Continuous Performance vs Annual Ratings: The Honest Trade-Off

In 2012 Adobe announced it was eliminating annual performance ratings. GE followed in 2015. Deloitte's 2015 HBR article 'Reinventing Performance Management' set the orthodoxy: ratings were biased, demotivating, and obsolete. A decade later, the picture is messier than the headlines suggested. The decisions that matter — promotions, pay, layoffs — still require comparative judgments about people, and continuous feedback alone does not produce them.

The 2010s movement and what actually happened

Mercer's Global Performance Management Study (2019) and Bersin by Deloitte's 2020 follow-up both found that the majority of companies that publicly eliminated ratings had quietly reintroduced some form of comparative evaluation by 2020 — often labelled differently ('contribution levels', 'impact tiers') but functionally similar. The reasons were practical: compensation committees needed differentiated inputs, layoff decisions needed defensible documentation, and high performers wanted to know where they stood.

What continuous performance does well

• Reduces the 'no surprises at review time' failure — feedback in the moment beats feedback months later.
• Improves development conversations by separating them from comp negotiation pressure.
• Surfaces problems early enough to act — PIPs become rarer because issues are addressed before they compound.
• Reduces recency bias by aggregating multiple feedback points rather than relying on one annual conversation.
• Aligns better with project-based and short-cycle work, where annual cadence is too slow.

What ratings (or substitutes) still do

• Force calibration across managers — without comparative scoring, leniency bias goes unchecked.
• Anchor compensation decisions in something other than 'the squeaky wheel'.
• Document defensible bases for promotion, demotion, and termination decisions.
• Help high performers see where they stand — opacity often demoralises top talent.
• Provide the comparative basis for talent reviews and succession discussions.

The hybrid model

Implementation pitfalls

1. Eliminating the annual artifact entirely — the comp committee still needs an input.
2. Calling tiers 'contribution levels' to avoid the word 'rating' — fooling no one, including the employees.
3. Continuous feedback without calibration — leniency drift kicks in within 2 cycles.
4. Tooling without ritual — buying Lattice or Culture Amp without training managers on the conversation it enables.
5. Decoupling pay from performance entirely — high performers leave when contribution and reward visibly disconnect.

Frequently asked questions

Do we still need annual performance reviews?

You need an annual artifact that documents the year and supports compensation and promotion decisions. Whether you call it a 'review' or a 'summary' is branding. The artifact has to exist.

Can we eliminate ratings entirely?

You can eliminate the label and the 1-to-5 number. You cannot eliminate the underlying comparative judgment — your comp committee, your promotion committee, and your potential layoff decisions all require it. Most companies that 'eliminated ratings' simply renamed the comparative evaluation.

How often should formal performance conversations happen?

Quarterly is the modern default. Monthly is too frequent (signal-to-noise drops); annual is too rare (recency bias dominates). Quarterly check-ins with continuous in-the-moment feedback in between is the rhythm that produces both development and documentation.

Should the development conversation be separate from the compensation conversation?

Yes. Even within the same review cycle, separate the two conversations by at least a week. Comp pressure crowds out development discussion; mixing them produces a worse version of both.

What is calibration and why is it non-negotiable?

Calibration is a meeting where managers review their proposed ratings or summaries together to remove rater bias. Without it, leniency bias dominates — every manager rates their reports above average, and the system loses its information value. It is the single most important ritual to preserve when redesigning performance management.

Where to read further