OKRs for Growing Companies: Setup and Review Guide

Most growing companies do not have a goal problem. They have a goal clarity problem. Ask five people in a 120-person company what the most important thing to achieve this quarter is, and you will usually get five reasonable, sincere, completely different answers. OKRs — objectives and key results — exist to close that gap. Done well, OKR goal setting turns a vague strategy deck into a short list of outcomes that everyone can name, argue about, and track without being asked.

Done badly, OKRs become the most expensive spreadsheet in the company. Someone in HR spends three weeks collecting goals, the leadership team reviews them once, managers forget about them by week four, and in month three everyone scrambles to backfill updates before the quarterly review. This is not a failure of the framework. It is a failure of rhythm, drafting discipline, and tooling — and all three are fixable.

This guide is written for HR leaders, people managers and founders at Indian SMBs and startups who are either rolling out OKRs for the first time or trying to revive a rollout that has gone quiet. It covers what an OKR actually is, how it differs from a KPI and from a traditional appraisal goal, why rollouts stall, how to write key results that survive contact with reality, the quarterly operating rhythm, scoring, the relationship between OKRs and increments, a 90-day rollout plan, and a set of ready-to-adapt examples across functions. Where numbers appear, treat them as illustrative — your targets should come from your own baselines, not from a blog post.

What OKRs Actually Are (and What They Are Not)

An OKR has two parts. The objective is a short, qualitative statement of what you are trying to achieve — memorable, directional, slightly ambitious, and written in language a new joiner would understand. The key results are the two to five measurable outcomes that tell you whether the objective actually happened.

The objective answers "where are we going?" The key results answer "how will we know we got there?" That is the whole framework. Everything else — scoring scales, cascading, check-in cadence, confidence ratings — is implementation detail layered on top of those two ideas.

The anatomy of a well-formed OKR

A complete OKR at a growing company usually has five components:

• Objective: qualitative, time-boxed, inspiring without being fluffy.
• Key results: two to five measurable outcomes with a baseline, a target and a metric.
• Owner: one named person, not a team, not a department.
• Timeframe: usually a quarter, sometimes a year for company-level objectives.
• Initiatives or projects: the work you believe will move the key results. These are not key results.

That last distinction is the one most teams get wrong in their first cycle, and it is worth spending time on. Launching a new careers page is an initiative. Reducing time-to-offer from 34 days to 22 days is a key result. The initiative is a bet; the key result is the scoreboard.

What OKRs are not

OKRs are not a task list. If a key result can be completed by one person ticking a box on a Friday afternoon, it is a task masquerading as a goal.

OKRs are not a description of your job. A customer support lead's job includes answering tickets. That is business as usual, measured by KPIs. An OKR is the thing that should be different by the end of the quarter.

OKRs vs KPIs vs Traditional Appraisal Goals

Confusion between these three is the most common source of resistance from managers, especially in companies that already run an annual appraisal with weighted goals. They are not competing systems. They answer different questions.

A KPI is a health metric you watch continuously. It has no natural end date. Gross margin, monthly churn, CSAT, offer-acceptance rate, DSO — these should be on a dashboard whether or not anyone wrote an OKR about them.

An OKR is a change agenda for a defined period. It exists because you decided a particular number needs to move, or a particular capability needs to exist, in the next 90 days.

A traditional appraisal goal (the kind that sits in an annual review form with weightages) is primarily an individual accountability and evaluation instrument. It mixes outcomes, behaviours, development goals and role expectations, and it is usually tied to rating and increment.

How the three work together in practice

A healthy setup looks like this. Your KPI dashboard runs all year. Once a quarter, leadership looks at the dashboard, decides which two or three numbers are not where they need to be, and turns those into objectives. Individual appraisal goals then draw on OKR contribution as one input among several — not as an arithmetic formula.

A note on Indian financial-year cycles

Most Indian SMBs run an April-to-March financial year, with appraisals and increments landing in Q4 or Q1. This matters for OKR planning: your Q1 (Apr-Jun) OKR cycle will collide with appraisal season unless you plan for it. Either start your first OKR cycle in Q2 (Jul-Sep) when calendars are calmer, or deliberately shorten the Q1 planning window and accept that leadership attention is split.

