By the time the average annual review lands, the information inside it is 10 months stale.
Most organisations already know this. Yet the system persists, because it is fused to compensation, because HR calendars drift, and because nobody wants to carry the political cost of the switch.
The good news: you do not have to blow up the annual review in one go. You can evolve to a quarterly calibration rhythm and quietly retire the annual ritual inside two years. The better news: the quarterly version is easier to run, produces sharper talent decisions, and is more defensible.
Why the annual review fails
It is not the manager's fault. The structural problems are:
- Recency bias. 60% of the evidence comes from the last 60 days.
- Forced calibration in one session. Trying to rank 30 people against each other in a 90-minute meeting is cognitively impossible.
- Compensation link. Ratings get nudged to fit the pool, which erodes trust.
- Zero coaching value. By the time feedback lands, the behaviour is either sunk or already changed.
- Inflation creep. Without a calibration discipline, ratings drift up year on year until they are meaningless.
What quarterly calibration replaces it with
The quarterly version is not "four times as much work". It is the same work, redistributed, and it replaces the painful annual cycle with a rhythm that is easier to sustain.
Each quarter, three inputs land for every employee:
- One short self-assessment on two to three focal capabilities (not the whole model, see the capability framework template for how to scope focal capabilities).
- A single-page manager note capturing one observable strength and one growth area, with examples.
- A 60-minute calibration session where peer managers sit together and compare calls.
The session produces a light-touch quarterly rating, tracked over time. You keep one formal moment a year where the four quarterly ratings are summarised into an annual picture, but that summary writes itself, because the work already happened.
How to sequence the transition
The shift works best when staged over four cycles.
Quarter 1. Keep the annual review in place. Introduce the quarterly snapshot in parallel, with no compensation tie. Call it a "development check-in". Goal: build the habit without political risk.
Quarter 2. Run peer calibration sessions for the first time. Two rules: peer managers sit side-by-side, and everyone defends one level call with a specific example. Goal: expose vague descriptors and lazy assessments early.
Quarter 3. Use the quarterly snapshot data to sanity check the mid-year adjustment cycle if you have one. Start socialising the direction publicly.
Quarter 4. Use the four snapshots to build the annual write-up. Managers report that the annual narrative now writes itself in 20 minutes. Start reducing the weight of the annual rating inside comp decisions, bring quarterly data in as the primary input.
Year 2. Retire the standalone annual review. Replace with a year-end synthesis conversation that draws on the four quarterly snapshots.
The rules that make it stick
Four non-negotiables:
- No forced distribution. Quarterly calibration works when managers are trying to describe reality, not match a curve.
- Peer calibration every quarter. Without peers in the room, managers drift back to their own biases inside two cycles.
- Keep the rating scale short. Three or four levels, plain English. "Off track / on track / ahead" works surprisingly well.
- Never combine the review conversation with the comp conversation. Split them by at least 30 days. The moment people smell compensation, they stop hearing feedback.
Why managers actually prefer it
After two cycles, managers we work with describe three changes:
- Year-end writing takes a quarter of the time (the notes are already there)
- Regretted exits drop, because issues surface in month three rather than month ten
- Promotion decisions feel defensible, because there is a trail of evidence
None of this is magic. It is what good performance management has always looked like, broken into pieces small enough to actually happen.
Where to start
Do not announce a system change. Run one quarter of calibration sessions in one function as a pilot. Measure three things: manager time, quality of the output, and employee sentiment. If all three move the right way, expand.