I previously wrote about the challenges of capacity building – hiring, training, and managing a team. We thought we would share some of the principles and practices we’ve come to believe are important to this goal, so that others can both learn from/consider them and provide their thoughts. As in the previous post on this topic, “we” primarily means Elie and myself, as it’s only fairly recently that other staff have taken on managerial roles.

An overriding theme is that of putting substantial work into capacity building (training, evaluating, and managing employees) and constantly integrating capacity-building considerations into our everyday work. In every piece of work we do and every assignment we make, we ask whether the way we’re going about it is optimized for long-term capacity building (training and evaluating other staff such that they end up in the best possible roles), and not just for getting the work done as quickly as possible. Maintaining this attitude can be a challenge, since it involves focusing on a longer-term (and hazier) goal than completion of the work at hand, but we believe that it is the right long-run approach to building a team.

• We try to think constantly about whether everyone, including and especially ourselves, is assigned work that’s a good fit. Whenever I start a task, I ask, “Am I the best person to do this task? Can someone else on staff, who has capacity, do it just as well? Why does it have to be me?”
• When we don’t have much sense of what we’re expecting from an assignment, we’re likely to take it on ourselves. By doing the work ourselves, we’re able to get a better picture of how relevant it is and what it consists of, which is necessary if we’re hoping to later assign similar work to more junior staff. (We do have a “Senior Research Analyst” distinction for other staff members that we’re comfortable assigning extremely open-ended work; currently we have one Senior Research Analyst, Alexander Berger.)
• When we’re doing work ourselves, we often explicitly approach the work with a goal of “getting the work to the point where other staff can carry it on” rather than with a goal of “completing the work as efficiently as possible.” This means a lot of reflection and (sometimes) writing about what we’ve been doing, and thinking about what skills and knowledge base it requires. For example, Elie recently completed a shallow investigation of developing-world infrastructure that we explicitly thought of as a way to reflect on how shallow investigations should be done in general; I am currently focused on learning enough about political advocacy and scientific research to create shallow investigation assignments in these areas for other staff. In general, we see much of our GiveWell Labs work as “learning how to do investigations (so that we can build staff capacity for them via hiring and training)” rather than as “simply trying to understand the topics we’re exploring.”
• When we assign work, we usually are optimizing for the goal of “training and evaluating staff” rather than simply “completing the work as efficiently as possible.” Ideally, we want each assignment to help the assigned employee learn, and to help us learn about the employee. We think about how each assignment fits into our picture of the employee’s strengths, weaknesses, areas for improvement, and possible long-term trajectory. Sometimes the primary purpose of an assignment is to evaluate an employee, including “stretch” work that gives the employee an opportunity to show that they can take on more responsibility than they have in the past.
• We try to have a reasonably clear picture of what constitutes “excellent,” “good” and “subpar” work so we can make accurate evaluations. Doing so can be very difficult, and is usually more tractable when we have more experience with a given assignment.
• When we have an assignment that’s particularly easy to evaluate, we’ll sometimes assign it to multiple employees over time for calibration purposes (i.e., getting a read on different employees’ strengths and weaknesses).
• When assigning work that is very different from what an employee has done before, we often follow the work very closely, and we are often prepared to spend more time managing the assignment than it would have taken to complete the assignment ourselves. We seek to thoroughly understand, and give our input into, the employee’s thought process and progress, so that the output is a meaningful representation of what will be produced on future similar assignments. We’ve found that a major determinant of an employee’s fit with GiveWell is their willingness to constantly check in with us and keep us posted on their thought process, so that we can stay on the same page.
• When we gain confidence in an employee’s ability to do a particular kind of work, we step back and reduce our involvement – but we still take periodic opportunities to check in at a greater level of depth. Occasionally Elie or I will “double-do” an employee’s work to see whether we pick up any new insights about how the work is being done and how it could be improved. Most conversations at GiveWell, external and internal, are recorded, and Elie has found it helpful to listen to them when he can.
• We put significant time and reflection into considering employees’ long-term trajectories and try to make assignments consistent with these.
• We spend significant time in discussions with employees, getting on the same page about both strengths and weaknesses of recent assignments and likely long-term trajectories. It’s important that employees know where we stand and have the opportunity to point out when they disagree with us on these matters.

We’d welcome thoughts from others on these principles and practices.