Why Most OKR Rollouts Fail

Having watched this pattern repeat, the failure modes are remarkably consistent. Almost none of them are about the framework itself.

Too many objectives

A team with nine objectives has no objectives. It has a to-do list with aspirational formatting. The whole value of OKRs comes from the argument you have while cutting the list from nine to three — that argument is the strategy work.

Output disguised as outcome

This is the most common drafting error. "Publish 12 blog posts," "close 40 support tickets a week," "complete the ERP migration" — these describe activity you control, not change you are trying to cause.

Output-style key results feel safe because they are fully within your control. That is exactly the problem: you can hit all of them and change nothing. A reasonable compromise for infrastructure-type work is to write the outcome as the key result and keep the delivery milestone as an initiative underneath it.

OKRs wired directly to bonus

The moment an OKR score determines a payout, two things happen immediately. People sandbag their targets during drafting, and people negotiate their scores during grading. You have converted a learning system into a bargaining system, and you will never get an honest confidence rating again.

No check-in rhythm

An OKR that is updated twice — once at the start and once at the end — is a forecast, not a management tool. Without a weekly or fortnightly touchpoint, there is no mechanism to notice that a key result has gone off track while you can still do something about it.

Copy-paste cascading

The mechanical version of cascading — every manager copies their boss's objective, adds their name, and passes it down — produces documents that are technically aligned and practically useless. The sales team ends up with the CEO's revenue number restated four times at four levels of the org.

Real alignment is not replication. A team's OKR should answer: "given the company objective, what is our distinct contribution, stated in our own metrics?" Marketing's contribution to a revenue objective is qualified pipeline, not revenue.

No one owns the system

Frameworks do not roll themselves out. Someone — usually an HR business partner, a chief of staff, or a founder in a smaller company — has to own the calendar, chase the drafts, run the review, and keep the tool tidy. In companies where this role is undefined, the rollout reliably dies in month two.

How to Write a Key Result That Works

Writing good key results is a skill, not a personality trait. Most managers get noticeably better after one guided drafting session and one review cycle.

Three tests every key result should pass

1. The measurement test. Can you name the exact source of truth — the report, the dashboard, the system — where this number will be read at quarter end? If the answer involves "we'll figure it out," the KR is not ready.
2. The outcome test. If this key result hits its target and nothing else changes, has anything meaningfully improved for the business or the customer? If not, you have written an initiative.
3. The surprise test. Is there any real chance of missing it? If the answer is no, it is a status report, not a goal.

The standard format

The most reliable structure is: verb + metric + from baseline + to target + by when.

"Increase trial-to-paid conversion from 11% to 18% by 30 September" leaves nothing to interpretation. "Improve conversion" leaves everything to interpretation, and you will spend the grading meeting arguing about it.

Weak versus strong key results, by function

The following rewrites are illustrative — the numbers are examples, not benchmarks. Use your own baselines.

| Engineering | Complete the platform refactor | Cut p95 API latency on the checkout flow from 1.9s to under 800ms |

| HR | Improve hiring | Reduce time-to-offer for engineering roles from 38 days to 24 days |

Notice what the strong column has in common: a number you can look up, a baseline, and a real possibility of failure.

Handling work that is genuinely hard to measure

Some important work resists clean metrics in the first cycle — a compliance overhaul, a brand repositioning, a first-time ISO audit. Two honest options:

• Milestone key results with defined quality bars. "Complete external security audit with zero critical findings" is a binary KR, but it is unambiguous and meaningful.
• Proxy metrics with an explicit caveat. "Raise internal policy-awareness score from 45% to 80% in a random sample of 50 employees" is imperfect but measurable.

The Drafting and Alignment Process, Step by Step

This is the part most first-time rollouts underinvest in. Budget two to three weeks of calendar time, not one afternoon.

