How do I run effective performance reviews and set clear expectations?

The review is the last step, not the first

Most founders who ask about performance reviews are really asking about expectations. A review is a comparison, and if there is nothing credible to compare against, the conversation collapses into impressions, recent memory and whoever was most visible that quarter.

Start instead with what the role is accountable for. In professional services the most damaging version of vagueness has a name: role corruption, where the job was never clearly defined and the strong performers quietly absorb the work of their coworkers. When Greg Alexander team ran diligence on a site selection boutique with forty percent annual turnover, they contacted former employees to find out why they had left. Role corruption came up repeatedly. The stars were doing other people work and burning out. The annual reviews, in the same firm, were described by former employees as a joke, a compliance-driven exercise. What they wanted was real feedback and they never got it.

That firm did not have a review problem. It had an expectation problem that the review process then failed to catch.

Cascade the targets until a frontline employee can see themselves in them

The mechanism that connects firm strategy to an individual review is a cascade. Buyers assess it directly. During diligence they will spend time with the leadership team and one layer below, and they are looking for a crisp, data-supported strategy, evidence that each person understands their role in it, and cascading targets that reach all the way to frontline employees.

This is a useful test to run on yourself before anyone runs it on you. Ask a random sample of your people what the firm goals are, how they personally are trying to help achieve them, which values win when a tradeoff has to be made, what behavior gets you hired here, what gets you promoted, and what gets you fired. In a start-up you get the same answers from everyone. As a firm scales you start getting different answers, and that divergence is the culture eroding in real time.

If the answers diverge, the review process cannot fix it, because there is no shared standard for the review to apply. Fix the cascade first.

Set the standard as capability, not as a score

Rating scales invite comparison between people. Capability standards invite comparison between a person and the work.

The practical way to build one is to work backwards from engagements you have actually delivered. Take a representative sample, break each project down, and identify the specific knowledge each required. Then look at how the work was performed at the task level and inventory the skills needed to perform each task. Convert that into an assessment, and categorize people against it. Alexander recommends the 101, 201, 301 model used in academia: below 101 means the upskilling effort is too large, 101 is junior, 201 is midlevel and 301 is expert.

Two things follow, and they are the reason this beats a rating form. The baseline tells you who can do what, which is what you consult when staffing engagements. And it puts every person on a named learning path with a defined endpoint, so development stops being a conversation about attitude and becomes a conversation about a specific gap.

It is also the objective evidence a review needs in order to be believable. Telling someone they are not ready is an opinion. Showing them which knowledge and skills the next band requires, and which of them they have demonstrated, is a plan.

Review what the firm actually needs, including how clients feel

Most review templates in professional services measure delivery quality and utilization, and stop. That leaves out the thing sophisticated clients increasingly buy on.

Quality is measured by the finished product: does the site function, is the analysis sound. Service is measured by how the client feels while working with you: do they understand what you are doing and why, do they feel like a priority or like one more account. Quality work is close to a commodity in most categories. Far fewer firms deliver outstanding service alongside it.

The way to make that real rather than aspirational is to put client experience feedback into performance reviews, and to let bonuses and promotions go to the people who excel at it. Behavior changes when it is measured and rewarded, and not before.

Be careful what else you measure, though, because you will get exactly what you score. If you reward responsiveness, you will get over-servicing. If you reward visible effort, you will get busyness. If you reward being indispensable, you will get people who hoard context and resist developing anyone behind them. For anyone who leads a team, one of the more honest measures available is the growth rate of the people they lead: teams that stagnate under a manager are usually under-led rather than unlucky.

Cadence, and why annual is usually wrong

An annual review in a fifteen-person firm is a ceremony. The information it delivers is stale, the recency bias is severe, and the employee has already decided how they feel long before it happens.

The useful pattern is to separate the conversations. Coaching runs continuously and is about the work in front of the person. Capability assessment runs on a defined cycle and is about movement between bands. Compensation is decided on its own schedule, so that the coaching conversation is not silently a negotiation.

The other cadence worth building is the post-project review, conducted by someone who was not on the project team. They interview each team member and review the objectives, profitability, timelines, budgets, deliverables and adherence to standard operating procedure. Done for every project, it accumulates into an archive that tells you what actually happened rather than what people remembered at review time.

What Era 3 changes

In Era 1, talent decisions were made under pressure, with no forward visibility. Development was informal, promotion lagged behind the responsibility people were already carrying, and reviews happened when there was time. Era 2 digitized the paperwork. Performance management tools formalized the review, engagement surveys gave employees a voice, and HR looked more sophisticated. The outcomes did not move. Reviews became more structured while promotion decisions stayed late and risky, and the presence of data was mistaken for insight.

What changes in Era 3 is the input to the conversation rather than the conversation itself. Promotion readiness can be treated as a signal that accumulates from performance and skill progression rather than a judgment assembled once a year. Attrition risk and burnout signals can surface early enough to act on. Time-to-productivity can be modeled by role and skill set, which tells you whether a development path is working. Succession risk in critical roles becomes visible while intervention is still cheap.

The division of labor is the point. The analytical, continuous work moves. What stays human is judgment, tradeoffs, cultural reinforcement, coaching and accountability. Nothing about Era 3 makes the difficult conversation easier, and nothing about it should make that conversation automated. Employees can tell the difference between feedback and a generated summary, which is the same instinct that made them dismiss the compliance-driven annual review in the first place.

When this answer flips

In a firm of eight people where the founder works alongside everyone daily, a formal review process adds ceremony without adding information. What is still required is the expectation: each person should be able to say what they are accountable for. Write that down and skip the form.

Reviews are also the wrong tool for a problem that is not about performance. People leave over pay they know is below market, over benefits that put too much of the burden on them, and over a firm with no purpose they believe in beyond making the owner wealthy. Alexander notes that almost everyone who left the site selection boutique landed a better-paying job, and the employees knew they were underpaid. A better review process would not have retained a single one of them.

And if the review process is being introduced to build a case for letting someone go, say so to yourself honestly. That is a legitimate need, but it is documentation rather than development, and running it as though it were development damages trust with everyone watching.

The short answer

Define the expectation before you build the review. Give every role a clear accountability so that strong performers are not silently absorbing other people work, and cascade the firm targets down until a frontline employee can explain how their work moves them. Set the standard as demonstrated capability, built backwards from engagements you have actually delivered, so the review compares a person to the work rather than to their colleagues. Measure what the firm needs, including how clients feel and not only what was delivered, and be careful, because you will get exactly what you reward. Run coaching continuously, assess capability on a cycle, and keep compensation on its own track. Use post-project reviews conducted by someone outside the team to build a record of what actually happened. In Era 3 the signals that feed the conversation can run continuously, but the conversation itself stays human, and a review people do not believe is worse than no review at all.