Implementing a Performance Management System for the First Time

An automotive components company sets a target of one defective part per million at the customer. The Quality Head is given a goal of “1 PPM”. So are the Plant Head, the Supplier Quality Manager and every line supervisor.

The goal forms are complete. Nobody yet knows what they personally have to change, which decisions they own or whose help they need. In March, everyone will be rated against a number that no single person controls.

That gap is what a performance management system (PMS) exists to close. It turns business priorities into clear commitments for teams and individuals. It gives managers and employees a way to review evidence, solve problems, recognise good work and build capability through the year.

An appraisal records a judgement at one point in time. Performance management is the year-round work that makes that judgement fair and useful.

If you only have two minutes

1. Start with Management by Objectives (MBO). Give each role 3–5 weighted goals that describe results, not activities.
2. Cascade from one company priority, and let teams push back on targets they cannot influence before anyone signs.
3. Talk monthly and review quarterly. A 20-minute check-in beats a 2-hour annual appraisal.
4. Pilot with two teams for 6–8 weeks before rolling out, and keep the first cycle separate from pay.
5. Train every manager to agree goals, give specific feedback and discuss a gap, before launch.
6. Calibrate instead of forcing a bell curve. Compare evidence across similar roles; do not fill rating quotas.
7. Add tools later. OKRs, the Balanced Scorecard, 360° feedback, the 9-box grid and software are useful once the basics run.

Which performance management approach should you use?

Most of what gets called a “type of PMS” answers one of three different questions:

• What should we deliver? Goal frameworks: MBO, SMART goals, KRA scorecards, the Balanced Scorecard, OKRs, Hoshin Kanri.
• How well did someone perform, and how? Assessment methods: self-assessment, rating scales, 360° feedback, the 9-box grid.
• When do we talk about it? Review rhythm: annual, quarterly or continuous check-ins.

A working system picks one answer to each. Companies get into trouble when they buy a tool for one question and expect it to answer all three.

Methods for setting and tracking goals

These methods combine well. A maintenance manager might have an MBO commitment to complete at least 95% of planned preventive maintenance on time each month, with safety-critical tasks never late. The same team could run a quarterly OKR to make a bottleneck line more reliable, cutting unplanned downtime from 12% to 9%. Both fit into the same monthly check-in.

A word on OKRs and pay. Google and most OKR practitioners keep OKR scores out of ratings and increments, because people stop setting ambitious objectives once the score affects their pay. If you use OKRs, keep them for direction and learning, and rate people on their MBO or KRA commitments.

Methods for gathering evidence and assessing performance

Continuous performance management is not a separate method. It is the habit of regular check-ins, coaching and recognition that every one of these methods needs. John Doerr’s CFR (Conversations, Feedback, Recognition) is a popular shorthand for it.

Why start with MBO?

MBO forces the three conversations a first-time organisation most needs. Leaders have to say what matters. Managers have to turn that into results each role can influence. Employees have to be able to say what good looks like and what support they need.

A one-page goal sheet and a scheduled conversation are enough to start. Bring in a Balanced Scorecard for the leadership team, OKRs for a specific programme or 360° feedback for manager development once the first cycle shows you need them.

Write goals that describe results, not activities

Five terms get mixed up in most goal sheets:

• A key result area (KRA) is an area the role is responsible for.
• A goal says what must improve or be sustained, and by when.
• A key performance indicator (KPI) is the measure that shows progress.
• An activity is something done to achieve the goal.
• A guardrail is a standard that must hold while the team chases the target, so that hitting one number does not damage another.

The activity trap. “Conduct four quality meetings”, “train operators” and “audit suppliers” belong in an action plan, not on a goal sheet. Try this test on any draft goal: imagine the activity done in full with no improvement in the result. If that is possible, the goal is incomplete.

When a project needs a milestone, say what will be accepted. “Train 20 operators” measures attendance. “By the end of Q2, all 20 designated operators can independently demonstrate the two approved setups and the stop-and-escalate procedure” measures capability someone can verify.

A complete goal has a result, a baseline, a target, a date, a scope and a source of evidence. It names one accountable owner and records its dependencies and guardrails.

How many goals, and how to weight them

Use 3–5 goals for most managerial roles, and no more than six for senior roles. Beyond six, the weights get so small that no single goal matters.

