Behind every evaluation form sits a set of competencies, whether or not the firm has written them down. When they are not written down, each reviewer uses their own, which is the usual reason two partners rate the same associate three bands apart and both believe they were being fair.
Why a general-industry framework does not transfer
Borrowed frameworks tend to arrive with competencies like strategic thinking, stakeholder management and driving results. None is wrong, and none describes what a fourth-year litigator does in a week.
Law firm work has features a general model does not account for. Work arrives through matters rather than through a reporting line, so an associate is assessed by several partners with partial views. Quality is judged against what the matter required rather than against a fixed standard. And progression is up-or-out on a known timetable, which means a framework has to say what adequate looks like at each year rather than only what good looks like in general.
The domains that matter
Six to eight competencies is the working range. Fewer and the ratings carry no diagnosis; more and reviewers stop reading the anchors.
- Technical quality. Work product measured against what the matter required, including whether it needed rework and how much.
- Legal judgment. The quality of the call when the answer was not in a precedent, and whether the associate knew to escalate.
- Ownership and reliability. Whether the thing got done without being chased, and whether problems surfaced early or late.
- Written and spoken communication. Including whether a partner could send the draft on with light edits.
- Client handling. Scoped to what the associate is actually exposed to at their level.
- Supervision and development of juniors. This starts mattering years before anyone is formally responsible for it, and it is the competency most frameworks add too late.
- Practice contribution. Business development, knowledge, recruiting, pro bono, scoped realistically rather than imported from the partnership criteria.
The part most frameworks get wrong: seniority
A single scale applied across all years produces a predictable artefact. Junior associates cluster low and senior associates cluster high, and the ratings describe tenure rather than performance.
Each competency needs a different expectation at each stage. For technical quality, a first-year is expected to produce work that is accurate and needs editing; a fourth-year to produce work a partner can send on with light edits; a senior associate to produce work that is ready and to catch what the junior missed. Those are three different descriptions of a 4, and a framework that does not say so is asking reviewers to invent the standard.
In practice this means a grid: competencies down one axis, bands of seniority across the other, with a short written expectation in each cell. It is more work to build, and a framework without it leaves every reviewer deciding privately what a 4 means.
One competency, worked across three bands
Technical quality is the easiest to illustrate because every firm rates it and few firms define it by year. A usable cell says what the work looked like, not how good it was.
- First and second year, meets expectation. Research is accurate and the authorities are current. Drafts follow the precedent and the house style. A partner edits for structure and judgment, not for errors. The associate flags what they were unsure about rather than leaving it to be found.
- Third and fourth year, meets expectation. Drafts arrive in a form a partner can send on with light edits. The associate identifies the issue the instructions did not mention. Research is scoped to the question rather than exhaustive, and the memo says what to do rather than what the law is.
- Fifth year and above, meets expectation. Work is ready. The associate catches what the junior missed before the partner sees it, decides what does not need escalating, and takes a position where the authorities are unsettled rather than presenting both sides.
Three different descriptions of the same rating, which is the point. An associate in their second year producing the third description is exceeding expectation, and a firm with one undifferentiated scale has no way to record that.
How to write a behavioral anchor
An anchor describes what somebody did, not what they are. Three rules make them usable.
Write them in the past tense about observable actions. "Raised the limitation issue before the filing deadline" rather than "demonstrates commercial awareness".
Anchor the ends and the middle, not every point. Describing a 2, a 3 and a 4 gives reviewers enough to place somebody. Describing all five invites them to read none.
Make the middle band acceptable in its own right. If a 3 reads as a failure, everyone gets a 4, and the scale collapses to two usable points. A 3 should describe an associate performing at the level their year expects.
Building it with the firm rather than for it
A framework written by one person and circulated for comment gets approved and ignored. The version that holds is built from what the firm already rewards.
A practical route: take eight to ten associates the partnership agrees are performing well at different seniorities, ask the partners who staffed them what specifically they did, and group the answers. The competencies come out of that, in the firm's own language, and the anchors come with examples attached.
Check the result against the exit data as well. If departing associates consistently cite something the framework does not measure, which supervision quality often is, the framework is describing what the firm notices rather than what determines whether people stay.
What it connects to
The framework is the thing evaluation forms rate against, the thing calibration compares reviewers on, and the thing a promotion decision should be able to point at. If a firm cannot say which competency a partnership decision turned on, the framework is not doing its job regardless of how well written it is.
It also has a shelf life. Practice changes, and a framework written before a technology shift or a change in client staffing expectations will keep measuring what used to matter. Revisiting it every two or three cycles is enough.
SRA builds competency frameworks with firms as part of instrument design, calibrated to practice area and seniority, and administers the upward reviews and 360s that rate against them. The methodology covers how the resulting data is held and reported.