Running a design critique people actually want to attend

I have run the Thursday critique at Brassfern for nine years. In that time it has been cancelled for deadlines, relocated to a pub, conducted entirely over video with a newborn asleep on the presenter's shoulder, and once — famously — attended by a client's CFO who thought he'd booked a different meeting and stayed for the whole hour. ("This is the only meeting here where people say what they think," he said, which we chose to receive as a compliment.)

It persists because it works, and it works because it is structured. Most design critiques fail the same two ways: the presenter gets vague validation ("nice!", which teaches nothing) or unstructured demolition ("I'd never click that," which teaches resentment). Both failure modes come from the same root — nobody agreed what the room was there to do. Our format fixes that with four mechanisms, none of which require talent, only enforcement. Here's the whole thing. Your team can run it next Thursday.

Mechanism one: the presenter frames the questions

The critique does not belong to the room; it belongs to the presenter, and they arrive owning it. Every session opens with two minutes from the person showing work: the stage of the work, the decision they're facing, and two or three framed questions they want answered. Not "thoughts?" — framed questions, like:

• "Does the pricing hierarchy read in under three seconds, or does the annual-toggle steal it?"
• "I'm choosing between these two empty-state directions — which one survives contact with a first-time user?"
• "The client asked for a dashboard; I think they've described a report. Is that in the work, or in my head?"

Framed questions do two jobs. They convert critique from open fire into aimed fire — feedback lands on the decision actually being made, so it's actionable by construction. And they convert the presenter from defendant into researcher: you're not protecting your work, you're using twelve people's eyes to answer your question. The emotional difference between those two postures is the difference between a critique people book and a critique people avoid. If you change nothing else about your reviews, change the opening. Ban "thoughts?".

Mechanism two: the silent read

After framing, the room goes quiet for three to five minutes. Everyone looks at the work — actually looks, screens or printed proofs — and writes their observations on sticky notes or in a shared doc before anyone speaks.

The silence is the whole mechanism. In a talking-first critique, the first confident voice sets the frame and the next twenty minutes orbit it; by the third comment you're not collecting reactions, you're ratifying a consensus produced by meeting dynamics. The silent read gives you the distribution before it collapses — including the quiet people, who routinely see the thing nobody else saw, because they were looking at the work while the loud people were composing.

We read the notes round-robin, one each, no rebuttals. The presenter stays silent too, which is the hardest rule in the room and the most valuable: a presenter explaining their design mid-critique is bias wearing an apron. You will have time to respond. First, collect the data.

Mechanism three: kind, specific, necessary

All spoken feedback passes a three-word filter before it's allowed into the room. It must be kind (addressed to the work, never the worker; critique the screen on the wall, not the designer at the table). It must be specific ("the contrast on the tertiary button fails at this size" is feedback; "it feels busy" is a weather report — useful only if you then say which clouds). And it must be necessary — in service of the presenter's framed questions, not a tour of how you would have done it.

"Necessary" is the one people struggle with, because it forbids the most psychologically satisfying genre of feedback: demonstrating taste. Every design team has someone whose comments reliably translate to I would have done it differently and I'd like that on record. At Brassfern, that sentence gets caught by the filter with a question from the facilitator — "which of the framed questions does that answer?" — delivered warmly and without exception. Warm, because tone is load-bearing: if the filter feels like a punishment, the room clams up. It's a good-natured redirect, and after three Thursdays, people start filtering themselves in the seconds before speaking, which is when you know the culture has landed.

If your team also wrestles with receiving feedback as a client-facing skill, the counterpart discipline is in how to give design feedback that speeds the work up — written for the other side of the table, and the two rituals reinforce each other.

Mechanism four: seniority speaks last, and we keep a worst-advice log

Two final pieces of plumbing. First, the speaking order runs most-junior to most-senior, with the facilitator and any directors at the end of the line. Senior voices anchor rooms even when they're wrong — especially when they're wrong, because confident wrongness from a principal is gravity. Juniors-first means the intern's genuine first impression — the closest thing to a user's genuine first impression in the building — gets recorded before my opinion can edit it. Some of the best catches in our critique's history have come from people in their first month, precisely because they hadn't learned what the work was "supposed" to be.

Second, the worst-advice log: a running doc where the presenter notes the feedback they chose to ignore and why, reviewed at the end of each project. This sounds petty. It is the opposite. Every critique generates advice that doesn't get taken — necessarily, because advice conflicts — and in most studios that advice just evaporates, so nobody ever learns whose instincts age well. The log closes the loop. It has revealed, among other things, that one of our directors' colour instincts improve every project they touch, that my own typography feedback is right embarrassingly often and my navigation feedback is right embarrassingly rarely, and that the single most ignored-and-vindicated category of advice is "the client will think this says X." Data about your own room is the only feedback training that transfers.

The logistics that make it stick

None of the mechanisms survive bad logistics, so: sixty minutes, weekly, same time, two presenters maximum — depth beats coverage, and a critique that triple-books is a status meeting with better visuals. Work can be any fidelity, but never polished-fake: we explicitly welcome ugly wireframes and forbid presenting work as more finished than it is, because the stage of the work sets the altitude of the feedback and a pixel-perfect mock of an unresolved flow collects type opinions when it needs flow opinions. Attendance is wide: engineers, writers, producers, strategists. The weekly demo is the whole team's window into a project; the critique is the design team's, and the cross-functional mix is what keeps the feedback honest — engineers catch the state nobody designed, writers catch the words pretending to be UI.

And the facilitator rotates. Critique is a skill with a feedback loop of its own; you only learn to run a room by running one. If you're building the wider culture around rituals like this, our studio page covers how the week fits together, and the way we align stakeholders without design by committee is the same philosophy pointed outward at clients: strong structure is what lets honesty stay kind.

Key takeaways

• A critique without framing defaults to vague validation or demolition. The presenter opens with the work's stage and two or three framed questions — "thoughts?" is banned.
• Run a three-to-five-minute silent read before anyone speaks; collect the distribution of reactions before meeting dynamics collapse it.
• Filter all feedback through kind, specific, necessary — and let "necessary" kill taste-demonstration, warmly and without exception.
• Speak junior to senior. First impressions from the newest person are the closest available proxy for a user's.
• Keep a worst-advice log and review it per project; it's the only mechanism that teaches your room whose instincts to weight.
• Protect the logistics: weekly hour, two presenters, honest fidelity, wide attendance, rotating facilitation.

Frequently asked questions

What if the presenter is presenting to stakeholders, not the design team? This format is for the internal room. Client reviews adapt the same bones — framed questions, seniority-last, specific-kind-necessary — but they add translation work and trade the silent read for pre-circulated context. Mixing the rooms is how you get designers performing confidence instead of testing uncertainty.

How do you handle someone who keeps violating the filter? Redirect in the moment, privately after the third time. Almost every persistent offender is optimising for status and will switch targets once taste-demonstration reliably fails to earn any. The rare remainder is a conversation about fit, not facilitation.

Does this work remotely? Yes — arguably better. Shared docs make the silent read effortless and simultaneous, round-robin is trivially enforceable, and recordings give the presenter the notes verbatim. What remote kills is the hallway debrief, so we keep ten minutes of loose time at the end for it.

Doesn't round-robin get slow with a big room? Cap the room at ten to twelve. Above that, split into two parallel critiques and swap the notes; quality of attention collapses faster than the seating plan suggests.

How long until a team feels the difference? Three sessions. The first is awkward (the silence feels enormous), the second is competent, the third is the one where someone says "I actually look forward to this" and means it. The mechanism is less a technique than a promise kept weekly.