR2: Manager effectiveness index rises from 3.5 to 4.0. - KR3: All 22 people managers complete the feedback and difficult-conversations module with a pass score.

Note the weak versions you would want to avoid here. "Conduct manager training" (task). "Improve engagement" (no number, no baseline). "Roll out an HRMS" (output with no stated outcome). "Ensure statutory compliance" (that is a KRA — it is your job every quarter, and it does not belong in an OKR unless something is genuinely broken).

Recruitment / Talent Acquisition

Objective 1: Hire senior engineers faster than our competitors can. - KR1: Reduce median time-to-offer for senior engineering roles from 38 days to 21 days. - KR2: Raise offer acceptance rate from 64% to 82%. - KR3: Raise the share of hires from referrals and direct sourcing from 22% to 45%, reducing agency dependence. - KR4: Maintain hiring manager satisfaction with shortlist quality at 4.2 or above.

Objective 2: Make our interview process something candidates recommend. - KR1: Candidate experience NPS rises from +11 to +40 across all interviewed candidates, including rejects. - KR2: 95% of candidates receive a decision within 4 working days of their final round, up from 58%. - KR3: 100% of interviewers for senior roles complete structured interview certification.

Before and after, for the same intent:

Payroll and HR Operations

Operations teams often protest that OKRs do not fit them because their work is repetitive. That objection confuses KRAs with OKRs. Payroll runs every month regardless — that is the KRA. The OKR is about what should be better about it in 90 days.

Objective 1: Payroll that never surprises anyone. - KR1: Reduce payroll-related tickets from an average of 47 per cycle to under 12. - KR2: Reduce post-payroll corrections and off-cycle payments from 9 per month to 1. - KR3: Move payroll input freeze from T-2 days to T-6 days with zero missed statutory deadlines. - KR4: 100% of employees can access their payslip, Form 16 and tax declaration self-service without contacting HR.

Objective 2: Close the compliance gaps before an auditor finds them. - KR1: Complete internal audit across PF, ESI, PT, TDS and shops-and-establishments registers; zero unresolved high-severity findings at quarter end. - KR2: Reduce statutory filing lead time from "day of deadline" to "3 days before deadline" for all 12 monthly filings. - KR3: Full-and-final settlement completed within 30 days of last working day for 100% of exits, up from 55%.

Sales

Objective 1: Build a pipeline that does not depend on heroics in the last week of the quarter. - KR1: Raise pipeline coverage at quarter start from 2.1x to 3.5x of target. - KR2: Raise the share of revenue closed in the first two months of the quarter from 38% to 60%. - KR3: Reduce average discount from 17% to 10% on new business. - KR4: Increase qualified opportunities created per rep per month from 6 to 10.

Note that the revenue quota itself is deliberately absent. It lives in the incentive plan. The OKR covers the health of how revenue gets made — the things that quota alone will not fix.

Objective 2: Win in the mid-market segment we keep losing. - KR1: Close 8 new logos in the 200-1000 employee segment, up from 2 last quarter. - KR2: Raise win rate against our two most common competitors from 28% to 40%. - KR3: Reduce mid-market sales cycle from 74 days to 55 days.

Engineering

Objective 1: Ship without breaking things. - KR1: Reduce change failure rate from 21% to 8%. - KR2: Reduce median time-to-restore from 4.5 hours to under 45 minutes. - KR3: Raise deployment frequency from weekly to daily for the core service. - KR4: Reduce P1 incidents from 7 per quarter to 2.

Objective 2: Make the product fast enough that nobody mentions speed. - KR1: Reduce p95 page load on the three most-used screens from 3.4s to under 1.2s. - KR2: Reduce support tickets mentioning slowness from 34 per month to under 8. - KR3: Reduce backend p99 API latency from 900ms to 350ms.

Weak engineering KRs to avoid: "complete the migration" (milestone with no stated benefit — say what improves), "write more tests" (activity), "improve code quality" (unmeasurable as stated), "reduce technical debt by 30%" (30% of what, measured how?).

Customer Support

Objective 1: Solve problems before customers have to chase us. - KR1: Reduce median first-response time from 6 hours to 45 minutes during business hours. - KR2: Raise first-contact resolution from 48% to 70%. - KR3: Raise CSAT from 4.0 to 4.5. - KR4: Reduce reopened tickets from 14% to 6%.

Objective 2: Cut the volume of tickets we should never have received. - KR1: Reduce total ticket volume per 100 active accounts from 31 to 20 without any drop in CSAT. - KR2: Deflect 25% of "how do I" tickets through self-serve help content, up from 4%. - KR3: Ship the top 10 support-identified product friction points to engineering with 7 of them fixed.

