Performance Management
A Nicosia managing partner asked for a KPI system for his lawyers. What his firm got instead was one properly prepared conversation a year, a mid-year check, one goal under each heading, and an email he was copied on. Here is why, and how it works.
Philippos Aristotelous is a law firm adviser who designs and runs performance-management systems inside law firms in Cyprus and Greece. The method described on this page is the one he uses: no overall score in the first year and nothing tied to pay, a short self-reflection before the meeting, one goal under each of the Five Pillars of Modern Legal Performance that applies to you, and a written follow-up copied to the managing partner. The rest of this page explains what that looks like in a real firm, what it costs the people who run it, and where it goes wrong.
That is me, so I will tell it as it happened. In January, Andreas Kyprianou wrote an action list after a partners’ meeting. He runs a twenty-six-person firm in Nicosia: litigation, corporate, property and employment. Item four read “set KPIs for staff”. The firm had never held an appraisal. There was an HR policy in a folder, borrowed from a larger firm years earlier, with a career ladder, a bonus formula and a thirty-criteria appraisal form, and nobody had ever applied any of it.
I wrote back and said: not yet.
Why do law-firm appraisals fail?
Usually because the manager’s half of the form is blank. Years ago I ran a properly scored system at a firm I worked with earlier in my career, details changed here: a five-point scale, a threshold for every indicator, weights, mid-year and year-end. When I went back through the archive, the employee half of every form was full and the manager half was mostly empty. The people meant to fill the forms in were busy; the forms were fine.
This is not a Cypriot habit. James Wilber of Altman Weil wrote in 2004 that “although most partners understand the importance of providing formal feedback, many do not devote the necessary effort to making the process meaningful.” The advice in the same piece was blunt: “If you promise associates that they will be evaluated, then do it and do it on time.” I still see the same pattern twenty years on.
The research underneath this is older than most appraisal forms. Kluger and DeNisi’s 1996 meta-analysis, covering 607 measured effects of feedback, found that feedback improved performance on average but that over one-third of feedback interventions made performance worse. Feedback is not neutral. Roughly a third of the time it costs you.
Should a law firm start with KPIs and ratings?
Not in the first year, and I would say that to most firms under fifty people. In my experience full scoring systems were built by much larger firms, over years of redesign, and even they struggle with them. A firm with no appraisal culture that starts with a rating scale gets one question in every meeting, asked before anything else: “What does this mean for my pay?” Once that question is in the room, the conversation about the work is over, and a busy manager’s hour has gone on the number instead.
So the first year is a pilot. Everyone reflects on their own work under the same headings. Everyone has one prepared conversation with their head of department. Everyone leaves with one goal per heading and an email confirming it. No overall score, no bonus link, no promotion link, and the firm says so in writing. The words are “at this stage”, because a link can come later, once managers can actually give feedback.
Andreas agreed, and said something I have heard from several managing partners since. He preferred constructive feedback and personal goals to numbers. He had just assumed a consultant would insist on them.
What has to exist before you can appraise anyone?
Reporting lines. You cannot appraise a lawyer if nobody can say who her manager is, and in a founder-led firm that has grown quickly the honest answer is often “whichever partner gave her the file”. When I ran a confidential questionnaire with Andreas’s staff before designing anything, the main thing that came back was role confusion, not performance: people unsure where their responsibilities ended, instructions arriving from three directions on the same matter, files waiting on a decision nobody owned.
So in week ten, before any form, we wrote down reporting lines. Every person got a named reviewer and a named person that reviewer escalates to. In a small firm your day-to-day boss, your reviewer and your head of department are usually one person. That is exactly why nobody bothers to tell them apart on paper, and the one time they differ, you will be glad you did.
One practical note: we attached the reporting lines to a policy people actually wanted, because it needed a department head’s sign-off to operate. The structure was in use for two months before anyone experienced it as a hierarchy.
Why does the self-assessment come before the meeting?
Because it produces a gap, and the gap is the meeting.
