I want to outsource technical interviews
You know who mostly outsources admissions testing? Colleges.
Do they do a shitty job? Sure, yes. But ‘good’ in this context isn’t well defined. The SAT was invented in the 20s by a pretty bad dude. But it’s cheap and nobody benefits if schools to do this themselves.
I’m not saying standardized testing is great, and I’m not even proposing standardized testing for tech recruiting. But external test centers would save teams time and money. Think about it: at the point where you’re ready to hire, you have more money than hands. Outsourcing makes a lot of sense.
Even though tech hiring is expensive, sometimes unfair, and often cruel (bad coffee to bathroom ratio), most people I ask aren’t willing to outsource the interview piece. What follows is my argument that you should.
- Information quality is higher
- It’s cheaper
- It’s less stressful
- It may be fairer
- Maybe don’t outsource the whole process
Information quality is higher
If outsourcing means that we send over a list of questions and get back a list of answers, it requires us to be more accountable about what information we’re collecting from the interview.
I enjoy reading triplebyte blogposts just like you, but I’m leery of claims that there’s any science to interviewing. But if we’re forced to write down interview questions ahead of time, suddenly you have a well-coded dataset that gives us a shot at clean stats.
Companies are still providing the questions in my fever-dream here; if the test centers worked off a limited menu of whiteboard problems, it would be pretty gameable.
My cokehead boss when I worked at a bank used to drag me to interviews and be like ‘come on, you’re going to learn how to interview’. (The implied ending of that sentence, like most of his sentences, was ‘from me’).
Assuming that junior devs benefit from being on interview panels (big if), outsourcing interviews doesn’t rob them of that experience. It just changes the learning process from ‘conducting an interrogation’ to ‘crafting good questions’.
If you want to train your people to have interpersonal skills, put them on the sales floor; if you want them to learn interviewing, focus on the information at play.
When you pay money instead of time, the costs of hiring become very clear. This is good news because now you can price out tradeoffs: maybe you bring in consultants. Maybe you take a shot on someone ‘okay’ rather than running more interviews.
Standard test centers create possibilities for sharing information across companies. There are downsides to this (see ‘fairness’ section below), but benefits too. The information can belong to the candidates, like medical records, saving them time and saving future interviewers money.
It’s less stressful
Test centers can normalize the hiring process and that’s probably a good thing. ‘We’re here to ask questions, not make a decision, so you can relax’ is something we all try to say on panels but in this context can be true. There can be norms around ‘okay, you failed this question, no big deal, let’s move on’.
‘Should we stop early if they’re not doing well’ is a question teams ask when interview cycles are dragging on too long. I had a candidate one time who got grilled by one of our senior devs and my section was next. They asked whether they should leave in the middle. I said ‘no you’re doing fine’ which I regret – this person was asking for permission to bail, I should have been more receptive.
As a candidate, I’ve left early at panels that didn’t feel right. But there’s a lot of social pressure against this. The team conducting the interview panicked a little when I ditched.
It may be fairer
Would third party interviewing eliminate bad bias? I don’t know. Outsourcing the interview step doesn’t mean I won’t interact with someone on the team, even if just a voice call or a text chat, before getting hired. And you can detect a lot in a conversation.
Reduction of stress could matter; if you believe research on ‘stereotype threat’, you may believe that stress is a mechanism that makes race + gender a factor in test scores. If ‘test centers’ are lower stress than an on-site interview, that’s a thing.
Reusable test outcomes are bad news if they create a ‘permanent record’. People have bad days. The questions may be the wrong questions. At minimum, results should expire.
As a hiring manager, I’m going to want to know about the candidate’s background, even if it’s anonymized like ‘big 5 tech co doing X job’, ‘college with X rank’. I certainly want to check out their github if they’ve done OSS work. I maybe want to see them code, even if something short.
I don’t think ‘fully blind hiring’ is better in all ways, but am open to the argument that it’s fairer and worth some sacrifices to give it a shot. Fairness has to be balanced against effectiveness; if my blind hires end up not lasting, that’s a bad outcome. But it’s something we should try. Maybe they’ll be better than candidates sourced through the normal on-site process.
Maybe don’t outsource the whole process
There are benefits to meeting each other. If I’m not getting the job, I learn a lot about the company from the interaction. If I am, I get to vet.
On-sites let you show off stupid human tricks. One time I touch-typed a coding section, carrying on a conversation while I did it. All I proved is that I’m a good typist, but I’m pretty sure it got me the offer.
Interviewing sucks on both sides but is a form of networking and can be fun; we should streamline it but not neuter it completely.