That last KR is a good example of a well-formed cross-functional dependency: support owns identifying and prioritising, engineering has a matching commitment to fix. Neither team can score it alone, and both know it going in.

Common Mistakes and Anti-Patterns

These are the failure modes that show up again and again. Most rollouts die from two or three of them in combination.

1. Sandbagging

Setting targets you already know you will hit. Symptoms: every score is 0.95 or above; targets look suspiciously close to last quarter's actuals; nobody ever misses.

The fix is structural, not motivational. Decouple scores from pay. Ask in the alignment session, "what would you have set if there were no consequences for missing?" and then set that. And have leadership publicly score their own OKRs at 0.6 and treat it as normal.

2. Task lists disguised as key results

Symptoms: KRs begin with verbs like "implement," "conduct," "create," "launch," "roll out." Every KR is binary. The quarter ends with everything checked off and no number moved.

The fix: for each KR, ask "and then what?" Launch the handbook — and then what? People know the policies. That is your KR. Force at least two-thirds of KRs to be metrics with a baseline.

3. Too many OKRs

Symptoms: 8 objectives per team, 5 KRs each, a document nobody can hold in their head. Everything is a priority, so nothing is.

The fix: hard caps, enforced. Three objectives maximum per team. If the team insists everything is essential, ask which one they would keep if the quarter were cut to four weeks. Then keep that one and two others.

4. Set and forget

Symptoms: OKRs are written in week 1, opened again in week 12. Confidence never updated. The check-in meeting is cancelled twice and never restored.

This is the most common cause of death, and it is undramatic. Nobody objects to OKRs; they simply stop looking at them. The fix is calendar discipline — a recurring, protected 20-minute slot — plus one senior person who visibly asks about OKR progress every week.

5. OKR theatre

Symptoms: beautifully formatted OKR decks, a dedicated OKR champion who chases updates, dashboards that are always green, and no observable change in what anyone actually works on. The process is impeccable and the outcomes are unchanged.

Theatre usually appears when OKRs were adopted for appearance rather than need. The diagnostic question: "name one decision we made differently this quarter because of an OKR." If nobody can answer, you have theatre.

6. Copying last quarter

Symptoms: the same objectives, quarter after quarter, with the target nudged up 5%. Usually a sign that the objectives are actually KPIs.

Some continuity is fine — a genuinely hard objective can span two quarters. But if an objective survives three quarters unchanged, it is business-as-usual and belongs on a dashboard.

7. Only lagging indicators

Symptoms: every KR is an outcome measured at quarter end. Weeks 1-9 have nothing to report. The team can only find out it failed in week 11.

The fix: at least one leading indicator per objective, ideally something measurable weekly.

8. OKRs written by managers alone

Symptoms: the team first sees its OKRs when they are published. Ownership is zero. Nobody argues about them because nobody feels responsible for them.

The fix: the team drafts, the manager edits. Not the other way round.

9. No owner

Symptoms: a KR belongs to "the team." When it goes red, there is a long silence in the check-in.

Every KR needs one named human. That person does not do all the work — they are accountable for knowing the number and escalating.

10. Confusing confidence with completion

Symptoms: "we are 70% done" reported as 0.7. Percentage of work completed is not the same as percentage of result achieved. Track the metric, not the effort.

A 90-Day First-Time Rollout Plan

If you are introducing OKRs for the first time, resist the urge to launch company-wide with individual OKRs in month one. This plan assumes you start about four weeks before a quarter begins.

Weeks -4 to -3: Decide and design

• Get explicit CEO commitment. Not approval — commitment. If the CEO will not publish and score their own OKRs, do not start. The single strongest predictor of a failed rollout is a leader who exempts themselves.
• Decide the scope of the pilot. Recommendation: 2-4 teams, including at least one operations team and one team led by a sceptic. A sceptic who is won over is worth ten enthusiasts.
• Decide the cadence dates and put them in calendars now, for the whole quarter.
• Decide explicitly and in writing that OKR scores will not feed appraisal ratings this year. Announce it. This removes the biggest single source of gaming before it starts.
• Choose where OKRs will live — HRMS goals module, or a shared document for a small pilot.

Weeks -2 to -1: Train and draft

• Run a 90-minute workshop. Half of it should be hands-on rewriting of weak KRs into strong ones using examples from your own company. Lecture time: 25 minutes maximum.
• Leadership drafts company objectives and shares them as a draft.
• Pilot teams draft their OKRs in facilitated 90-minute sessions.
• Hold the cross-team alignment session. Expect the first one to be messy and to run long.
• Finalise and publish.

