Hide table of contents

Deliberate Performance in People Management


Summary: Managerial skills are consistently listed as one of the biggest talent gaps in effective altruism, including in this years' survey. This article summarizes some techniques to more rapidly increase people management skills, using the Deliberate Performance framework.


A few months ago, I took up chess. In that time, I've spent dozens – perhaps hundreds – of hours playing. Despite that time investment, I've barely gotten any better.


This phenomenon is not surprising. Experts have long understood that merely having experience with something (even having 10,000 hours of experience with something), doesn't necessarily make you any better at it.



Despite my practice, 11-year-old Praggnanandhaa Rameshbabu enjoys a healthy 99.74% chance of beating me, based on our rankings


The most valuable type of experience is what has become known known as deliberate practice. Deliberate practice differs from normal practice in that it focuses on activities specifically designed to improve performance deficiencies. For example, a musician may deliberately practice over and over again just the portions of a piece which give them the most trouble, as opposed to playing whatever they feel like.


Unfortunately, the business world is not very well-designed for deliberate practice. Very few of us can drop all of our other responsibilities to spend a week solely focused on repeating one small task over and over again to get better at it.


The situation is even worse for those of us who wish to improve our management skills. As an individual contributor, it’s sometimes possible to spend extra time on “throwaway” projects which don’t achieve any business value but help you understand the domain. But as a manager, we (hopefully) don’t have “throwaway” team members.


Michael discusses one of the downsides of using team members for HR practice.

Deliberate Performance

Fadde and Klein (2010) have introduced the notion of “Deliberate Performance” to attempt to address these shortcomings. Deliberate Performance activities are like deliberate practice, except they:

  1. Are tied to everyday job performance

  2. Do not impinge on the performance of the job task at hand (including not requiring excessive amounts of time)

  3. Offer varied repetitions with timely feedback, and

  4. Do not require expert judgment for feedback

Fadde and Klein have further refined the notion of deliberate performance into the concept of “Action Learning Activities” (ALAs). In this article, I will give some examples of ALAs which are related to management. I will give specific examples from my own work, focused around recruiting and training talented team members.

Action Learning Activities

Like all good business strategies, Action Learning Activities can be broken down into an alliterative typology. I go through the following ALA’s here:

  1. Extrapolation

  2. Estimation

  3. Explanation

Throughout, these activities will be guided by the principle:

I should try to be surprised as frequently as possible

It is only when you are surprised that you can learn something new.


An important part of deliberate practice is working at the edge of your comfort zone. The ideal practice tasks are those which are possible for you to handle, but challenging to do so.

In a business environment, it’s rare to have tasks which are consistently this challenging. Hopefully, most of us are not in environments where everything is constantly on fire, but a lot of us do have regular “near misses” which can provide learning through extrapolation.

For example, when interviewing job candidates, it’s (hopefully) pretty rare to find someone who absolutely refuses a job offer due to your vacation policy, but you might notice a candidate frowning when you are describing it to them. Extrapolation tries to leverage this for learning: what caused them to dislike your vacation policy? What if they had been completely firm on their preferred policy – how would you have handled that?

Extrapolation Examples

Note: in this and all other examples, names and identifying details have been changed or de-identified. Additionally, they are all pastiches of multiple events, so that no person can be identified.


What happened


Training team in (technology)

(Person) called me, upset at the slow progress of transitioning to (technology). I listened actively, which seemed to solve 90% of the problem. Also reiterated goals behind (technology).

What if they had threatened to quit unless we stop using (technology)?

Would have focused almost entirely on active listening. Would not have discussed any specific action plans at all (like them quitting or moving to a different project) until things had cooled down. Generally (person) seems reliable, so my bias would be to assume that they have solid reasons for being frustrated and that we should genuinely consider changing directions.

Upgrading to newer version of (technology)

We were a week behind schedule, and (person) was struggling to learn the domain.

What if next week we still have no progress?

The first step is to ask (person 1) to pair with (person 2) and check in after the first day. If (person 1) thinks that the pace is acceptable, then we may just have misestimated the complexity. If a lot more gets done during that paired day than usual though, consider swapping the projects that (person 1) and (person 2) work on. Also list this domain as a specific quarterly goal for (person 2) in quarter 2.


Detailed timeline estimation is a pretty standard part of a project manager’s skill set. But detailed estimation is also crucial to improving our ability to manage people, because it enables us to become surprised more frequently, and surprise leads to learning.

For example, suppose we hire a new engineer. I have a 30-minute conversation with her, after which I give her an evaluation of "average". Three months later we decide that her performance is below where she needs to be, and we have to let her go.


I'm probably pretty unsurprised by this. “Average” team members don’t work out 20% of the time, and 20% of events happen 20% of the time. Even though bringing her on board only to let her go was a hugely expensive endeavor (from both parties' perspectives), I've learned very little.


Alternatively, suppose I grill her during the interview process and make very specific predictions: her knowledge of the programming language Ruby will be the second best on the team, she will have a six-week period in which she struggles to adapt to the faster pace of our company as compared to her previous one, but after that she will do fine, etc.


No matter what happens after this, I'm almost certainly going to be surprised. My prediction was so specific that getting everything right would be like guessing next week's lottery numbers. As a result, I will have very specific ways in which I can improve my assessment of candidates.


