Setting the Performance Bar: What Nobody Tells You About Missed Expectations

Fifteen years of teaching taught me that most missed expectations are a failure to set them. What my classroom experience taught me about defining and holding a performance bar as an engineering manager.

TL;DR

• If someone misses a bar I never described, that is a management failure (mine!), not a performance failure. Run the retro on yourself first.
• Separate the instructions from the task. People should be evaluated on the work, not on their ability to decode what was wanted.
• Describing “good” is not enough. Show it. Give exemplars. A weak version next to a strong version teaches more than a rubric does.
• Name it immediately. Letting it slide (deadlines!) or delaying feedback lets the low bar become the standard.
• Fix the system, not just the instance. Set people up to succeed rather than catching them after they fail.
• Build the standards with the team, not at them. Shared agreement on what “good” looks like is what makes a bar hold, not my oversight.
• Standards and goals are different things. Skipping tests is a performance issue. A model that misses its number is not.
• Empathy is not a reason to avoid the hard conversation. It is how you deliver straight talk as someone who wants them to succeed.

Teacher to EM

I spent 15+ years as a teacher & university professor. I have roughly 10,000 hours of teaching experience. I learned a lot about how to manage performance from this, trying new things or tweaking my approach with every class and every term. I think a lot of this is relevant to managing performance as an EM.

If the Instructions Weren’t Clear, You Are the Problem

I once proctored a final English test at a high school in Osaka. I don’t remember details but it was some kind of listening test. My duty was to take attendance, hand out test papers and then start a recording for an initial listening test. This was a timed test and I was no permitted to stop or restart the recording.

After starting the recording I stood to the side of the room to observe. I immediately noticed an issue. The instructions for what to do for this listening test were not written on the test sheet - they were delivered orally … in English on the recording. What I noticed was that one girl had misunderstood the instructions and was on the wrong track.

Afterwords I talked to her teacher (Dan) and told him what happened. I thought it was unfair and suggested perhaps he could take that misunderstanding into consideration when grading her test. He said he didn’t think that was necessary. “It’s ALL part of the test” was his opinion, including instructions.

I disagreed… but it wasn’t my place to say or do more so I left it at that. But the experience has stuck with me.

I had learned (during teacher training) that separating instructions from the actual assessment is a fundamental part of the test pedagogy. If it were me I would have realized that I was responsible for not delivering clear instructions and … I don’t know what. Probably have some human understanding and take that misunderstanding into consiration when grading her. I would definitely take that as a learning experience and change the approach next time (i.e. provide written instructions, probably written in Japanese, rather than English.) Holding a quality standard doesn’t mean you have to be a hardass.

One reason this incident hit home is that I, myself, have a form of aphasia that causes delayed processing. I’ve been in this exact situation a few times where oral instructions are given that I didn’t completely get.

A Bar I Never Actually Set

I’ve also been on the other side as a teacher more than once. I have been the teacher who gave an assignment that was not completed well and, on reflection, I realized I could have been more clear and proactive about clarifying expectations and deliverables. In one case, I was teaching a 1st year (university) class. The assignment was that all students had to give a short presentation (in pairs) about one topic from the syllabus. The first pair got up and … did a truly terrible job. But the thing is they had obviously done research and prepared materials and had collaborated well. And they were actually enjoying themselves. (Gregarious, natural performers.) But the thing was, there was not there … there. It was light on substance. More like just a list of facts and descriptions of photos they prepared. It most definitely didn’t meet the bar. While they were talking I was doing a retro in my head trying to determine how this had come to pass. I decided by the time they were done talking that I bore some responsibility. I could have given examples of good vs bad, the type of insights I was looking for and what “good” looks like.

After they finished I thanked them and told everyone to give me ~5 minutes to write my reflection. That gave me some breathing room to think. I came to a decision. I think it is always better to address issues as soon as possible and be as straight as possible. But on the other than doing so would mean calling them out in front of their peers - something I’d normally consider a red line. But I did just that. My rationale was that (1) I needed to correct the problem NOW before other student’s took this as the standard, (2) I needed to clarify expectations and give better instructions and and (3) these girls were a bit more extroverted and I thought I could manage the correction without psychological damage.

Showing Them What “Good” Looks Like

I looked through my materials and found a topic that hadn’t been chosen for a presentation. Great - I’ll use that as an example. I went to the front of the room and said I’d like to give some feedback as well as advice for the rest of the class. I framed it as a “compliment sandwich” (putting the critique between ‘slices’ of compliments.)

“You both have very clear speaking voices…. You made great eye contact and connection with the audience… You used speaking notes and didn’t just read straight off a page…” etc etc. “But … there are some issues with the presentation. This isn’t what I was looking for… “

Their faces fell. I told them I feel I didn’t prepare them well enough so I’d like them to deliver a “part 2” in one of the following classes. I started talking to the class about what kind of content I was looking for, the level of analysis, the surfacing of nuance. I wrote notes on the board as I went. Students started to take their own notes. I told them I’ll give a brief example off-the-cuff on another topic. I pretended to be giving two different presentations. “Here’s the weaker version….” and delivered a surface level speech. “And here’s a stronger version…” and talked about the same content but at a deeper level of analysis, and connected themes together. The two girls started nodding. “Ah naruhodo” they said. They said they got it now. I asked them to recap for everyone else. They stood up and did so and gave great examples of how they could have done a better job. I closed the compliment sandwich by praising their quick understanding. And I told everyone my point here was not to embarrass them but to set everyone up for success. They did the “part 2” version a couple weeks later. To be honest it wasn’t the best, but it did meet the performance bar now that it had been clarified.

Fixing the Assignment, Not Just the Presentation

In subsequent terms/years I’ve iterated on this assignment. I give students my rubric, explain shallow vs deeper content, give examples of those and ask students to meet with me before their presentation date and have them show me an outline and their general direction. In those subsequent classes most students did a great job. (Because I intentionally set them up for success.) Live and learn. First fix the instance. Then go back and fix the system that produced it.

The Same Problems, With Engineers

I have found A. LOT. of my teaching experience and development has provided a good foundation for being an engineering manager.

I have experienced and learned first hand the need for clarifying expectations, being clear about the performance bar, being transparent about performance evaluation … and continuously reflecting and iterating to hone my overall approach and customize it to the situation I am leading.

When Nobody Sets the Standard, Everyone Sets Their Own

A couple times in my career as an EM I’ve found the whole team isn’t really meeting the performance bar due to no one having set standard practices. In those cases that lead to people merging directly to main, or self-approving PRs or approving PRs as a rubber stamp, all of which is a recipe for NOT holding any bar. In these cases created protocols and did code review for every PR for a while. I added documents on clean code/architecture, GitOps flow, testing requirements and a Definition of Done to the repo and went over everything as a group. I invited the team to contribute or suggest changes to these → the larger point is about establishing a shared practice of engineering and shared agreement of what “good” looks like.

Performance Standards Are Not Project Goals

With all teams I’ve been clear that performance standards are separate from project goals. You need to follow the standards and meet the performance bar. That is non-negotiable. But if you do that and we do not meet our goals, that is separate. If a model doesn’t hit a certain number that doesn’t reflect poorly on your performance. If evals have regressed and we have to push back a release, that doesn’t reflect poorly on your performance. But if, for example, you don’t write and run tests but do finish a task quickly that DOES reflect poorly on your performance.

Empathy Is Not Avoidance

With the pair that did a poor presentation I named the problem right away. I try to that that with my team. If someone is below the bar name it. I am empathetic, but that doesn’t mean I avoid tough conversations. Rather I use my empathy to deliver straight talk from someone who wants to see them succeed.