How to make peer feedback useful, not awkward

Peer feedback is usually the most valuable input a firm collects and then wastes. The colleagues who pair on a build, share a client call, or carry the slack when someone coasts see the day-to-day: reliability, craft, how a person works on a team. A rotating lead never witnesses that. It turns awkward and useless for two fixable reasons: no clear purpose and fear of consequences. Fix both. Give peer feedback one job, helping a colleague grow, and keep judgment out of it. Ask specific behavioural questions instead of "rate your peer 1–5." Collect it from the people who worked together while the engagement is fresh. Keep it separate from the rating so nobody sanitises or colludes, and make sure the person turns it into an action. Structured that way, peers give honest, rich, useful feedback and stop dreading the form. The awkwardness comes from the design, which means you can change it.

The stakes are in the numbers. A single reviewer's read is unreliable: about 62% of rating variance is the idiosyncratic rater rather than the person (Scullen, Mount & Goff, Journal of Applied Psychology, 2000). That is why the independent, close-up viewpoint of a peer matters. Feedback still isn't automatically useful. Across 607 studies it lifted performance only modestly (d ≈ .41), and over a third of interventions made performance worse (Kluger & DeNisi, Psychological Bulletin, 1996). Purpose and delivery decide which side of that line your peer feedback lands on.

Why is peer feedback worth the effort?

Because peers are the closest witnesses to the work. In a firm of rotating leads, they are often the only consistent observers of how someone operates day to day. A partner or engagement manager sees the visible surface: the deck, the client outcome, the milestone. The peer sees the process: who unblocked the junior at 6pm, who quietly redid a sloppy handoff, who went silent when the scope slipped.

That close-up view matters more than it looks, because the rater's own idiosyncrasies dominate one reviewer's rating: about 62% of the variance is the rater rather than the person (Scullen, Mount & Goff, Journal of Applied Psychology, 2000). A second, independent vantage point gets you nearer the truth. Peers also cover a blind spot leads can't reach: collaboration, generosity, reliability, the behaviours that make a team fast and never show up in a solo deliverable. Waste that input and you're rating people on half the evidence.

Why does peer feedback so often turn awkward or useless?

For two fixable reasons: no clear purpose, and fear of consequences. When people don't know what the feedback is for, they default to bland ("great to work with!") or generic. When they suspect it will feed a rating, a bonus, or a partner-track decision, two failure modes appear. They sanitise, because nobody wants to hurt a colleague they'll be staffed with again next quarter. Or they collude and retaliate: you protect my score, I'll protect yours. Either way the signal collapses into noise.

Feedback carries risk even when it flows. Over a third of feedback interventions reduced performance (Kluger & DeNisi, Psychological Bulletin, 1996). Candour also depends on safety, and people withhold when speaking up feels risky (Edmondson, Administrative Science Quarterly, 1999). Drop an anonymous 1–5 form onto busy, billable people and you get what you'd expect: thin, guarded, or weaponised feedback. Peers aren't cowards. Nobody told them what the form was for, and they are afraid of what it will do.

Developmental or evaluative: what should peer feedback be for?

Developmental. Give it one job: help a colleague grow. Once peer input feeds the rating, comp, or ranking, honesty gets expensive and people manage the score instead of telling the truth. Keep peer feedback on the development side of the line, and leave the grade to the calibrated cycle owned by the leads who staffed the work. Feedback and appraisal are different jobs, and fusing them makes candour costly (DeNisi & Murphy, Journal of Applied Psychology, 2017). Developmental feedback can still be blunt. It aims at next time and leaves the verdict on last time to the leads.

Don't expect the form itself to change anyone. After multisource feedback, measured improvement is small across every source: largest from direct reports, smaller from peers, close to nil for self-ratings (Smither, London & Reilly, Personnel Psychology, 2005). Coaching and a concrete goal afterwards move performance. Feedback is receiver-driven, so it works only when the person can take it in and use it (Stone & Heen, Thanks for the Feedback, 2014). Peer feedback earns its keep when someone turns it into an action.

What questions produce useful peer feedback?

Behavioural ones. Ask what the person did and what it was like to work with them, and skip the verdict or score. "Rate teamwork 1–5" produces noise: people rate others' abstract traits unreliably, and most of that score reflects the rater rather than the person (Buckingham & Goodall, "The Feedback Fallacy," Harvard Business Review, 2019). Ask instead about the rater's own experience of the work.

1. What should they keep doing? The specific strengths to build on, each one named. "Good team player" doesn't count.
2. What would make working with them easier? The one friction, framed forward. It's the single most useful thing a peer can offer.
3. Describe one moment. Anchor to a single observable situation and describe a behaviour instead of attaching a label (Center for Creative Leadership, SBI model).
4. Where did they help you or the client? Surfaces the invisible collaboration a lead never sees.
5. One thing for the next engagement. Forces a single, actionable takeaway instead of a report card.