The headings are the four that apply to everyone at the firm, Financial Focus, Client Focus, Learning & Growth and Internal Collaboration, plus People Leadership for anyone who runs a team. I have written about what they measure in The Five Pillars of Modern Legal Performance. One design decision matters here: for an associate, Financial Focus is about behaviour, not a billing target. It asks whether you agree scope before you start, bill when the milestone is reached, chase at day fifteen and day thirty, say something when the job has grown past what was agreed, and bring in any work of your own. The revenue figures stay with the partners, where they always were, and the pay conversation stays there too.
In week eighteen every person at the firm received a short online form: ten statements per heading on a one-to-five scale, one or two open questions, and a proposed goal under each heading. The form says ten minutes. The scores are there to make you think. The open questions are where it bites: How do you know your clients are satisfied? Give one or two concrete examples. If you ran your own desk as a business, what would you do differently? Could a colleague pick up any file of yours tomorrow and know where it stands? Nobody answers those honestly in ten minutes, and the people who gave them an hour got the most out of what followed. Wilber’s 2004 line applies: “A self-appraisal by the associate should be required.”
Christos Pavlou, a fourth-year litigator, rated himself five out of five on client focus and four on internal collaboration. Maria Constantinou, who heads the litigation department and reviews him, separately rated him three and two. On learning and growth she rated him higher than he rated himself.
Christos did not see Maria’s ratings before the meeting. That is a rule, not a courtesy: a rating seen beforehand kills the conversation before it starts. He saw them afterwards, so nothing sits in a drawer he cannot open.
What do you do when the lawyer rates themselves higher than the manager does?
You think about context before you conclude anything, and then you open with the gap as a question. Maria’s preparation notes, printed and annotated by hand, said: new to running his own hearings this year; overloaded since March; nobody ever showed him the firm’s file-note convention. In the room she asked him to talk her through the five on client focus. He described two clients who had said thank you. She described a third who had rung her, not him, twice in April, because he had not replied. They spent twenty minutes on it. He did not agree with the two on internal collaboration until she opened one of his files on the screen and asked him where it stood. He could not say from the file alone.
That is the conversation the whole system exists to produce. In a firm that has never held an appraisal, finding out that two people hold completely different pictures of the same job is the first year’s work. There is still a verdict at the end of the year, and the supervisor has the final say. What we removed was the verdict before the conversation.
How do you write a performance goal that can be checked?
First, have few of them: one under each heading, four for most people, five for team leaders. The written goals are deliberately few. The conversation covers the whole heading: what is improving, what is getting harder, what the person needs to learn. Few goals on paper, because paper is what you are held to in December.
The classic bad goal is the one Christos proposed on his form in May, under Financial Focus: “improve business development”. Every associate in the world has written it. It is not a goal, it is a mood. Nobody can tell at year-end whether it happened, so nobody has to admit it did not.
A goal that works says when, in what area, something a person could watch you do, and how often. Then the part everyone forgets: a second visible act that leaves a trace somebody else can see. Christos’s agreed goal was: “This year, for the three insurance brokers I already act for, one written follow-up after each completed matter, and a two-line note to Maria on what came back.” Now Maria can tell whether it happened without asking him. A good deal of legal work has no number in it, so the test is observable rather than measurable, and if you cannot picture the year-end conversation about a goal, it is not a goal.
One instruction matters more than the rest. Everyone keeps a running personal file through the year: matters, emails, incidents, examples, filed against their goals. Then December rests on evidence, not impressions. Wilber gave the reviewer’s half of the same instruction in 2004: “Keep a log on performance issues for each associate you will be evaluating with examples of good and not so good performance.” It is the cheapest thing in the whole system and the one most often skipped.
How should a department head prepare for a performance meeting?
Maria received a pack a week before her meetings: her seven people’s completed forms, in confidence, and very practical instructions. Print them. Read them with a pen. Mark every place your view differs from theirs, higher or lower. Pick one or two points per heading, not every line. Think about context before you conclude.
