Most organisations that fail at performance management do not fail at designing the framework. They fail in the twelve weeks after it is signed off. The KRAs are written, the document is circulated, and then nothing happens differently — because nobody planned the rollout with the same care they gave the design. This is what the first ninety days should actually look like.
Before day one: three things that must already be true
A rollout that starts before these are settled will consume goodwill you cannot get back. If any of the three is missing, fix it first; the delay is cheaper than a failed launch.
- The promoter has their own KRAs written, and has agreed to be reviewed against them. This is not symbolism. In a promoter-led business, everyone is watching to see whether this applies upward. If it visibly does not, the exercise is read as paperwork for staff, and managers will treat it accordingly.
- Every role in scope has an agreed set of result areas and measures, with a named source for each number. If the measure depends on someone remembering, it will not survive a busy quarter. The methodology for getting this right is set out in Dr. Babu’s guide to KRA and KPI design — that is the design step, and this article picks up after it.
- A decision has been taken, in writing, on whether ratings affect pay this cycle. Our strong recommendation is that they do not, in cycle one. Where ratings drive increments from the first cycle, managers inflate them to protect their teams and the data is worthless. Say so openly at launch and you will get honest ratings.
Weeks 1 to 3: start with one group, not the whole organisation
The instinct is to launch everywhere at once, so nobody feels singled out. It is the wrong instinct. A full-organisation launch means every mistake in the design is made simultaneously across every department, and you discover them all in the same week.
Pick one function of fifteen to forty people. The best candidate is a department whose head is respected, moderately sceptical, and willing to say when something is not working. An enthusiast will tell you it is going well when it is not.
In these three weeks the pilot group does one thing: the manager sits with each person and confirms that the written result areas match what that person actually does. This surfaces the design errors that no workshop catches — the outcome two people both think they own, the responsibility nobody owns, the measure that turns out not to exist in any system.
Expect to change ten to twenty per cent of the framework here. If nothing changes, the conversations were not honest.
Weeks 4 to 6: extend, and let the pilot group do the explaining
Now widen it. The important choice at this stage is who does the communicating. A rollout announced by HR or by a consultant is a policy. A rollout explained by a manager who has already been through it, in their own words, including what they disliked about it, is credible.
Have the pilot managers brief the next group. Let them say the awkward parts out loud. The scepticism you allow into the room at this stage is scepticism that does not go underground later.
Two practical points. Keep the form short enough to complete in twenty minutes — a form that takes an hour will be completed carelessly at 6pm on the deadline. And put it in a system rather than a spreadsheet, because from here the version-control problem becomes real; this is where the performance management system and the HRMS underneath it start to earn their place.
Weeks 7 to 9: the first goal-setting conversations
This is the point at which the exercise becomes real for employees, and where most rollouts quietly go wrong. The failure is not refusal; it is compliance without engagement. Targets get accepted in the meeting and disowned in private.
The single strongest safeguard is the requirement that a target must be argued with before it is agreed. Tell managers explicitly that a conversation in which the employee accepts every number without question has failed, and should be reopened. Managers find this counter-intuitive — they read agreement as success — which is why it needs saying more than once.
What managers need before these conversations
- Practice at writing an observation rather than an adjective. Missed the dispatch cut-off on four of eleven consignments in October is usable. Not proactive is not.
- Permission to renegotiate a target that changed circumstances have made meaningless, and a defined route for doing so. Without it, people quietly stop tracking against a number everyone knows is dead.
- A clear separation between the development conversation and the compensation conversation. Held together, the second one crowds out the first entirely.
- Somewhere to put the material — course content, examples, the framework itself — that people can return to weeks later, which is where an LMS stops being a nice-to-have.
Weeks 10 to 12: the first check-in, and the corrections it forces
A short structured check-in at the end of the first quarter is the moment the system either becomes a habit or becomes an annual ritual. It should take twenty to thirty minutes per person and cover three questions: what has actually moved, what is blocked, and does anything in the target need to change.
Leadership should look at the distribution of these conversations, not their content. The useful signals are administrative rather than substantive:
- Which managers completed them and which did not. Non-completion in cycle one predicts non-completion permanently, and needs addressing as a management issue immediately rather than at year end.
- Where every rating is identical. A manager who rates their entire team the same is avoiding the conversation, not observing an unusually uniform team.
- Where targets were renegotiated and why. A cluster of renegotiations in one function usually means the original targets were set without the information the function actually had.
- Where the evidence field is empty. Ratings without evidence are opinions with a number attached, and will not survive being challenged at appraisal time.
The rollout mistakes that cost the most
| Mistake | What it causes | What to do instead |
|---|---|---|
| Launching everywhere at once | Every design flaw surfaces simultaneously and confidence collapses | Pilot one function of 15–40 people first |
| Linking ratings to pay in cycle one | Managers inflate to protect their teams; the data is unusable | Decouple for one cycle, announce it openly, link in cycle two |
| HR or a consultant fronting the launch | Reads as policy imposed from outside | Pilot managers brief the next group in their own words |
| A form that takes an hour | Completed carelessly against the deadline | Twenty minutes maximum, five to seven result areas |
| Waiting for a perfect framework | The launch never happens | Launch something defensible; correct at the mid-cycle review |
| No route to renegotiate a target | People stop tracking against numbers everyone knows are dead | A defined, legitimate renegotiation step at mid-cycle |
| Promoter exempt from the process | Read as paperwork for staff only | Promoter KRAs written first and reviewed like everyone else |
Measure the rollout, not just the people
For the first two cycles the useful question is not how employees performed. It is whether the system is being used honestly. Four indicators tell you that, and all four are available from the system itself rather than from opinion:
- Completion rate by manager, not organisation-wide. An overall figure of eighty per cent can hide a department at zero.
- Rating spread within each team. Identical ratings across a team is the clearest signal of avoidance there is.
- Proportion of ratings carrying written evidence. This is the number that determines whether a promotion decision will be defensible a year from now.
- Time from cycle close to conversation held. Long gaps mean the review is being treated as an administrative submission rather than a discussion.
Pulling these together is what HR analytics is for at this stage. Not dashboards for their own sake — four numbers that tell leadership whether the thing they paid for is actually running.
Frequently asked questions
Ninety days seems fast. Is it realistic?
Ninety days covers the rollout of an agreed framework, not its design. Designing result areas for a whole organisation takes longer and comes first. If the design is not finished, the clock has not started.
Can we roll out to everyone at once if we are small?
Below roughly forty people, the whole organisation is effectively the pilot, so yes. Above that, pilot first. The threshold is not really headcount — it is whether one person can still personally observe every conversation.
What if a manager refuses to engage?
Address it in cycle one, explicitly, as a management performance issue. A manager who does not complete reviews in the first cycle will not complete them in the third, and the rest of the organisation is watching to see whether the standard is enforced.
Do we need software to do this?
Not to design it. You will want it by week four or five, when version control and completion tracking become real problems. Spreadsheets are workable for the pilot and start to hurt as soon as it widens — the practical thresholds are set out in HRMS vs spreadsheets.
Where to go from here
If you are designing the framework, start with KRA and KPI design and, for the wider question of whether your organisation is ready, what a performance management engagement involves. If the framework is agreed and you are planning the rollout, our case study walks through a full engagement phase by phase, and the performance management system is what the framework runs on afterwards.
BPro is based in Kochi, Kerala and works with organisations across India. Call +91 80869 08876, email care@bpropms.com, or use the contact form.
Pingback: From Appraisal to Training Plan: Turning Performance Gaps into Learning - BPro PMS