Behavioural, first-person, specific. That's the difference between "he's a 4/5 on communication" and "he rewrote the client email so the junior wasn't thrown under the bus. Do that earlier and louder."

When, and from whom, should you collect it?

From the people who worked together, around the moment an engagement wraps, while memory is fresh. Skip the blanket, everyone-rates-everyone blast, which yields thin feedback from people who barely overlapped.

1. The right people. The two or three colleagues who collaborated on the engagement: paired, handed off, shared the client. Proximity is the whole point. Someone who shared one stand-up isn't a witness.
2. The right moment. At or just after the engagement debrief. Eleven months later at annual review is too late. Detail decays fast, and captured fresh it stays specific.
3. The right amount. A couple of short, focused requests beat a mass survey. Multisource feedback makes a difference only when it's done sensitively, with questions that are short, clear and relevant to the job (CIPD, 360-degree feedback).
4. Once per engagement. Capture a small note each time so it accumulates into the growth picture and nobody has to reconstruct, or invent, it under deadline.

Signed or anonymous: how do you keep it safe without breeding venting?

Signed by default, and protected only when there's a real reason. Anonymous-by-default feels safe, but it breeds venting and score-settling. In a small agency team, anonymity is a fiction anyway, because everyone knows who was on the engagement. Attributable feedback that the person sees builds trust and forces the giver to be constructive. You write differently when your name is on it.

Signed feedback feels safe when there is psychological safety, the shared sense that honesty won't be punished (Edmondson, Administrative Science Quarterly, 1999), plus the guarantee that the feedback is developmental and won't be fed into someone's rating. You don't need to hide names. Keep an escape hatch: a confidential channel to raise a serious concern (harassment, ethics) with a lead. Don't make anonymity the default for everyday growth feedback.

1. Signed and shared with the person → trust, accountability, and better-written feedback.
2. Anonymous-by-default → venting, sanitising, and nobody owning their words.
3. A separate confidential channel → for genuine concerns only, kept apart from growth feedback.

How do you turn peer feedback into action across rotating project teams?

Here peer feedback becomes part of how the firm grows people. In an agency or boutique, people are staffed across engagements under different leads, rated per engagement, and no single reviewer saw the whole year. The peer's close-up view is often the most continuous record you have. Make it operational:

1. Capture per engagement, lightly. Two or three behavioural notes from real collaborators at each debrief: fresh, specific, low-cost. That cadence makes peer input comparable at calibration and spares you a once-a-year scramble.
2. Roll it into the growth picture, away from the rating. Peer themes feed the person's development conversation. The calibrated grade stays with the leads who staffed the work (DeNisi & Murphy, Journal of Applied Psychology, 2017).
3. Close the loop into one or two actions. Feedback pays off only when someone acts on it. The follow-up moves performance, and the form on its own does little (Smither, London & Reilly, Personnel Psychology, 2005). End every round with a concrete next step, and check it at the next engagement.
4. Weight by proximity, and read divergence as signal. The colleague who paired daily saw more than the one who shared a stand-up. When two peers disagree, they're telling you something about context. Don't average it away.
5. Protect the staffing reality. These people will be on a team together again next quarter, so keeping peer feedback developmental and away from comp is what keeps it honest.

A quick peer-feedback self-check

FAQ

Should peer feedback be anonymous?

Not by default. Anonymous-by-default breeds venting and score-settling, and in a small team it's a fiction anyway: everyone knows who was on the engagement. Signed, developmental feedback that the person sees builds trust and reads better. Safety comes from psychological safety and from keeping the feedback out of the rating (Edmondson, Administrative Science Quarterly, 1999). You don't need to hide names. Keep a separate confidential channel for genuine concerns like harassment or ethics.

Should peer feedback affect someone's rating or bonus?

Keep it developmental and out of the rating. Once it feeds comp or ranking, people sanitise or collude and the signal collapses. Feedback and appraisal are different jobs (DeNisi & Murphy, Journal of Applied Psychology, 2017). Let the leads who staffed the work own the calibrated grade, and let peers inform growth.

What questions should we actually ask peers?

Behavioural, first-person ones: what should they keep doing, what would make working with them easier, describe one moment, where did they help you or the client, one thing for next time. Avoid "rate teamwork 1–5": people rate others' abstract traits unreliably, and most of the score is the rater rather than the person (Buckingham & Goodall, "The Feedback Fallacy," Harvard Business Review, 2019).

Does peer feedback actually improve performance?

Only modestly, and only if someone acts on it. After multisource feedback, measured improvement is small across sources and near zero for self-ratings (Smither, London & Reilly, Personnel Psychology, 2005). The follow-up moves the needle: coaching and a concrete goal. The form only starts the change.

How often should we collect peer feedback?

Per engagement, while the work is fresh: a couple of short, focused requests from real collaborators, in place of one big annual blast. Captured lightly each time, it accumulates into a fair growth picture, and nobody has to reconstruct it under deadline (CIPD, 360-degree feedback).