Weeks 1-4: Build the habit

• Weekly check-ins, every team, no exceptions. Someone senior attends the first two of each team's check-ins to model the tone — curious, not prosecutorial.
• Expect the first three weeks to feel awkward. People will report activity instead of numbers. Redirect gently and consistently.
• After week 3, collect feedback: what is confusing, what feels like busywork? Fix the process, not the people.

Weeks 5-8: Recalibrate and prove value

• Run the mid-quarter recalibration in week 6. Publicly adjust or drop at least one thing. This teaches the organisation that OKRs are living documents.
• Find one concrete story where an OKR changed a decision, and tell it at the all-hands. Adoption spreads on stories, not on policy documents.
• Start informally socialising the idea of expansion to other teams.

Weeks 9-12: Score and learn

• Week 11: remind teams that a 0.6 is a respectable score, before they start scoring.
• Week 12: teams score their own KRs. Managers may challenge the arithmetic, not the honesty.
• Week 13: retrospective. Four questions: What did we learn? Which KRs turned out to be the wrong measure? Where did the process waste our time? What are we changing next quarter?
• Then expand: add two or three more teams for quarter two. Full company by quarter three or four. Individual OKRs, if ever, in year two.

A blunt expectation-setting note: your first quarter's OKRs will be mediocre. Roughly a third will turn out to be tasks in disguise, a couple will be unmeasurable, and one team will forget about theirs entirely. This is normal and it is not a reason to abandon the framework. Quality of OKR writing improves sharply in cycle two and again in cycle three. Judge the system on quarter three, not quarter one.

Running OKRs at 20 People vs 200 People

The framework is the same; almost everything around it changes.

At 20 people, the honest advice is to keep it light. Three company objectives, nine key results, one weekly conversation. If a founder can hold the whole plan in their head, you do not need software, you need discipline. The risk at this size is over-engineering.

At 200 people, the constraint flips. Nobody can hold it in their head, the CEO cannot personally review 40 objectives, and dependencies between functions become the main source of failure. Here you need three things you did not need at 20: a real system of record, a named process owner, and formal dependency negotiation. The risk at this size is that OKRs become an administrative ritual disconnected from work.

The transition point is usually somewhere between 60 and 100 people, and it announces itself in a specific way: someone says "I did not know that team was working on that" about something that directly affects their own goals. That is the signal to formalise.

What to Track in Your HRMS

If OKRs live in a spreadsheet, they will drift. A goals module in your HRMS earns its keep by making four things effortless.

1. The OKR record itself. Objective text, owner, period, parent objective (the alignment link), status, and for each KR: description, type (metric or milestone), baseline, target, current value, unit, owner, and confidence.

2. History, not just current state. The single most useful chart in an OKR system is a KR's value over 12 weeks. It shows whether progress was steady or all crammed into week 11 — which tells you far more about the team's operating health than the final score does. Keep an immutable log of every update, who made it, and when.

3. Process compliance signals. These are the leading indicators of whether OKRs will survive:

• Percentage of KRs updated in the last 7 days
• Percentage of teams that held their check-in this week
• Number of KRs with no owner
• Number of KRs with no baseline
• Average number of objectives per team (watch for creep)
• Number of KRs that were adjusted mid-quarter, with reasons

4. Connections to the rest of the people system. OKRs should be visible inside the 1:1 agenda, referenced in the review form, and linked to development plans. If a manager has to open a different tool to see their report's goals, they will not.

How OKRs Connect to Reviews, 1:1s and Development Plans

OKRs are not a standalone ritual. They are most valuable when they quietly supply material to the three conversations HR already runs.

In 1:1s

Spend five minutes, not thirty. The manager's job in a 1:1 is not to audit progress — the weekly check-in already did that. It is to ask the questions the check-in cannot:

• "Which of your key results are you least confident about, and what would help?"
• "Is anything on your plate right now not connected to these? Should it be?"
• "What did you learn from the KR that is going badly?"

The most useful function of OKRs in a 1:1 is as a workload reality check. If someone's week is full of work that appears nowhere in the team's OKRs, either the OKRs are wrong or the week is.

In performance reviews

Use OKRs as evidence, never as a formula. A good review conversation pulls three things from the quarter's OKRs:

• What was attempted. Ambition is visible in OKR history. Someone who consistently takes on hard objectives and lands at 0.6 is often more valuable than someone who lands at 1.0 on safe ones — and without OKR history you would never see it.
• How they responded to trouble. The record of confidence updates and escalations shows whether someone flagged problems early or hid them until week 11.
• Where the gaps are. A KR that failed for a skill reason is the single best input into a development plan.

Write it into the review template as a prompt, not a score field: "Which OKRs did you own this quarter, and what did you learn from them?"

