How to iterate your way to great feedback: an engineer’s guide to performance reviews

Disclaimer: This is not a commentary on the value of performance reviews[0] but assumes you’re working somewhere where you simply have to write them. Instead, it’s a working method that can help you finish your reviews without getting stuck, by iterating and distilling feedback down to what really matters. Note that this may be most helpful for writing downward reviews (for people you support on your team) and most useful if you have several of them to write.

For context, I first became a people manager back in 2011, and since then have written hundreds of reviews (self, upward, peer, downward) including almost a hundred in downward alone. I think few people find writing reviews fun, but early in my career I remember struggling to find the right words and spending hours per review in drafting and editing.

Illustrative example of the joy of writing reviews, as posted by a friend on Facebook

Over time, I developed a process for outlining, structuring, and prosifying performance reviews that’s since helped me cut through the pain and write focused reviews much more quickly. As I walk through the steps below, I’ll use a generic review outline I made up as an illustrative frame.

Step 1: Catalog your work

To start, collect all the information you need and put it in one place. This can increase your efficiency and create a space to quickly jump through different reviews or content and iterate quickly. This information might come in the form of guides or instructions from your company’s people team, but can also be project plans, design docs, key tasks, or even large PRs that were particularly important from the review period.

Some things I do:

  • Write a basic checklist, grouped by review types and names
  • Create a folder or space for all of your review drafts
  • Have a copy of your company’s review prompts in a doc you can easily reference — you should refer back to this frequently and use it as a guide and clear rubric

What my desktop roughly looks like for my Q3 2017 review cycle.

Step 2: Sketch out a basic outline

Using your company’s prompts, start a draft review doc and include sections you’re trying to complete. Jot down initial thoughts as you sketch it out, stick to bullets or phrases, and don’t worry too much about how “good” the feedback is yet.

This is a purely illustrative review outline I made up, here if you want to see it.

Step 3: Read through any supporting feedback or additional data sources

Collect all of the information you want to incorporate into your review, and read through it very carefully. As you read, toss notes and bullet points into your review outline. Don’t self-censor, but do focus on things that have clear, explicit supporting examples or come up consistently across sources. When you come across a great quote from someone, pull that out into your outline and remember who wrote it — I believe great reviews represent more voices than just yours as the manager, and it’s important to reflect that authentically in the end result.

As an aside, writing feedback is something I don’t think people get recognized for enough. When reviews are complete, I often go back and personally thank people who took the time to write great feedback for their peers.

Repeat this step a few times until you think you’ve properly distilled down what you’ve read.

Step 4: Edit and iterate on your outline

Now that you have your outline, sit with it for awhile and edit it while asking yourself different questions. Here are some that have worked well for me:

  • Did you end up pulling out a few areas that fit naturally into a broader theme? What is that theme?
  • Are some of the areas ones you’ve talked about before? If so, how has the feedback changed since prior reviews?
  • Do the quotes you pulled out represent this feedback well or give a concrete example to explain it? If not, do you have a concrete example you’ve seen that can help supplement the point?
  • Is this feedback important enough to highlight as part of the review? Was it a one-time error or unusual mistake? If it’s a strength, is it one that the person demonstrates unique ability in or has shown improvement in since the last cycle?
  • How well do the themes resonate with what you’ve seen of their work? If you are seeing things in people’s feedback that surprise you, think about why that is and take another pass through the information you have to get more context and better understand it.
  • Do you have any concrete ideas for ways to address the feedback that you can capture and share? How can you help this person grow and be even better in the next cycle?

Typically, I try to zero in on 3–5 major themes or key messages per review prompt. These should feature concrete examples, resonate well with you and the feedback you’ve gathered from others, and contain some specific ideas to address the feedback going forward. Without those, you run the risk of using generic phrases that might not mean much of anything. There are also a few frameworks[1] out there that can help you deliver more effective feedback like this.

Step 5: Prosify

Now take some time to structure your thoughts in writing. I often leave certain sections in outline form (e.g. bulleted lists for the key 3–5 themes) as they can be easier to process, but use prose to capture how themes relate to each other and how best to communicate them overall. Don’t focus on flowing dialogue that turns each review into a short novel — think about the person or people who will read it, and what messages you want them to take away from it, as clearly as possible. If you soften the language too much, bury it in needless filler content, or omit concrete examples or details, you run the risk of your key messages simply being lost and ignored.

Step 6: Proofread and profit

Finally, take a quick final read-through to make sure the feedback resonates with you. Hit submit and you’re done!

Meta-Step: Mix it up

As one final efficiency tip, try mixing it up or pipelining your reviews. I often jump between people and finish outlines for everyone before writing any prose, as I find that last step the hardest to iterate on. Step-wise, my process looks something like this:

Step 1: Catalog your work
For each teammate, in any order:
* Step 2: Sketch out a basic outline
* Step 3: Read through any supporting feedback or additional data sources
For each teammate, in any order:
* Step 4: Edit and iterate on your outline
[Take a break and scan through your outlines until you feel good about them.]
For each teammate, in any order:
* Step 5: Prosify
* Step 6: Proofread and profit

I hope this guide helps you increase your efficiency as you write reviews, but sadly I’ve yet to find a cure for procrastinating on them in the first place. I.e. Instead of grilling sausages over a firepit, I decided to write this post.

Question for the reader: I’d love to hear what you think of these tips! What works (or doesn’t work) well for you?

[0] If you are curious about performance reviews in general, here’s an interesting article for that:
[1] Here’s one example we use at Stripe:
[also] thank you to
Gloria Lin Michael Schade Jorge Ortiz and Michelle Bu for helping me with this :)