Weight each goal by its importance to the business this year, not by how much time it takes. As a rule of thumb, no goal should carry less than 10% or more than 40%. A typical Quality Head sheet might look like this:

• Customer PPM: 35%
• Repeat-incident prevention: 25%
• Critical-control effectiveness: 20%
• Team capability and succession: 10%
• Cost of poor quality: 10%

Keep safety, quality and conduct standards outside the weights as non-negotiables. A serious breach is handled directly, regardless of how well the weighted goals went.

Decide upfront whether each goal is a shared result, an individual result or a development goal.

When there is no reliable baseline, start with a learning goal. For example: “By the end of the pilot, establish a validated baseline and identify the two sources of variation that cause the most rework.” Agree the performance target once the data is credible.

More than 35 years of goal-setting research by Edwin Locke and Gary Latham point the same way: specific, challenging goals work best when people understand the task, have the resources and ability to act, and get regular feedback on progress.

Worked example: cascading a 1 PPM quality goal

Step 1. Define the company result precisely

“Reach 1.0 customer PPM or below on the named programme by the financial year-end, measured over the trailing 12 months and reviewed monthly.”

Here “PPM” means defective finished parts that reached the customer. If “final quality” instead means defects caught at final inspection, that is a different measure with a different denominator, and it needs its own goal.

Customer PPM = (customer-scored nonconforming units ÷ eligible units received) × 1,000,000

In the baseline year the customer scored 16 defective units out of 2,000,000 received: 8.0 PPM. At the same volume, two scored units equal 1.0 PPM. But one scored unit in a 50,000-unit month equals 20 PPM, so always show the numerator and denominator.

For the rolling rate, add up scored units and eligible receipts across the window and calculate once. Show “N/A” if there were no receipts. A short run with zero defects is weak evidence of performance at 1 PPM.

Agree the counting rules in writing before launch. Settle the scoring source, the reporting cut-off, and how late claims, returns and rejected lots are treated. Reconcile the internal dashboard to the customer scorecard every month, and follow the customer’s own rules for anything contractual. Many year-end arguments about a quality goal turn out to be arguments about counting.

Step 2. Set the owner’s goals: the result plus leading measures

The Quality Head owns the customer result and leads the cross-functional work behind it. Give the role the end result and earlier measures that show whether prevention is working.

Read the three together. A low PPM figure with weak control evidence still needs attention. A safety-critical escape triggers immediate containment and escalation, whatever the overall rate.

Step 3. Translate the result into each role’s contribution

Start from the actual failure modes. Suppose the baseline investigation points to three causes: bore variation in machining, defective castings, and assembly setup errors. In one joint session, decide which function controls each cause and who can change it.

Incoming PPM, internal nonconformity and customer PPM count different things. Review each on its own basis. To show how much each cause contributed to customer PPM, give every scored unit one validated primary cause, keep an “unresolved” category and use the same denominator.

Step 4. Let the cascade work upwards too (catchball)

In Hoshin Kanri this back-and-forth is called catchball: a target is passed down, questioned and passed back until each level can commit to it.

In the example, the shift in-charge reports a variation the proposed control would miss. Engineering revises the control. The Plant Head commits machine time for validation. The Quality Head takes unresolved capacity or supplier decisions to the sponsor. Record each of these commitments against the goal it supports, so people can raise a dependency before it becomes a missed target.

The same five steps work outside manufacturing. Agree the company result and its owner. Find the processes that drive it. Give each function a measurable contribution. Turn that into goals teams and individuals can influence. Then bring the owners together to check that targets, resources and timelines fit.

Get leadership buy-in in three working sessions

A PMS fails quietly when leaders sign off the design but never use it. For a first implementation, run three leadership workshops, then one goal-setting clinic per function.

At each function clinic, give managers 90–120 minutes to draft goals with the people who know the work. Ask which data the team trusts, whether the target fits the resources and what another function must deliver. Share each draft with the employee and invite corrections before anything is scored.

In a family-run business or MSME, the promoter usually holds the clearest view of what matters, often in their head. Session 1 is where that view gets written down. If the promoter does not set and review their own direct reports’ goals, nobody else will take the system seriously.

Who does what

Pilot a small version first

Pick two teams with different work patterns, such as a production shift and a support function. Train their managers first and run the pilot for 6–8 weeks: write goals, hold at least one check-in and run a mock review with anonymised cases. Keep the pilot entirely separate from pay.

