Hiring as an incomplete information game
Hiring is an incomplete information game and to succeed you have to treat it as such. What that means is:
Know what to measure — Your interview process has to be designed to systematically measure the things that matter to succeed in that role. To do that, you need to know what great looks like, and spell that out as concretely as possible, not in the abstract. It’s amazing how people consistently skip this step, instead sticking to boiler plate job descriptions copied from the internet. Writing down what great looks like in a role ends up telling you not just what to measure but also how (see next point). What’s worked best for me is to define a rubric: 8-10 dimensions that matter in the role, each with a set of 3-4 concrete examples for what both good and bad performance look like along that dimension. [1]
Optimize for signal-to-noise ratio — A typical interview process will top out at 4-5 stages, giving you a finite number of noisy samples to work with. Make them count. Every stage should be designed to home in on 2-3 of the dimensions that matter. Don’t leave it up to individuals on the hiring team to ask random questions. Agree on a pool of interview questions and draw from that. Each question should tie back directly to one or multiple of the dimensions that matter, and it should be clear what signal you expect to get from it. Again, it’s surprising how people consistently make errors here, preferring to stick to a set of interview questions they think “work” based on gut feeling or have asked AI to generate for them. Designing good questions takes time and it’s an investment you need to make. Great questions aren’t binary — they stretch candidates and allow exceptional ones to shine.
Calibrate — Hiring is a team sport and it takes time to arrive at a shared view of what strong performance looks when starting up a new process. It’s not uncommon for two people with the same background to come out of an interview with totally different views on a candidate. Take the time to calibrate your team to converge on what good looks like. This is also true when working with external recruiters — the feedback loop should be tight, especially early on.
Accept false positives and learn how to deal with them — You will have false positives no matter how good your process. What matters is your hit rate, and how you handle false positives. Cut your losses early. Take probation periods seriously. Aim for 90% hit rate. Anything lower than 80% is problematic. Understand that the cost of a false positive is higher than the cost of a false negative (see next point).
Accept false negatives and understand their relative cost — False negatives represent opportunity cost, which can be inherently hard to quantify, especially compared to the more explicit costs of a false positive. All else equal, I prefer false negatives to false positives. Sure, missing out on a great hire sucks, but the counterfactual is difficult to get a handle on: Someone you didn’t hire might turn out a superstar at another company, but cultural fit and the environment are huge determinants and there’s no guarantee they would’ve thrived on your team.
Speak to references early and take the calls seriously — Speaking to references is a simple and reliable way to expand your dataset, as long as you calibrate for selection effects. Take them seriously, and do them as early as possible in the process.
Work with high quality recruiters — The recruitment market is fragmented and full of mediocre players. But there are some that really earn their fees. There are recruiters I’ve worked with for many years, across different companies and domains, and they’ve consistently delivered. They might be expensive but the fees amortize much more quickly than people think. I’ve never regretted paying them, and besides most of the time there are clawbacks during probation (see point 3).
Be willing to pay top of market — Great hires produce many multiples of value over the cost of their comp, sometimes orders of magnitude. When you have high conviction in a hire, act decisively and make a strong offer. The amount of false economy I see in offer package construction is baffling. Shaving off 5% in base salary does not de-risk a hire at all (false positive) but makes you much more likely to lose a great one (false negative). Getting a clear read on a candidate’s comp expectations prior to making an offer is critical (this is where high quality recruiters shine). Benchmarks, both internal and external, are also very useful. Finally, note that there’s also signal at offer stage: It’s natural for candidates to negotiate, but those who drive an unreasonable bargain can end up disqualifying themselves late in the process.
Recruit from your network — This one is obvious. Having worked with someone before, even to a second degree, dramatically increases information completeness. No hire is risk-free, as cultural fit and alignment matter a great deal, but the risk is just so much lower if you have first hand experience working with someone.
Invest in your onboarding process — Hiring and onboarding are on a continuum. A poor onboarding process can almost completely negate the work you’ve put in recruiting someone great. And by the same token a strong process can help you identify false positives early on. The goal is to get new joiners to be as productive as quickly as possible. And what they need most for that is context — an understanding of how your team operates and what you are looking to achieve. Provide a high degree of hand holding early on, set specific goals and set the hire up for success. For example, a simple goal we set for all engineering hires is to have code running in production by the end of week 1. All hires have hit it so far. It’s a tangible success experience that gives them a 360 view of our development and deployment process. Plus it’s a good forcing function for us to think hard about what someone needs to know to do that.
Thanks to Paul Sobey for reading drafts of this and making valuable suggestions.
Footnotes
[1] As a corrolary to that point, I think this is also why people have a hard time hiring for roles they haven’t performed themselves before in some capacity. You won’t know what’s hard about it and what it takes to be good at it. This is especially worth paying attention to if you’re building your own business. If your startup is successful, you will need to hire senior folks at some point to run and scale different functions of your business. The more of their jobs you’ve winged before in some way, the better. The rubric will write itself.
