At Thoughtworks, "cultivation"—helping one another grow—is central to how teams work. Collaborative software development creates constant opportunities to exchange views on how people think, solve problems and interact. Teams where colleagues give candid, timely and instructive feedback tend to be the ones where people grow the most, yet this kind of peer-to-peer exchange remains uncommon across the industry. What follows is what the authors have learned about it from long experience.
Core principles
Four ideas underpin the practice of sharing feedback.
All feedback is positive
Patrick Kua helped one of us see through the myth that feedback comes in positive and negative flavours. His claim was simple: "All feedback is positive." He listed only two goals for sharing it.
- Strengthen confidence: Reinforce what someone does well so they know which behaviours to repeat; a.k.a. praise.
- Improve effectiveness: Identify what the other person could do differently to perform better at their job or engage better with others; a.k.a. criticism.
Both goals are positive. If you are communicating to achieve some other goal, you may not be sharing feedback at all. Treating all feedback this way shifts your perspective and makes it easier to accept.
Radical candor
Kim Scott's Radical Candor framework summarises the approach well: it happens when we care personally about coworkers and challenge them directly about their work and interactions. Scott also describes the anti-patterns to avoid.
- Obnoxious aggression: challenging directly without showing personal care.
- Ruinous empathy: caring without challenging directly.
- Manipulative insincerity: neither caring nor challenging directly.
Feedback is a peer activity
Teams include people at varying levels of seniority and positional authority, but when sharing feedback, we treat each other as peers. Titles and grades can get in the way, yet in principle anyone can strengthen a colleague's confidence or improve their effectiveness. This places a heavy onus on senior colleagues to be open to feedback and to seek it out.
The most important words a leader can say are, "I screwed that up".
-- Dave Cooper
Retrospectives and after-action reviews reinforce this dynamic by fostering reflection, psychological safety and collective improvement.
Feedback as telemetry
Teams are social systems, and like software systems they need telemetry to know how they are behaving. Feedback is one such signal. A service without logs and metrics can appear healthy until it fails; a team without feedback can appear harmonious until a deadline, conflict or reorganisation exposes problems everyone had noticed but nobody had named. Continuous, peer-to-peer feedback serves as that telemetry signal for teams and individuals.
Keeping feedback in the flow of work
The early-and-often loop applies to peer-to-peer feedback as much as it does to releases. Knowing sooner that something could be better, or that something is already working well, lets us adjust or repeat the behaviour. Reflective feedback that looks back over a stretch of time is instructive, but feedback delivered in the flow of work is the most useful: the closer it lands to the interaction or the piece of work, the easier it is to understand and learn from.
Agile teams already generate plenty of these moments. Pairing is not only a way to solve problems together but also an opportunity to exchange feedback — and while teams increasingly pair with AI agents, two humans pairing for a few hours still learn and grow in ways AI does not replace. Beyond pairing, a shared design document, a joint presentation or a workshop run with others all serve as occasions for feedback.
Anchor feedback in observation
The central discipline in giving feedback is separating what you observed from the story you told yourself about it. Take this example:
When you presented today, I noticed you skipped several slides. Skipping slides makes your audience feel that you're either delivering a canned talk or are underprepared. Next time, if you have slides that aren't relevant to the audience, how about hiding them before you begin presenting?
It contains a situation (today's presentation), a behaviour (skipped slides), an impact (how the audience perceives it) and an optional recommendation (hide the slides beforehand). Leading with situation and behaviour keeps you on objective ground; the impact is your perception, and that perception is the value you offer. Situation-behaviour-impact, or SBI, is a popular format — but seeking context first avoids some of the tension a feedback conversation can accumulate.
Ask before you judge
No one screws up on purpose, so there is always context behind an action. Rather than launching into the SBI, open with a question: "Hey, when you were presenting today, I noticed that you skipped a number of your slides. Why was that?" Perhaps a late start or unplanned interruptions forced the presenter to cut content. Feedback is still possible, but now it is about elegant ways to skip slides rather than better preparation.
Seeking context also creates safety: you position yourself as curious rather than judgmental. A CSBI — context-situation-behaviour-impact — pattern builds that curiosity in before any judgment.