Before expanding, check four things. Can employees explain their goals? Can managers find the evidence? Do similar roles face comparable expectations? Do check-ins end with actions, owners and dates? A shared template with controlled access is enough for the pilot.

A first-year calendar for an April–March financial year

Most Indian companies appraise on the financial year, with increments in April or July. Working back from that, a first implementation might run like this:

1. October–November: leadership sessions 1 and 2, goal template agreed.
2. December–January: 6–8 week pilot with two teams; manager training.
3. February: fix what the pilot exposed; leadership session 3; function clinics.
4. March: all goals agreed and signed before the year starts.
5. April–March: monthly check-ins, quarterly reviews, a mid-year review in October.
6. Next March–April: self-reviews, calibration, final conversations, then increments.

Starting in October gives the first full cycle a clean start in April. A company starting mid-year can run a shorter “practice” cycle to March, with no link to pay.

Train every manager to run the conversation

The goal sheet is the easy part. Whether the system works depends on about 20 minutes a month between a manager and each team member.

Give every line manager a three-hour practice workshop before launch and a 60–90 minute follow-up after their first round of check-ins. Keep groups small enough for everyone to practise. Brief employees separately on goal setting, how reviews work, who sees their records and where to raise a concern.

Two models worth teaching by name

SBI for feedback (Situation, Behaviour, Impact). Describe when and where, what the person did, and what effect it had. It works for praise and for problems, and stops feedback from becoming a comment on someone’s personality.

“In Tuesday’s customer review (situation), you brought the scrap data by failure mode instead of the monthly total (behaviour). The customer agreed our containment plan in the meeting instead of asking for another visit (impact).”

GROW for coaching (Goal, Reality, Options, Way forward). A four-step structure for check-ins and for discussing a gap. The manager asks more than they tell: What are you aiming for? What is happening now? What could you try? What will you do, and by when?

Four situations to rehearse

1. Agreeing a goal. Start with the business priority, ask what the employee can influence, and write the result, evidence and support together.
2. Recognising a contribution. Use SBI. Then ask what helped them succeed and what they could repeat.
3. Discussing a gap. Show the evidence, hear their account, and work out whether the cause is clarity, skill, resources, a dependency or conduct. Agree the next action and when you will check it.
4. Handling a disputed assessment. Compare the evidence with the original standard, record the employee’s view and send anything unresolved through the agreed process.

Rehearse in threes: one plays the manager, one the employee, one observes. The observer checks four things. Did the manager use specific evidence, invite the employee’s view, offer support and agree a dated action? At the follow-up, bring real anonymised cases and practise the conversations managers found hardest.

Here is what a good next step sounds like:

“This quarter we recorded seven setup-related rejects per 10,000 units against a target of three. The increase started after the fixture change. What happened on those shifts? I’ll get Engineering to confirm the fixture settings by Thursday. Let’s review the next five changeovers together on Monday and see whether the fix has held.”

The manager then follows through on the engineering support. That matters as much as recording the missed target.

Give performance a rhythm people can keep

Start with this cadence and adjust it to the work. A production team may review quality data every shift or every week, and still keep a separate monthly conversation about each person’s goals, support and development. Raise significant issues when they happen, not months later at review time.

If performance is below expectation, agree what improvement looks like, what support you will give and when you will review it. Handle any formal performance improvement plan (PIP) or disciplinary process through company policy and HR, and keep a record of progress and the employee’s response.

Make the first review fair and useful

Calibrate twice. Before the cycle, compare the goals themselves across similar roles: are they equally demanding and equally within each person’s control? At year-end, compare how managers applied the standards to the evidence. Ask whether goals changed, whether people had comparable opportunities and whether a dependency blocked delivery.

Use a simple scale. If you need ratings, define each level with examples before the cycle starts. Three levels (below, meets, exceeds expectations) are enough for a first cycle when the standards are clear. Record both results and observable behaviour. For shared goals, record each person’s contribution and the evidence for it.

Agree the edge cases in advance. Decide upfront how to assess people who joined mid-year, changed roles or were on long leave. A common rule: assess anyone with at least three months in the role against goals for that period, and record the decision.

In the quality example, the plant may miss its customer target while the shift in-charge on one cell correctly stops production, holds suspect material and escalates a risk. Record the plant result and the shift in-charge’s contribution separately, and assess each person against the part they owned.

