A first-year associate arrives in September, works for four months, and is evaluated on an instrument designed for somebody with a year of comparable output behind them. The ratings come back compressed and uninformative, which everybody expects, and the exercise is treated as a formality.

It is also the associate's first direct experience of how the firm assesses people, and they draw conclusions from it that last.

Why the annual cycle does not fit

Three reasons, all structural.

First-years arrive together but start producing at different times, depending on practice group, matter flow and whether they landed on something live in week two. Four months in, two associates in the same cohort can have had entirely different amounts of assessable work.

They also improve fast enough that a rating is stale almost immediately. An assessment of somebody's drafting in October describes a different lawyer by January, which is not true of a sixth-year.

And the people who know most about a first-year's work are often not partners. The senior associate who reviewed the research memo and sent it back twice has the observation the evaluation needs, and most forms do not ask them.

What to look at, and when

At 30 days, nothing about performance. The only useful questions are about whether the arrangement is working: do they know who to ask, have they been given work, do they understand what the matter is for. A performance judgment at week four measures how quickly somebody found their feet in a building.

At 90 days, work habits rather than legal judgment. Responsiveness, whether they flag problems or sit on them, whether the work arrives when they said it would, whether they ask before going a long way in the wrong direction. These are observable this early and they predict a good deal.

At 180 days, the first real assessment. Work product against what the matter required, with the seniority expectation written for a first-year rather than borrowed from the general scale. This is also the first point at which a self-assessment is worth collecting, because the associate now has something to assess.

Who should be asked

For a first-year, the rater pool should include the senior and mid-level associates who actually supervised the work, not only the partners on the matter. This is the single change that most improves first-year evaluation quality, and it is usually resisted because the firm's form has a partner field and no other.

It has a second benefit. Asking senior associates to assess juniors surfaces who is good at it, which is the competency most frameworks start measuring years too late.

What the checkpoints are really for

Partly development. Mostly retention.

Across SRA's benchmark, engagement runs at 82 in the first year and 68 in the second. The first year is the high point, and what happens in it determines how far the fall goes. The signals that predict an early departure are visible in the first six months: work that never arrived, a supervisor who was never available, a sense that nobody had noticed either.

A 180-day checkpoint that asks the associate about those things, and not only the firm about the associate, is the cheapest retention instrument a firm has. It is also the one most likely to be cut when the cycle is busy.

What onboarding should surface that evaluation cannot

An evaluation asks how somebody is doing. The questions worth asking a new associate are different: whether the work they were promised at recruitment resembles the work they are doing, whether they have had any contact with the partners whose practice they joined for, and whether the firm has done what it said it would.

Those answers are uncomfortable and they are the ones that explain first and second-year attrition. A firm that discovers a recruiting-promise gap at the exit interview has learned it from the only person who can no longer act on it.

SRA runs milestone check-ins of exactly this shape for lateral partners, at 30, 90 and 180 days, and the same structure applies to an associate intake. Engagement measurement covers the cohort view, and the tenure figures above are listed with their definitions on figures and sources.