In that same week she and I went through her team list person by person and agreed the one message each of them should leave the room with. Not five. One. If you cannot say in a sentence what this person most needs to hear, you are not ready to hold the meeting.
The honest cost, for a head with seven reports, is about two days in the first year, three if she is conscientious. That includes the reading, the pre-brief, seven meetings of forty minutes to an hour, and seven emails. Set it against the inherited thirty-criteria form with its rating column, which nobody ever filled in, so it cost zero hours and delivered zero management. Two days a year to actually manage seven people is the job.
What happens after a law-firm performance meeting?
Within days, one email per person. It opens with the person’s specific strengths, in Maria’s words. Then the agreed goal under each heading. It says this is a working document, it can change, and it is the basis of the year-end check. Andreas is copied.
A goal the managing partner has seen is a commitment. A goal in a drawer is a wish.
Everything after that email is making it stick. The mid-year check, in week thirty-seven, is short, informal and unscored: are the four goals still right? Christos’s broker goal survived. His internal-collaboration goal was rewritten, because the firm reorganised how files were recorded over the summer and the original wording no longer made sense. That is what the mid-year check is for. It is where you rescue a goal instead of failing it in December.
Two things came out of Maria’s seven meetings that nobody had planned: a departmental procedures note, because three of her people had asked for the same thing and had never said so in a room where saying it was expected, and a weekly twenty-minute case meeting. Both had been wanted for a year. McKinsey’s 2024 survey of over a thousand employees found that only 21 per cent of those who had no development conversations felt motivated by their employer’s performance management. For those who got ongoing feedback it was 77 per cent. That number is about conversations, not forms.
What goes wrong when a law firm introduces performance conversations?
Resistance in a law firm does not argue. It reschedules. Every time I have introduced this there has been no open objection from staff. What there is: zero forms a few days before the deadline and nearly all of them in by it; heads postponing the pre-brief because a hearing came up; goal emails a week late. The partners’ own meetings are the last in the diary and drift into the holidays. The last reflection in is usually the most senior person’s, and everybody notices. Someone has to say that to them. In Andreas’s firm, that was me.
Someone will also try to bolt money onto it, usually whoever runs the bonus spreadsheet, usually with the best intentions. Not this year. The tracking system I built for the firm has a pilot mode that will not produce a score at all. You can flip that switch. You cannot flip it quietly.
And the process has to be visibly the firm’s. I drafted the all-staff email; it went out under a department head’s name, not mine. I drafted the format of the goal emails; the heads wrote them. If the staff think it is the consultant’s process, it ends when the consultant leaves.
Where does AI fit in law-firm performance reviews?
Preparation, not judgement. The laborious part of a good conversation is assembling the evidence, what matters the person handled, which deadlines were met, what was invoiced when, and that lives in the case-management system where nobody has time to pull it together. That is the job for AI: the personal evidence file made automatic, and a check on whether a draft goal leaves a trace. What it must not do is rate people, rank people, decide pay or replace the conversation. Reflections and reviewer notes are about the most sensitive data a firm holds. So where the tool runs, and who can see it, is a confidentiality question first.
Illustrative example. The firm, the people, the ratings and the goals are invented, and no real firm or matter is described; the earlier scored system is likewise described with details changed. The method, the sequence and the friction are not invented: they are drawn from performance-management work inside law firms in Cyprus and Greece, with details altered and combined so that no single firm is depicted. Sources: Kluger & DeNisi, “The effects of feedback interventions on performance: a historical review, a meta-analysis, and a preliminary feedback intervention theory”, Psychological Bulletin 119(2), 254–284, 1996; James Wilber, “Conducting Effective Associate Evaluations”, Altman Weil Report to Legal Management, July 2004; Komm, Weddle & Maor, “What employees say matters most to motivate performance”, McKinsey, August 2024.
Related reading: the Five Pillars of Modern Legal Performance, on what the headings measure and why; three hours of decisions, thirty-one days of waiting, on the reporting-line problem from the partner’s side; and legal business consulting, which is where this work sits. If your firm’s appraisal system produces a score but not a conversation, book a call.