1. Start with the annual picture. Before anyone writes a quarterly objective, leadership should agree on the three to five things that must be true by the end of the financial year. Without this anchor, quarterly OKRs drift toward whatever is loudest.
2. Review last quarter honestly. Pull the previous cycle's scores, the misses and the reasons. Carry forward what is still important; consciously kill what is not. Unreviewed OKRs quietly become permanent.
3. Draft company objectives in a small room. Three to five objectives, drafted by the leadership team in one or two working sessions. Argue about the cut list. Write them in plain language.
4. Share the draft company OKRs widely before they are final. Send them to all managers with an explicit note that they are drafts. This single step does more for buy-in than any amount of training.
5. Let teams draft their own contribution. Each team lead answers: "What is our distinct contribution to these objectives, in our own metrics?" Give them a week. Resist the urge to write it for them.
6. Run a horizontal alignment session. Get team leads in one room to read each other's drafts. This is where you catch the dependencies — marketing promising leads that sales has not planned for, engineering committing to a launch that support cannot staff.
7. Cut, then cut again. Most teams will arrive with twice as much as they can deliver. The facilitator's job is to force the question: if we could only do one of these, which one?
8. Set baselines and data sources for every key result. For each KR, record the current value, the target, the source system and the named owner. Any KR without a baseline goes back for rework.
9. Set confidence levels at kickoff. Ask each owner for a starting confidence (for example, on a 1-10 scale or as a simple red/amber/green). This becomes the reference point for every check-in.
10. Publish, then freeze the set. Once the quarter starts, the OKR set should not change except through an explicit mid-quarter reset. Continuous editing destroys the ability to learn anything from the scores.

The Quarterly Operating Rhythm

An OKR system is a calendar, not a document. If you get the calendar right, the documents mostly take care of themselves.

The default 13-week cadence

Weekly check-ins that people actually do

The check-in should take under ten minutes per person and answer four questions:

• What is the current value of each key result you own?
• What is your confidence that you will hit the target? (Same scale every week.)
• What changed since last week?
• What is blocking you, and who needs to unblock it?

The confidence rating is the most useful field and the most commonly skipped. A key result sitting at 40% progress in week six is ambiguous. A key result at 40% progress with confidence that has dropped from 8 to 4 over three weeks is a clear signal that something needs attention now.

The mid-quarter reset

Around week six, leadership should sit down with the full OKR set and make three explicit decisions for every objective:

• Keep as is — on track or recoverable, no change needed.
• Re-resource — still important, but needs people, budget or a deprioritised competing project.
• Kill or park — the assumption behind it turned out to be wrong.

Killing an OKR mid-quarter is a feature, not an admission of failure. The alternative is a team spending six more weeks on work that leadership already knows does not matter, which is far more expensive than the awkwardness of the conversation.

Keeping the rhythm alive after the honeymoon

Cycle one is easy; everyone is curious. Cycle three is where systems die. Three practices that help:

• Put the OKRs on the wall of every recurring meeting. Literally open them at the start of the weekly leadership meeting, every week, without exception.
• Have the CEO or founder reference them in writing. A monthly all-hands note structured around the company objectives does more than any HR reminder.

Scoring and Grading OKRs

Scoring is where opinions diverge sharply. What matters is that you pick one approach, explain it clearly, and apply it consistently.

Common approaches

For a first rollout at an Indian SMB, percentage complete plus a red/amber/green summary is usually the easiest to explain and the hardest to game. Decimal scoring can come in cycle three, once people trust that scores are not a pay lever.

Reading a score honestly

A score is a conversation starter, not a verdict. The useful interpretations:

• Consistently above 90% across a whole team: the targets were too safe. Praise the delivery, then push the ambition next quarter.
• 60-80%: usually the healthy zone for genuinely stretching goals. Real progress, real learning.
• Below 40%: something broke. It may be the target, the resourcing, the assumption, or the market. Find out which before deciding what it means.

The question to ask in every scoring session is not "did you hit it?" but "what did we learn about our ability to predict?" A team that missed at 55% and can explain precisely why is in better shape than a team that hit 100% and cannot explain how.

OKRs, Performance Reviews and Increments

This is the question every HR leader gets asked in the first town hall: "So will this affect my appraisal?"

The honest answer is: it will inform it, but it will not calculate it. Here is why that is the right design, and how to explain it.

The case for loose coupling

Stretch goals and compensation are structurally incompatible. If missing a target costs you money, you will set targets you are confident of hitting — which means the company loses the ambition that made OKRs worth adopting.