Get consent first
Feedback lands best when the other person is open to it, so don't ambush colleagues. "Is now a good time to share some feedback with you?" suffices for a short conversation. For something weightier, state purpose and duration: "I have an observation about how the workshop went. It may be useful to talk about it for 20 minutes. Is now okay, or should we find another time today?" If they aren't ready, agree on another time. Persistent reluctance to hear feedback may point to other issues in the working relationship — address those first.
Go private, go synchronous
Feedback aimed at improving effectiveness is usually safer in private. Public criticism adds an audience, forcing the receiver to manage both the content and their status in the room; psychological safety takes an immediate hit.
Real time — in person or on a video call — is preferable, because text is a poor default for feedback that carries tension: the written word doesn't promote conversation and lacks the pauses, body language and facial expressions that defuse tension. If both parties agree, document the feedback afterwards, but a private, synchronous conversation remains the sensible default.
Praise is the exception. People don't mind recognition in public — at Thoughtworks, the internal social network where colleagues share kudos is a popular way to recognise value-aligned behaviour, and such peer recognition can be both public and asynchronous.
Invite the solution
When sharing criticism, prescribing a fix comes naturally. Prescriptions can work when you're helping someone align to team norms, but in most cases involving the receiver produces a better way forward. Instead of telling the presenter what to do, ask: "How can you prepare your presentation in a way that allows you to be flexible to the situation you find yourself in?" Open-ended questions help colleagues find solutions that work for them — and since they, not you, must take the action, they are likelier to follow through. Together you may land on something better than your original idea.
Scale the feedback to the impact
Sharing feedback early, often and close to the work lets you match it to the level of impact, which can escalate across successive conversations.
- Content. The first time you notice a behaviour, address the specific incident — for example, a commit message that isn't clear or descriptive. Show the colleague how to write an effective one. Content-level feedback is the easiest to give and the one most often ignored; sharing it in the flow of work, or soon after, prevents repeated infractions from growing into greater impact.
- Pattern. If the same issue recurs, refer back to the earlier conversation before naming the pattern. "Hey Harish. If you remember, we discussed your commit messages last week, and I'm now seeing a pattern in your commits. I noticed that your last commit message was still quite high-level, and I had to open the diff again to understand what changed. As I explained, on our team we frame commits around the outcome or the reason for the change. In your latest commit, you should have written that you're rejecting expired tokens before creating a new session…" Referencing the prior discussion and establishing the pattern raises the stakes and makes the feedback harder to ignore.
- Relationship. If content feedback and a named pattern still bring no change, the stakes escalate further: repeated infractions fray working relationships, and the conversation can no longer be about the content or the pattern but about how you think of one another. Facing a colleague who has ignored two rounds of feedback, it might sound like this: you raised that messages such as "update code" did not explain intent or outcome; you then flagged the recurring pattern and repeated that commit messages should give the team enough context to understand a change without opening every diff; since then you have seen "fix tests", "address review comments" and "update validation". You note that it worries you the feedback isn't being acted on and that you don't share the same standard for communicating work, and that you now scrutinise the changes and rely less on the commit history, which makes collaboration more effortful — closing with "How can I help you follow the team standard for commit messages?"
If escalating to the relationship level doesn't work, someone else — a manager or HR — may need to intervene. That is where peer-to-peer feedback reaches its limits, and where establishing team norms may require positional authority.
Praise fits the same models
It is easy to imagine feedback as criticism only, but sharing feedback solely to improve effectiveness can hurt peer-to-peer relationships and dampen morale. Colleagues are often doing many things right with little acknowledgement; without encouragement they may change those behaviours or stop demonstrating them — neither a good outcome for the team.
CSBI and the content-pattern-relationship hierarchy apply as much to praise as to criticism. John Reid-Dodick, chief people and leadership officer at Thoughtworks, speaks of "catching someone doing something right": staying alert to what teammates do well and praising those behaviours matters as much as sharing criticism.