Normalisation is not the same as a bell curve

Many Indian companies inherited a forced bell curve: a fixed 10–20% at the top, most in the middle, a fixed share at the bottom. In a team of eight it forces someone into the bottom band whatever they achieved, and it teaches people that ratings are rationed, not earned.

Normalisation done well is calibration. Managers sit together, compare evidence for similar roles and correct ratings that are too lenient or too harsh. Distributions are looked at as a check, not imposed as quotas. If one department rates 80% of its people as “exceeds”, that is a question to discuss, not a number to cut.

Watch for the common rating biases

Linking reviews to increments

In year one, keep the link loose or keep it out entirely. Once people believe a rating decides their increment, goal setting becomes negotiation over easy targets.

When you do link them, treat performance, market pay and budget as three separate inputs. A pay matrix showing increment ranges by rating and position in the salary band is fairer than a single percentage per rating. Tell employees how the link works before the cycle starts, not after the letters go out.

Explain decisions before they affect people. Employees should know who assesses them, what counts as evidence, how to correct a record or raise a disagreement, and how performance relates to pay. Each employee sees their own goals and records; access to others’ records and confidential feedback is restricted.

Check that the system is working. After the first quarter, look at completion first: how many goal sheets are agreed and how many check-ins have happened. Then look at quality. HR samples goal sheets and check-in notes and asks four questions. Can employees state their priorities? Do goals have credible measures? Are managers delivering the actions they promised? Are similar roles judged on comparable evidence? Share the answers with leadership and fix the process before the next cycle. High completion with poor quality only proves that forms were filled in.

Decide who the system covers, and how it fits each group

A first PMS does not need to cover everyone on day one. It needs a clear boundary, and a format that suits each group inside it.

In manufacturing, start with supervisors and shift section in-charges. These are the first people who own a result for a line, a shift or a section: output, first-pass quality, safety, attendance on the shift, and the skill levels of the people they lead. Their goals come straight from the cascade above. For shop-floor operators, set up a separate system that is simpler and easier for them to understand.

Include entry-level roles in sales, customer service and back office, with a lighter format. These roles usually have clear, countable outputs. A simpler version works well:

• three or four KPIs, tracked monthly, such as collections, new accounts, calls resolved, accuracy or turnaround time;
• one behaviour goal, such as following the visit plan or the escalation process;
• a short monthly conversation of 15–20 minutes, with the half-year and annual reviews built from those monthly records;
• incentives kept separate from the review, so the review stays a conversation about how to improve rather than a pay negotiation.

New joiners and people who change roles mid-year

New joiners. Set probation goals within the first two weeks: what the person should learn, deliver and show by day 30, 60 and 90. Confirmation then rests on evidence, not impression. Bring the person into the regular cycle once they have spent at least three months in the role.

Transfers and promotions. Close the old goals with a short note from the previous manager, then set new goals with the new one. At year-end, weight each part by the months spent in it. This stops a mid-year move from wiping out half a year’s work, and it keeps the previous manager accountable for their part of the record.

When to buy software

Run the first cycle on a shared spreadsheet or the simple goal and review forms in your existing HRMS. Buy dedicated software only once the pilot has shown three things: managers are holding the check-ins, goals are being written well, and you know what you need the tool to do. Software makes a working process faster. It does not fix a process that managers have not yet adopted.

When you do choose a tool, look for goal cascading and alignment, check-in notes, a calibration view, mobile access for managers who are not at a desk, and an easy export of review data into your payroll or increment sheet.

Ten mistakes that sink a first PMS

1. Copying a large company’s system. A 40-page manual and five rating scales overwhelm a first cycle. Start with MBO and one simple form.
2. Writing activities as goals. “Conduct training” is a task. “Cut setup defects to 3 per 10,000” is a result.
3. Too many goals. More than six goals spreads attention too thin. Keep three to five for most roles, and weight them.
4. Launching without leadership agreement. If the leadership team hasn’t agreed the company goals, the cascade has nothing to start from.
5. Skipping manager training. The form doesn’t make the system work; the conversation does.
6. Reviewing once a year only. Year-end surprises damage trust. Monthly check-ins prevent them.
7. Forcing a bell curve on small teams. In a team of eight, a forced distribution produces ratings nobody believes. Calibrate instead.
8. Tying every rupee to the rating in year one. Link pay through a simple matrix, and only after one calibrated cycle.
9. Changing the rules mid-year. Keep the measures and weights fixed unless the business itself changes, and record any change in writing.
10. Buying software first. A tool won’t fix a process managers haven’t adopted yet.