In development plans

The link is mechanical and underused. Take each failed or struggling KR and ask why. Sort the reasons into three buckets:

• Capability — the person did not yet have a skill. This becomes a development goal with a named learning action.
• Capacity — the person did not have time or resources. This is a management problem, not a development one.
• Clarity — the KR was the wrong measure or badly defined. This is a process fix for next quarter's drafting.

Doing this exercise once a quarter turns OKR failures into the most concrete development inputs you will ever get. It is far better than asking someone in an annual review to nominate a development area from a dropdown.

Frequently Asked Questions

How often should OKRs be set — quarterly or annually?

Quarterly for team and function OKRs, with an optional annual layer of three to five company objectives that stay stable through the year. Ninety days is short enough to force focus and long enough for something meaningful to change. Companies with very long delivery cycles — heavy manufacturing, regulated products, long-cycle enterprise sales — sometimes run four-month or six-month cycles. That is fine, but keep the weekly check-in regardless of cycle length.

Should every employee have individual OKRs?

No, and especially not in your first year. Team OKRs plus good 1:1s cover the ground for most companies under 100 people. Individual OKRs multiply administrative effort by headcount, tend to collapse into task lists, and are the leading cause of OKR fatigue. Introduce them selectively for senior individual contributors whose work is genuinely separable from team output — and even then, cap at two objectives.

What if a team misses all its OKRs?

First, distinguish between the three causes. If the targets were fantasy, that is a leadership calibration problem. If the team was blocked and did not escalate, that is a process and psychological safety problem. If the priority genuinely changed mid-quarter and the OKR was never updated, that is a recalibration discipline problem. Only rarely is "the team did not try" the answer. Treat a total miss as a diagnostic event, not a disciplinary one — and if you treat it as disciplinary once, you will never get an honest score again.

Can OKRs work for operations and support functions?

Yes, and they often work better there than in sales, because the improvement opportunities are so concrete. The trick is to separate the recurring work (the KRA — payroll runs every month) from what should improve about it (the OKR — payroll tickets drop from 47 to 12). Every operations function has cycle times, error rates, ticket volumes and manual-effort measures that can move.

How do OKRs work with the Indian April-March financial year?

Map quarters to your fiscal calendar rather than the Gregorian one: Q1 April-June, Q2 July-September, Q3 October-December, Q4 January-March. Two practical adjustments: build the drafting window for Q1 in the second half of March when annual planning is already happening, and expect Q3 to be disrupted by the festive season across October-November — set slightly fewer objectives that quarter rather than pretending the disruption will not happen.

Should OKRs be linked to increments and bonuses?

Preferably not directly. Coupling the score to money produces sandbagging within one or two cycles and destroys the ambition that makes the framework worth adopting. Use OKRs as rich evidence in the performance conversation, and let the appraisal assess the whole person — judgement, collaboration, growth — rather than an arithmetic average. If your context forces some linkage, separate "commit" OKRs (targeted at 1.0, may touch variable pay) from "aspirational" OKRs (targeted at 0.7, never linked).

What is a good OKR score?

Around 0.7 on average for aspirational OKRs. Consistent scores above 0.9 usually mean targets are too soft; consistent scores below 0.4 mean targets are unrealistic or teams are under-resourced. The score is a conversation starter, not a verdict — the more useful question is always "what did this number teach us?"

How long does it take before OKRs actually work?

Plan for three cycles. Quarter one produces awkward, task-shaped OKRs and patchy check-ins. Quarter two is noticeably better because people have seen a full cycle end. By quarter three, the drafting sessions get sharper, the check-ins get shorter, and you start hearing OKRs referenced in decisions rather than only in meetings about OKRs. Judge the system then.

Bringing It Together

OKRs are not complicated. An objective worth caring about, two to four key results that would convince a sceptic, a weekly twenty-minute conversation, an honest score, and the humility to change what did not work. That is the whole framework.

What makes them hard is everything around them: the courage to publish an ambitious number, the discipline to keep the check-in when the week is busy, the restraint to cap the list at three objectives, and the organisational maturity to score a 0.6 without anyone reaching for the appraisal form. Those are cultural muscles, and they take about a year to build.

Start smaller than feels right. Two or three teams, one quarter, decoupled from pay, with a leader who publishes and scores their own goals in public. Get one cycle finished, run the retrospective properly, and expand from there.

If you would like the mechanics handled for you — objectives and key results with baselines and targets, alignment links between company, function and team goals, weekly check-in reminders, progress history, and goals that sit alongside your 1:1s, reviews and development plans rather than in a separate spreadsheet — that is exactly what the performance management module in CozyHR is built for. Have a look, set up one quarter with a couple of teams, and see whether the rhythm sticks.