Estimation Examples



What actually happened

Take away

Hiring (person)

Biggest concern is that they will not be motivated by the work. If they are motivated, then they will have absolutely no problems accomplishing everything. Their (technology 1) knowledge will be greater than everyone else’s on the team by a large margin, and it will take about three months before they get up to speed on (technology 2), after which they will be at least as good as (person). They will be able to contribute to architecture decisions immediately in (technology 1), and have some input into (technology 2) architecture within two months.

After checking references, it turns out that they very frequently struggle with completing tasks on time and do not have a strong sense of ownership. We made the decision not to hire them.

Despite the current engineering hiring vogue, resumes are actually an important thing to check over. Should ask about dates and reasons for leaving previous jobs, even if it seems pedantic, with every person.

Also, it is worth having a five-minute conversation with the few most recently-listed places on the candidate's resume, just as a sanity check. Lastly, even though many people I respect don’t do reference checks, I should’ve had a stronger Chesterton’s fence response against not doing them myself.

Guiding team through (training project)

I think this project’s tasks will be easier than the first project’s, and everyone will get them done on time. However, people will have challenges with problems D and E, which will require asking me for help.

Everyone did finish it on time, and only one person reached out for help. However, the instructions on D and E were ambiguous, which caused everyone to implement the solutions incorrectly.

I had noticed some slight ambiguity, but thought it would be understandable. Should make an extra effort when writing training instructions to be sure that everything is completely unambiguous.


Many organizations have a "post-mortem" process—when something goes bad, everyone discusses what went poorly and how it can be done better in the future.

"Explanation" takes this idea and applies it to individual-level areas for improvement: Why was I unable to clearly articulate the design requirements to my team? Why did our newest hire get up to speed twice as fast as the usual hire? It is important to note that you should be able to come up with explanations both for why things went more poorly than expected as well as those which went better than expected.


An important type of explanation is "intellectual theft": when someone does something better than you, figure out why they were able to do it better and how you can copy them.


Note that a good explanation is just a prediction in disguise: if you say that X happened because Y, you are implicitly predicting that Y will cause X in the future. If an explanation does not enable you to make a prediction, then that’s a signal that it is not specific enough.


Explanation Examples


What happened


(person) underperformance

I had been expecting (coworker) to let (person) go based on their underperformance. But they didn’t.

(Coworker) instead asked the recruiter if the refund period could be extended and the recruiter agreed. I had not thought of asking for the period to be extended, but it makes sense that the recruiter agreed.

User Interface for (project)

(Person) completed this much faster than expected

After talking to them, it turns out that they paired with (person 2). A noteworthy aspect is that this does not appear to have impacted (person 2)’s productivity. Should consider doing paired programming more frequently, as well as considering (person 2) for more mentoring roles.

Hiring (person)

(Person) was very excited to join us, and even turned down a higher-paying offer from another company.

It’s possible that (person) is making up the compensation for the other offer, but based on information on Glassdoor.com it seems likely that they are being honest. It seems good to me that we did “unique” things like the victory cheer during the interview process, and offered them the lead on a greenfield project, which was motivating to them.


Implementing these processes

I constantly have a spreadsheet open to which I append whenever something relevant happens. A template similar to the one I use is available here.

I also keep a notebook and whenever something surprising happens, I make a note to add it to the spreadsheet later with my explanation or extrapolation. Some common triggers for me:

  1. A member of my team is upset

  2. A member of my team is unusually happy

  3. A customer is upset

  4. A customer is unusually happy

  5. One of my coworkers did something other than what I expected


Some of us are fortunate enough to have managers who care deeply about our growth and provide great coaching. Most of us, though, will eventually have the “you aren’t ready for more responsibility but I can’t really explain why” conversation with your boss, and for many of us that conversation comes when we move into management.

It is crucial to take your career development into your own hands, and the Deliberate Performance framework is a helpful way of doing this. The framework can be applied to a wide variety of skills, and I hope the examples here provided insight into how to use Deliberate Performance to more rapidly improve your people management skills.






More posts like this

Sorted by Click to highlight new comments since:

This is a great article, and I will make one of those spreadsheets!

Though I can't resist pointing out that, assuming you got 99.74% out of an Elo calculation, I believe the true probability of them beating you is way higher than 99.74%. :)

This is a great system for setting up lots of tight and actionable feedback loops for yourself; thanks for sharing!

Another way to get timely feedback (whether you're a manager or not) is to literally ask as many people as you can to always give you immediate feedback. Ask people to immediately after a meeting or other interaction to tell you if they thought you did something wrong, or if you could improve something, or even if you did something particularly well. This way you can improve much more quickly than having to wait for the next one-on-one or quarterly performance review or so (or even never hearing the feedback at all).

There are entire books on feedback (like http://a.co/hIbhNZs) but this is a simple thing you could start doing tomorrow.

I use PredictionBook for keeping track of informal predictions I make but this provides a great way to formally apply it more in the workplace and could be extrapolated to interpersonal relationships as well. Thanks for writing!

I'm kinda surprised I haven't seen more information about deliberate practice as a manager. The specific issue mentioned here seems to be predictions made around people. Maybe in 10 years AI systems will be better than top managers are doing the sort of thing?

Curated and popular this week
Relevant opportunities