There is also a fairness problem. A key result can be missed because a customer went bankrupt, a regulation changed, or a co-founder reprioritised the team mid-quarter. Attaching pay to an uncontrolled outcome is not accountability; it is a lottery.

What "loosely coupled" looks like in practice

• OKR history is visible in the performance review as context — what the person owned, what moved, what they learned.
• Managers write a narrative assessment that references OKR contribution alongside behaviours, collaboration and role expectations.
• There is no formula converting OKR score to rating or increment percentage.
• Calibration discussions can reference OKRs, but a manager who says "she scored 0.7 so she gets a 3" should be challenged.

Explaining it to employees without sounding evasive

Say it plainly: "OKRs are how we decide what the company works on and how we track it. Your appraisal looks at how you performed overall, and your OKR work is a big part of that story — but we are not doing arithmetic on scores, because we want you to set targets you might miss."

Rolling Out OKRs in Phases

Company-wide big-bang rollouts have a poor track record. Phasing works better because it lets you fix your own process before exposing 200 people to it.

Phase 1: leadership only (one quarter)

Only the leadership team writes OKRs. The rest of the company sees them but writes none. The goals for this phase are to establish the drafting habit, find out how bad your data hygiene actually is, and give leaders the experience of being scored before they ask anyone else to be.

Phase 2: leadership plus two or three teams (one to two quarters)

Pick teams with a willing manager and reasonably clean metrics — often sales, support or marketing. Avoid starting with the team that is in crisis; you will conflate OKR problems with firefighting.

Phase 3: all teams (from the following quarter)

Roll out to every team with team-level OKRs. Keep individual OKRs off the table unless a specific team has a strong case.

Signals that you are ready for the next phase

• Over 80% of check-ins are completed without a chase.
• Managers are cutting their own OKR lists without being asked.
• At least one OKR was killed at the mid-quarter reset, deliberately.
• Scoring happened without anyone negotiating a number.

If any of these are missing, stay where you are for another quarter. There is no prize for rolling out fast.

The 90-Day OKR Rollout Plan

This assumes you are starting from scratch with no goal system, or replacing an annual-goals-only process. Adjust the dates to your financial year.

1. Days 1-5: Get sponsorship and name an owner. Secure explicit commitment from the CEO or founder, and name one person as the OKR champion with the time to do it. Without both, stop here.
2. Days 6-12: Audit what already exists. Catalogue current goal documents, KPI dashboards, the appraisal template and where each metric actually lives. You will usually find that half the numbers you want to use are not reliably measured anywhere.
3. Days 13-20: Fix the measurement gaps you can. For each metric you intend to use in cycle one, confirm a single source of truth and an owner who can produce the number on demand. Drop what you cannot measure today.
4. Days 21-30: Define the model and write it down. Decide cadence (quarterly), scoring approach, check-in frequency, who writes OKRs in phase one, and the explicit policy on the link to pay. Keep it to two pages. Circulate to leadership for sign-off.
5. Days 31-38: Train the leadership team. One 90-minute working session on writing key results, followed by a live drafting exercise using real company priorities. Do not run a theory session with generic examples — draft the actual OKRs in the room.
6. Days 39-50: Draft, align and cut the first OKR set. Run the drafting process described earlier. Expect two rounds of rework. The first draft will contain too many objectives and several initiatives disguised as key results; that is normal.
7. Days 51-55: Set up the tool. Configure OKRs in your HRMS: objectives, key results, owners, baselines, targets, check-in frequency and reminders. Do this before the quarter starts so week one is not spent on data entry.
8. Days 56-60: Announce company-wide. One all-hands, one written summary, one place people can see the OKRs without asking. Explain the pay policy explicitly in this session; if you do not, the rumour will fill the gap.
9. Days 61-90: Run the first cycle in public. Weekly check-ins from day one, reviewed in the existing leadership meeting. The champion chases missed check-ins personally for the first three weeks; after that, habit does most of the work.
10. Day 90 onward: Review, adjust, expand. Run the mid-quarter reset on schedule, score honestly at quarter end, hold a process retrospective separate from the results retrospective, and decide whether to expand to phase two.

