How to Give Feedback 💛
How to empower your teammates and build trust by giving feedback the right way.
Feedback is one of the most used words in business, and also one of the most literal.
If you look it up in a dictionary, feedback occurs when the outputs of a system are routed back as inputs. So the system literally feeds back into itself.
The simplest systems to include mechanical feedback were invented in Egypt more than 2000 years ago, and were basically ancestors of modern float valves. The first adoption of the word feedback, instead, is attributed to Nobel laureate Karl Ferdinand Braun, who used it in some early 1900s papers around electronic circuits.
Feedback today can refer either to systems or people. They both receive feedback to assess their performance and make it better over time.
As managers, it’s on us to create this loop for our reports. It’s our responsibility to clearly articulate what is expected of them, what is working well, and what they can do differently instead.
Giving feedback, though, it’s hard.
Negative feedback is especially tough to deliver, for two reasons. As humans:
We want to be liked — we want other people to like us, and we instinctively believe that negative feedback makes us liked less.
We don’t want to hurt feelings — we think people may react negatively to criticism, and we don’t want them to be sad, or angry.
Both these concerns can be countered by giving feedback the right way. You can effectively create a culture and processes where 1) people like you more because you give them feedback, and 2) people are eager to receive it because it supercharges their growth.
This article is meant to be a comprehensive guide about how to give feedback in a way that enables these qualities. It covers:
🔢 The feedback quadrant — core principles about good feedback.
✍️ How to formulate feedback — how to be clear, but kind.
🔍 What you should give feedback on — the relationship between positive and negative feedback.
🗓️ When to give feedback — processes and tactics to make sure you have a constant exchange.
As always, it draws from my own experience, from plenty of conversations with the best managers I know, and from real-world case studies and research.
Let’s dive in 👇
🔢 The Feedback Quadrant
Kim Scott has seen it all.
She led teams at Google, coached Twitter and Dropbox CEOs, managed a pediatric clinic in Kosovo, and started a diamond-cutting factory in Moscow.
She spent a lifetime managing the most diverse people and leading them to success. Later, she wrote about her learnings in one of the world’s most popular books on management and feedback: Radical Candor.
Kim breaks feedback into four quadrants. On the horizontal axis you have unclear to clear feedback, and on the vertical you have negative to positive feedback.
Whether positive or negative, the best feedback is always clear and specific. On the contrary, it sets up our reports for failure.
Softening negative feedback is human — we want to avoid adding even more pressure on our teammates. Such cruel empathy, though, doesn’t give them the tools to do things differently. It dampens their growth.
Unclear positive feedback is equally ineffective. Simply saying “you are doing a great job” doesn’t cut it, because it doesn’t bring any specific learning. In the worst case, it feels artificial and a way to counterbalance the bad one.
So how can you give feedback that is clear and kind, even when it might not be nice?
✍️ How to formulate feedback
At its core, feedback is an actionable observation about a recipient’s behaviour. We give feedback to encourage one of two things:
🟡 Do something differently — propose to change this behavior into something more effective with respect of some goal.
🟢 Continue doing what you are doing — signal that the behavior is working well and that they should continue with it.
To be effective and formulate feedback correctly, I use a simple mental model that includes four items:
Request / Question
Let’s see them all 👇
This is the who, when and where.
Is this something that happened on a particular occasion? Is this a recurring behaviour? Do you notice this happening when some condition is met?
The what. Your observation of someone’s behaviour. Focus on facts, leaving aside any kind of judgement or assumptions.
Consequences of such behaviour, explained in an objective way.
It is best if you are able to tie this to some business impact. Example:
When you create a pull request with thousands of lines, the reviewer may need to spend hours on it, so it will likely pass a lot of time before you get a proper review. That may slow down the release. If you are able to split the work into several, smaller PRs, instead, it is more likely you will get a speedy review and you won’t get bottlenecked.
If you are unsure about the impact, you can tell them your honest feelings about it.
Framing the impact around your feelings is powerful for two reasons: 1) feelings are always objective, and 2) they are your personal reaction, so the other person doesn’t feel directly attacked.
Two examples of the same feedback, with different delivery:
In the latest meeting you diminished me by failing to address <our important achievement>
When in the latest meeting you didn’t address <our important achievement>, I felt diminished because we all worked so hard on that.
While they basically say the same thing, the first example is more likely to create a debate, because you are labelling the other person's behavior and she might not agree with your interpretation.
The second example, instead, only addresses your reaction, without an explicit judgement. This is effective because it relies on the assumption that the other person cares about you, and will want to sort this out. You can easily follow up on this with “Was that your intention?”, to set an even more collaborative ground.
Question or Request
Feedback should always end with some action item. This can be either a request, or a question.
The difference between the two is similar to the one between mentorship and coaching.