Your feedback is not universal truth
Context matters, and sometimes feedback reveals more about the giver than the receiver. One of us, Sumeet, often advises colleagues not to delegate communication to AI. The advice is often sound, but no one is obliged to follow it, since there are as yet no team or company norms around using AI to communicate — the feedback reveals his communication preferences more than an established standard. He holds strong views but doesn't expect everyone to agree, and keeps sharing them with anyone who cares to listen.
Unless your feedback concerns an established principle or standard, treat it as a gift. Gifts are sometimes useful and sometimes not; leave your colleagues to decide how useful yours is.
Receiving feedback well is harder than giving it. Praise is easy to accept; criticism rarely is. But if we recognise that all feedback serves one of two ends—strengthening confidence or improving effectiveness—criticism becomes easier to hear. Feedback is telemetry for the team's social system, and losing access to it is worse than hearing something uncomfortable.
When you see yourself doing something badly, and nobody's bothering to tell you anymore, that's a bad place to be. You may not want to hear it, but your critics are often the ones telling you they still love you, care about you, and want to make you better.
-- Randy Pausch
The "assume positive intent" (API) principle is the right mindset here. Feedback that fails to land with us usually isn't malicious; it may be incomplete, clumsily phrased, or shaped by a different experience of the same event.
Habits that make feedback land
There are only two responses to feedback — "Thank you," and "Thank you, please tell me more"
-- John Reid-Dodick
Start with thanks
Offering feedback demonstrates care. The giver may be wrong about what we need, but they care about us getting better or staying awesome. A thank-you signals openness and keeps colleagues sharing feedback in the future. The same applies to praise: some of us deflect it reflexively. Sumeet grew up in a tough-love culture and would answer praise with "Oh, it's nothing, really." With conscious practice, he replaced the deflection with a simple, equanimous "thank you."
Clarify without defending
"Thank you, please tell me more" is powerful. Many people are cagey about giving feedback and clumsy at articulating it, so questions help both sides. The challenge is the fine line between clarification and defence:
- "Can you give me an example?" is clarification.
- "But I had to do it because nobody else would" is a defence.
Defence is a natural first reaction to feedback we don't understand or find unpleasant, but suspending judgment and reflecting serves us better. Useful clarifying questions include:
- When did it happen?
- What did you observe?
- What impact did you see?
- What did you expect instead?
This is one of the biggest benefits of exchanging feedback in a real-time conversation: there is room for such questions.
Read written feedback with a smile
Written feedback can feel unpleasant at first blush. Reading it while imagining the giver speaking it with a warm tone and a smile helps a lot. Recently, Anuja has pasted written feedback into Google Docs and had Gemini read it back in a friendly tone. Text carries no body language or expression, so the harshness we experience often lies in our interpretation rather than the giver's intent.
It's fine to sleep on it
Sometimes, despite our best efforts, we can't reconcile feedback with our view. That's OK. It's also OK to tell the giver we don't understand the feedback but will think about it: thank them, reflect for a while, then decide whether there is anything to act on. If giver and receiver both treat feedback as a gift, these reflective pauses are comfortable for both.
Close the loop
Imagine someone using a gift you gave them and telling you how useful it was. The feedback gift works the same way: acting on feedback and closing the loop with your colleague demonstrates that their input mattered, and encourages them to share more. Closing the loop matters even when nothing is actionable:
Thank you for the feedback. After reflecting on it, I still don't see a way to act on it, but I'd love for us to keep this conversation going.
Don't wait—ask
Sharing criticism is hard anywhere; it is easier when someone explicitly asks. Asking for feedback therefore gets us better at what we do, even from colleagues who aren't skilled at giving it. Be specific about why you're asking—about a shared project or a particular skill:
We worked together on the release and the trunk migration. I'd love your feedback on how I collaborated with you and how I can make my changes easier to review and integrate.
The clearer the question, the more useful the answer.
Antipatterns to avoid
Feedback tied to performance reviews
Many teams collect peer-to-peer feedback during performance appraisals. In our experience it serves neither personal growth nor the review. Reviews track performance against goals agreed with a manager; peer feedback that doesn't address those goals can confound the process. Because reviews feed into salary increases and promotions, honest feedback gets suppressed—nobody wants to hurt a colleague's prospects. Written formats, already a difficult channel, carry even higher stakes here, and looking back over six months or a year makes specific examples hard to recall, so platitudes and generalisations dominate.