The most common reason this plan fails is step 1. An OKR rollout without a visibly committed founder is an HR project that competes with revenue work, and it loses.

Manager Enablement and Common Coaching Conversations

Managers are the transmission mechanism. If they treat check-ins as admin, so will their teams.

What managers actually need to be taught

Keep training short and practical. Four capabilities cover most of it:

• Writing a measurable key result from a vague ambition.
• Running a ten-minute check-in that surfaces blockers rather than rehearsing status.
• Saying no to work that does not serve the quarter's objectives.
• Scoring honestly, including scoring their own team's misses without defensiveness.

Run this as a workshop using the team's own draft OKRs, not case studies. Managers learn the framework by applying it to work they care about.

Conversations managers will have, and how to handle them

"Everything on my plate is important." The response is not to argue about importance but to make the trade-off visible: "If we commit to all five, which one do you expect to slip first? Let's just decide that now instead of in week ten."

"I don't control this number." Usually true and usually fixable by reframing. If a support lead cannot control churn, they can control first-response time and reopen rate, which are their contribution to churn. Find the part of the outcome they do influence.

"Can we change the target? The situation changed." Yes, at the mid-quarter reset, with a written reason. Not silently in week nine.

Running OKRs in an HRMS Instead of a Spreadsheet

Almost every company starts OKRs in a spreadsheet, and there is nothing wrong with that for one leadership-only cycle. The problems begin when you scale.

Where spreadsheets break down

• No history. Someone overwrites a target in week seven and there is no record of the original. At scoring time you cannot tell what you committed to.
• No reminders. Check-ins depend on someone remembering to chase, every week, forever.
• No visibility. Employees cannot see how their team's work connects upward without opening a file most of them will never open.
• Disconnection from people data. When someone changes roles or exits, their key results become orphans that nobody notices until quarter end.
• No audit trail for reviews. At appraisal time, you have a number with no context — no check-in notes, no blockers, no record of what changed.

What to look for in a goals module

• Objectives and key results with owner, baseline, target, current value and progress calculation.
• Alignment links between company, team and individual goals, with a visible tree.
• Scheduled check-ins with automated nudges and a confidence field.
• Historical snapshots so you can see how a key result moved week by week.
• A clean handoff into the review cycle — the ability to pull goal history into a review form without retyping it.

Why keeping goals next to people data matters

When goals live in the same system as your org structure, attendance, leave and payroll data, a few useful things follow without extra effort. Reporting lines stay accurate, so alignment trees do not break when someone moves teams. Exiting employees' key results are flagged for reassignment automatically. And the performance review cycle can draw on the actual check-in history rather than on what everyone remembers from the last three weeks.

CozyHR takes this approach — goals, check-ins and reviews sit alongside the employee record, so the quarterly rhythm runs on the same system that already holds your people data.

What tooling will not fix

Software cannot make a leadership team choose. It cannot turn eleven objectives into three, and it cannot make a founder open the dashboard on Monday. Tooling removes friction from a working process; it does not create one. Get the rhythm right on paper for one cycle, then automate it.

Ready-to-Adapt Example OKRs by Function

Use these as starting structures, not as targets. Every number below is illustrative and should be replaced with your own baseline and a target you have argued about internally.

Company level

Objective: Turn our best customers into our growth engine. - Increase net revenue retention from 96% to 112%. - Raise the share of new logos sourced from referrals from 8% to 20%. - Improve logo retention in the top revenue decile from 88% to 96%.

Sales

Objective: Win more of the deals we are already in. - Increase win rate on qualified opportunities from 18% to 27%. - Reduce average sales cycle for mid-market deals from 74 days to 52 days. - Raise the share of deals with a documented champion and a mutual action plan from 30% to 75%.

Engineering

Objective: Make the product fast enough that speed stops being a sales objection. - Cut p95 page load on the three highest-traffic screens from 3.4s to under 1.5s. - Reduce p95 API latency on the checkout flow from 1.9s to under 800ms. - Bring the count of performance-related support tickets from 45 to fewer than 12 per month.