Your launch checklist

• Company goals for the year agreed by the leadership team
• Approach chosen (MBO with a weighted scorecard is the usual start)
• Coverage defined: who is in, who is out, and the format for each group
• Goal-writing clinics held for every manager
• Managers trained in SBI and GROW, with rehearsal
• Pilot run in one or two departments
• Check-in calendar published for the full year
• Rating scale and calibration process agreed
• Pay matrix agreed and shared with leadership
• One-page record for each employee set up

The one-page performance record

Frequently asked questions

What is the difference between performance appraisal and performance management?

An appraisal is the year-end judgement of how someone performed. Performance management is the full cycle around it: setting goals, holding regular check-ins, giving feedback, supporting development and then reviewing. In a good system, the appraisal summarises conversations that have already happened, so nothing in it comes as a surprise.

How long does it take to implement a PMS for the first time?

Plan for five to six months from the first leadership session to signed goals. That covers about two months of leadership and design work, a 6–8 week pilot, and a month to fix what the pilot exposed and train every manager. The first full cycle then runs for a year.

What is the difference between a KRA and a KPI?

A KRA is an area the role is responsible for, such as customer quality. A KPI is the measure that shows progress in that area, such as customer PPM. A goal joins the two with a baseline, a target and a date.

How many KRAs should each role have?

Three to five for most roles, and no more than six for senior roles. Give each a weight between 10% and 40%, with the weights adding up to 100%.

Does a small company or MSME need a formal PMS?

It needs one once the promoter can no longer review everyone personally, which usually happens when a layer of managers appears between the promoter and the team. Start with the simplest version in this guide: a one-page goal sheet, monthly check-ins and one calibrated year-end review.

Should ratings decide increments in the first year?

Not directly. Keep the first cycle unlinked, or only loosely linked, so people set honest goals. After one calibrated cycle, link pay through a matrix that also takes market pay and budget into account.

Do we need a bell curve?

No. Calibrate ratings by comparing evidence across similar roles, and use the distribution as a check rather than a quota. A forced curve is especially unfair in small teams.

Should OKRs replace KRAs?

Usually not in a first PMS. Use KRAs and MBO goals for accountability and assessment. Add OKRs for a specific improvement programme once the basics are working.

Getting the first cycle right

A performance management system works when managers trust it, employees understand it and the results connect to the business. The first cycle sets that tone for years.

TNSL’s HR Centre of Excellence helps companies design and launch their first PMS: setting company goals with leadership, writing role goals, training managers and running the first calibration. If you are planning your first cycle, talk to us about where to start.

Sources and further reading

• Peter F. Drucker, The Practice of Management, Routledge. The origin of management by objectives.
• George T. Doran, “There’s a S.M.A.R.T. way to write management’s goals and objectives”, Management Review, 1981.
• Robert S. Kaplan and David P. Norton, “The Balanced Scorecard: Measures That Drive Performance”, Harvard Business Review, 1992.
• John Doerr, Measure What Matters, Portfolio, 2018. On OKRs and CFR.
• Google re:Work, Set goals with OKRs.
• Yoji Akao (ed.), Hoshin Kanri: Policy Deployment for Successful TQM, Productivity Press, 1991.
• Edwin A. Locke and Gary P. Latham, “Building a practically useful theory of goal setting and task motivation: A 35-year odyssey”, American Psychologist, 2002.
• Patricia C. Smith and Lorne M. Kendall, “Retranslation of expectations: An approach to the construction of unambiguous anchors for rating scales”, Journal of Applied Psychology, 1963. The origin of behaviourally anchored rating scales.
• John C. Flanagan, “The critical incident technique”, Psychological Bulletin, 1954.
• John Whitmore, Coaching for Performance, Nicholas Brealey. The GROW model.
• Center for Creative Leadership, the Situation–Behaviour–Impact (SBI) feedback model.

Related reading from TNSL

• Performance management should be a year-round conversation, not an annual event
• The manager effect: building better managers to build better workplaces
• When should an MSME build a formal HR function?