Requests are for direct advice. When there is little doubt about the topic or the way forward, formulating a simple request is a clear way to wrap up the conversation.
Sometimes, though, we are not sure why some behavior happens or what is the best way to address it. In this case, we can be curious instead of critical, and ask questions to learn more.
Questions help us both to come to an actual request later, and to facilitate some introspection by the other person, who may come to a solution by herself.
🔍 What you should give feedback on
For those who receive it, feedback should act as a way to assess their overall behavior. This means identifying things that they get right, things that they don’t, and adjust accordingly.
This balance isn’t always met — many managers I know give an outsized amount of negative feedback with respect to the positive one.
There is a tendency to believe feedback should be mostly corrective. This is false and causes two obvious problems:
Self-evaluation — if most feedback people receive is criticism, they will think their performance is bad.
Bias against feedback — people will associate feedback with negative emotions, will start avoiding it, and handicap their own growth.
Last week I sat down with Matt Van Itallie, CEO at Sema, who taught how to give feedback at McKinsey and Harvard. He is clear about this balance:
The ideal praise-to-criticism ratio has been studied extensively. Research indicates that in the highest-performing teams this ratio is >5:1, that is, people receive more than five positive comments for every negative one.
If this seems outlandish to you, think about the people on your team. How much do they really do wrong, in percentage of their total actions? Chances are even your average performers do 80% of things right. You should acknowledge those things as well.
Matt also shared the 📑 template he uses at his company in one-on-ones, and in special meetings he holds monthly to give and receive feedback from everyone on the team.
He identifies three areas:
⭐ Values — feedback around company principles.
💬 Communication — feedback around how people interact with each other.
🔨 Work — feedback around specific pieces of work.
These are in descending order of importance, and ascending order of how frequently such kind of feedback comes up.
For each area, you have two kinds of items:
WWs (Working Well) — Reinforcing feedback about what the person is doing well. About 80% of items.
DDs (Do Differently) — Corrective feedback about what the person can do differently. About 20% of items.
🗓️ When to give feedback
There are three situations in which most feedback is generally shared:
In my experience, these are in descending order of effectiveness. Let’s see them all 👇
The best time to give feedback is on the spot.
The closer you are to the action that generates the feedback, the more context both you and the receiver retain, and the more clarity and alignment you are able to create.
If you don’t already, try to give at least one specific piece of feedback every day. Developing such habit is incredibly powerful:
It makes you more attentive to other people’s behaviour.
It develops your own introspection, by making you reflect on your own judgement.
It creates a virtuous cycle where you create stronger relationships with your teammates and accelerate their growth.
Brendan, founder and product manager, shared his take on the community 👇
Give feedback frequently (i.e. multiple times a day — it should be a non-event), that is mostly positive (~70% is ideal) and actionable (specific and objective).
Formula is something like “Can I give you some feedback?” (And respect the answer), then “when you <objective and specific action>, <objective impact of that action>
An example might be “when you move a small bit of complex code into a method with a clear name, like you did here, it’s easy to understand what the code is doing and why.”
The second best time to give feedback is during one-on-ones.
These are crucial to having a safe, recurring space for honest conversations with your reports. I have written more about them in a previous article:
One-on-ones are also effective for you as a manager to receive feedback. If this doesn’t come easy for reports (spoiler: it usually doesn’t), you may solicit it by asking direct questions:
What can I start doing?
What should I stop doing?
What should I keep doing?
Reviews are of course a canonical place where to share feedback, but my take is that these should mostly ratify what has been already discussed privately.
If your report is surprised by what they read in the review, or feel they are discovering that for the first time, then it means you did something wrong.
If you share frequent feedback on the spot and in one-on-ones, then you can use reviews to draw the big picture based on such past feedback.
📑 Ideal Praise-to-Criticism Ratio — Harvard research about how to balance criticism with positive feedback, backed by data about top performing teams.
📑 Feedback Equation — Lara Hogan shares the equation she uses to give feedback: observation + impact + request.
📑 My Management Lessons — Kim Scott shares lessons around management and feedback she learned over her extraordinary career.
📝 The State of Frontend 2022
The talented guys at TSH are building a comprehensive survey around the State of Frontend, interviewing thousands of developers. I have worked with them on some of the questions and will later write my own take on the results.
If you have 5 minutes to spare, it would mean a lot if you filled out the survey! 🙏
Hey, what I often miss from the guidance on how to give feedback is also on "how to receive feedback". I feel like we can be trained and coach others on how to give feedback but if there is lack of understanding on the other end as to why feedback is beneficial for them, but also guidance on how to react or not to react is super important, such as: that you do not need to react to it right away, that feeling upset or angry is ok, taking time to process is ok and also asking for follow up.
Hey, Luca! Great article, i just have one questions. You say that the feelings are always objective. What do you mean by it? As i see it they may be biased or affected by multiple things. Thanks in advance.