HR and People

Objective: Hire faster without lowering the bar. - Reduce time-to-offer for engineering roles from 38 days to 24 days. - Raise offer acceptance rate from 68% to 85%. - Keep 90-day new-hire performance ratings at or above 4 of 5 for 90% of joiners.

Objective: Make the first 90 days the reason people stay. - Cut voluntary attrition in the 0-12 month tenure band from 24% to 15%. - Raise the 30-day onboarding experience score from 3.4 to 4.4 out of 5. - Ensure 100% of new joiners have a documented 30-60-90 plan within their first week.

Objective: Get payroll and compliance off the risk register. - Reduce payroll queries raised after salary credit from 31 to fewer than 8 per cycle. - Achieve zero late statutory filings for three consecutive months. - Cut manual payroll processing time from 3 days to under 6 hours per cycle.

Customer Support

Objective: Resolve issues before customers have to chase us. - Improve first-response SLA compliance from 71% to 92%. - Raise first-contact resolution from 48% to 65%. - Reduce ticket reopen rate from 14% to under 6%.

Finance

Objective: Give the business cash visibility it can act on. - Reduce DSO from 64 days to 45 days. - Bring invoices raised within 24 hours of delivery from 55% to 95%. - Publish a rolling 13-week cash forecast weekly, with variance under 10%.

Frequently Asked Questions

How many OKRs should a team have in one quarter?

One to three objectives, with two to four key results each. A team with more than three objectives is describing its workload rather than its priorities. If a leader insists everything is essential, ask which one they would protect if half the team were unavailable for a month — the answer is usually clarifying.

Should every employee have individual OKRs?

Not in your first year, and often not at all in companies under 150 people. Team-level OKRs plus clear role expectations deliver most of the alignment benefit with a fraction of the administrative load. Introduce individual OKRs only where a person's outcomes are genuinely separable from the team's.

What is the real difference between OKRs and KPIs?

KPIs are ongoing health metrics that you monitor whether or not anything is changing; OKRs are a time-boxed change agenda. A KPI tells you the patient's temperature; an OKR is the decision to bring the fever down this quarter. Most OKR key results are drawn from KPIs you already track, which is why a reliable dashboard should come before an OKR rollout.

Can we link OKR scores to increments or bonuses?

You can, but you probably should not link them directly. Once a score determines money, people set targets they know they can hit and negotiate scores at quarter end, and you lose the honest signal that makes the system useful. Use OKR history as context in a human judgement about performance instead of as an input to a formula.

What do we do when an OKR becomes irrelevant mid-quarter?

Kill or park it at the mid-quarter reset, with a written reason. Business conditions change, and persisting with an obsolete objective for six weeks is far more costly than the discomfort of cancelling it. What you should avoid is silently editing targets week by week, because that destroys your ability to learn anything from the cycle.

How long does it take before OKRs feel natural?

Plan for three cycles. The first is awkward and produces too many objectives and several initiatives disguised as key results. The second is noticeably better because managers have seen scoring happen. By the third, drafting takes half the time and the check-in habit largely sustains itself.

Do OKRs work for non-target functions like finance, HR or legal?

Yes, though the key results look different. Support functions often have fewer growth metrics and more cycle-time, quality and risk-reduction metrics — close-by-day-5, time-to-offer, zero late filings, query volume after payroll. The discipline is the same: state the outcome, name the baseline, and pick a number you might miss.

Conclusion

OKRs are not complicated. Two to three objectives, a handful of measurable key results, a named owner for each, a ten-minute weekly check-in, an honest mid-quarter reset, and a scoring session where the misses get discussed as openly as the wins. The framework fits on one page.

What is hard is the discipline: cutting the list when everything feels urgent, writing outcomes instead of activity, resisting the temptation to wire scores to pay, and showing up to the check-in in week nine when the novelty has worn off. Companies that get OKR goal setting to work are not the ones with the cleverest objectives — they are the ones that kept the rhythm for four quarters in a row.

If you decide to move your goals out of a spreadsheet, CozyHR handles OKRs, check-ins and performance reviews in the same place as your employee records, attendance and payroll — so the quarterly rhythm runs on data you already trust. It is worth a look when your rollout outgrows the sheet.