Decouple the two. The appraisal should focus only on performance against agreed goals. The reviewer shares reflective feedback, but none of it should surprise the reviewee—all of it should have arrived in the flow of work over the preceding months. The most effective review processes have managers holding regular feedback discussions with their direct reports, culminating in the appraisal.
The sandwich method
Sandwich every bit of criticism between two layers of praise.
-- Mary Kay Ash
The sandwich dates to the 1980s, and it doesn't work. The praise feels insincere, the criticism feels diluted, and the receiver is left confused about the point. Instead, headline both: tell the receiver you have praise and criticism to offer. Our ex-colleague Pat Kua would spend part of a conversation strengthening confidence before moving to improving effectiveness; both landed, and the receiver left with clarity rather than confusion. Catching people doing something right is worthwhile—but don't let it dilute the intent to improve effectiveness. Separate praise and criticism; the message lands better.
Orchestrated feedback events
Teams seeking a healthy feedback culture sometimes stage events that do more harm than good.
Public 360 feedback
In a public 360, people praise and criticise each other in an open meeting. Most will not be honest with criticism in public: we are social creatures and don't want to damage anyone's status. Sharing criticism publicly makes psychological safety the casualty, and there is little upside. Team dynamics benefit from open conversation—that is where agile retrospectives come in—and retrospectives are not a forum for peer-to-peer feedback.
Speedback
Speedback applies speed-dating to feedback: teammates pair with a random colleague every few minutes and exchange feedback. It rarely helps. Feedback benefits from shared work context and powerful questions; paired with someone they don't share context with, people may have nothing to say. The two-to-three-minute constraint creates pressure, and people skirt around conversations with shallow observations. A deep, 30-minute conversation is often far more valuable than such contrived events. Feedback doesn't need special events; embed it in the natural flow of work—every collaboration is a chance to seek and offer it, and so is a one-on-one.
Feedback farming
Asking for feedback works when the other person has enough shared experience to observe something. Feedback farming is soliciting comments from anyone who might provide a favourable signal and treating the volume of responses as evidence of performance. It is common where peer-to-peer feedback is conflated with performance reviews: the goal shifts from learning to accumulating appraisal currency. The results are strange situations—colleagues receiving formal requests from people they haven't collaborated with enough, leaving them to respond with platitudes or ignore the request. Both are bad options.
None of these antipatterns is malicious or ill-intentioned. They are misdirected efforts in search of a shortcut.
Why the shortcuts fail
The performance-review pile-up, the sandwich, the public 360, speedback, farming feedback for appraisal currency—every one of these antipatterns tries to compress something that cannot be compressed. Each is an attempt to manufacture trust in a single event, while genuine trust accumulates in ordinary, everyday moments. A feedback culture cannot be installed in an afternoon any more than trust can be.
So the single takeaway is to stop looking for the shortcut. Don't wait for the retro, the review cycle, or the offsite. This week, each time you collaborate with a colleague, look for ways to strengthen their confidence and improve their effectiveness, and ask them for feedback as well. Say thanks when you receive it, then repeat the exercise the next week, and the next. Done often enough within the ordinary flow of the work, this may leave you with no need for the frameworks, the events, or the workarounds—you will have contributed to your team's feedback culture.
Acknowledgments
Ifeoma Onwughalu, Jade Daubney-Bennett, Bruna Pons, Inger Dickson, Karolina Najdek, John Reid-Dodick, Yewande Ige, and Gayathri Moghe reviewed drafts of this article. Martin Fowler provided editorial oversight, and the Thoughtworks cultureShapers group strengthened the authors' confidence with their reactions to the initial drafts.
Grammarly was used to proofread and copy-edit the essay, and Claude helped verify that all feedback received on the article had been addressed.
Disclaimer: The statements and opinions in this article are those of the author(s) and do not reflect the positions of Thoughtworks.
24 September 2026: published



