Hacker News Reader

> A quality hiring process would know with at least a 50% confidence level that the person is worth hiring before bringing them in for the onsite.

After having done a decent amount of hiring interviews and managing in my life, I don't think it's possible to have 50% confidence before you meet someone. The way you gain that much confidence is by 1- talking and listening to them, and 2- comparing them to other candidates.

> I'd still expect them to achieve their results with fewer than 3, and preferably fewer than 2 onsites per hire.

This seems hyper-focused on the minor efficiencies of interviewing, at the expense of employment efficiency; it's feels like an engineering approach that might not have all the requirements at hand.

The employee is (probably) going to be paid a six figure salary, and will have a large impact on the team. Great candidates can improve the efficiency of the people around them for years and years. Bad candidates can slow entire teams down. I've watched some excellent and prolific and opinionated engineers do major productivity damage -- millions of dollars worth, and I've watched some young CS graduates outlearn and outperform some veterans. Resumes & github accounts simply don't have enough information to know whether a candidate can help your company.

In my experience, it has been worth the time to let the hiring process go slow enough to be able to compare people's personalities, and not just their on-paper engineering credentials.

I think far too many engineers here on HN and out in the workforce dramatically underestimate how important attitude and people skills are in the hiring process, and dramatically over-estimate how important specific math & programming skills are.

As another person with some experience with hiring, I agree with you fully.

To put one of your points in my own words, your engineers are going to be spending the majority of their waking hours with this new person five days a week. Take the time to make sure it's a good fit.

As yet another person with some experience with hiring, let's not get theory and practice confused.

Taking the time to make sure it's a good fit is a worthy goal, but the industry is operating under some kind of delusion that there's a Ballmer Peak (http://xkcd.com/323/) to interviewing people somewhere between the 4th and 5th hour of the onsite interview.

Returns start diminishing very quickly after the first five minutes. This is why multiple people earlier in the process are necessary.

> Returns start diminishing very quickly after the first five minutes.

That's the opposite of a scientific approach. One of the best things I ever learned from an experienced interviewer is the spend the rest of the interview trying to disprove your first impression. Anything else is an exercise in confirmation bias. I'm not saying that I've never ended an interview early when it's obvious there is a bad fit. My time is important and so is other people's time.

A scientific approach is to pose a theory (this candidate would make a good hire) and perform a measurement (whoa, this guy's good) and then perhaps perform the measurement again, without bias, to see if the measurement is consistent and reproducible.

What you're suggesting is to spend the next several hours performing additional measurements in the hopes of finding one contrary to your first measurement. That is precisely the opposite of science. It's also how casinos stay in business.

> and then perhaps perform the measurement again

Not perhaps. Definitely measure again.

And that's exactly what I said. Don't trust your initial impression. Measure again. Get as many data points as you can.

> What you're suggesting is to spend the next several hours

I didn't suggest that and I'm not sure where you got that idea from. My interviews are usually less than an hour and then I hand the candidate over to the next interviewer. If they are a strong no from me, I'll even talk to the next interviewer to explain why and see if they even want to continue with the candidate. They usually don't want to continue with a "strong no" candidate, but sometimes the next interviewer might not share my concerns and they do want to continue with the candidate. Could be the person is a good fit for the other interviewer's team.

On the last point we agree, but you're missing the main point here.

Videoconferencing and phone calls can take care of not all but a large measure of the personality issues that can crop up. Many a brogrammer ends up wasting an onsite by skipping this step when the candidate could have saved them both the trouble, had they just taken the 5 minutes to have a Hangout.

The efficiencies of interviewing are anything but minor, and I find it odd how the wrong six-figure hire is bandied about but the existing 2-six-figure manager spending half his workweek interviewing nonhires instead of working is overlooked.

> Videoconferencing and phone calls can take care of not all but a large measure of the personality issues that can crop up.

You seem to be making large assumptions and arguing straw man issues. What makes you think that both the author and I don’t use videoconferencing or phone calls? I didn’t say anything about it, and the article mentioned screening calls right at the top; literally the 4th sentence.

> I find it odd how the wrong six-figure hire is bandied about but the existing 2-six-figure manager

Please explain. What’s “wrong” about my suggestion that an engineer will “probably” make something near six figures? Average salary for all engineering in the US is north of $85k. The average engineering manager salary is around ~$130k.

> spending half his workweek interviewing nonhires instead of working

You seem to be assuming that the manager is always interviewing, as opposed to, say, filling a few positions, working without hiring for a few months, filling a few more positions.

The guy in the article is a recruiter, so recruiting is his full time job. If a normal engineering manager has other work, then recruiting isn’t their full time job, and they might not be hiring at all times. I haven’t ever worked in a department that was constantly hiring, neither in startups nor in huge global companies. Have you?

Suggesting that the cost of interviewing a dozen people is larger than the cost of employing one not-so-great engineer seems, from my perspective, to be off the mark.

> What makes you think that both the author and I don’t use videoconferencing or phone calls?

Maybe it would help if you understood it as "use [enough] videoconferencing or phone calls".

Answering that question: For the author, the aforementioned 5X ratio. For you, your apparent insistence that a) you have to talk and listen to them, but b) this somehow has to take place in person to get to 50% confidence. Maybe that's true for you, but I doubt it's universally so.

> What’s “wrong” about my suggestion

I think you parsed my sentence incorrectly. "wrong" modified "six-figure hire", as in, the wrong guy who costs six-figures.

> I haven’t ever worked in a department that was constantly hiring, neither in startups nor in huge global companies. Have you?

Oh yeah. It's more like everyone had two full-time jobs, and one of them was 30 hours a week of interviewing.

> Suggesting that the cost of interviewing a dozen people is larger than the cost of employing one not-so-great engineer seems, from my perspective, to be off the mark.

Let's try from my perspective. Say you hire a not-so-great engineer, who's maybe worth $100k, for $90k. A great engineer would be worth $1M, for $90k. Both are wins.

O that it were only a dozen people. Say it was 5 hours a week of wasted interviews, for 50 weeks, and the interviewer earns $100k. This puts a loss of >$10k onto your bottom line, neglecting the "real work" the interviewer could be doing, and the other employees also doing the same thing.

A low win beats any loss.

> Maybe it would help if you understood it as “use [enough] videoconferencing or phone calls.”

Again, what makes you think the author isn’t using enough? Why are you so certain you have an important point here that anyone else who’s hiring doesn’t already know? I think you’re making an incorrect assumption that the author is being unreasonable.

> the aforementioned 5x ratio

That was the ratio of candidates to jobs. It doesn’t say anything about how much screening was involved. Interviewing 5 candidates for a job seems completely reasonable to me, and not out of the norm of hiring for any company I’ve ever worked or hired for. I might screen 10 or 15 before interviewing 5 on-site.

> I think you parse my sense incorrectly.

I did, thanks for the fix.

> Say you hire a no-so-great engineer, who’s maybe worth $100k....

I already mentioned I’ve seen people lose companies millions. Engineers who are great on paper, and great in interviews. The bottom end here is not a small but positive win, it’s an enormous loss.

> A great engineer would be worth $1M

Where does this number come from?

> Say it was 5 hours a week of wasted interviews, for 50 weeks, and the interviewer earns $100k. This puts a loss of >$10k onto your bottom line.

This thinking still seems like penny-wise and pound-foolish to me. I don’t consider any interviews wasted. I’m learning in all of them, and I’m comparing the candidates. All samples are valuable. Also, you’re talking about the cost of 125-250 interviews over the course of a year, if I assume it takes 1-2 hours to interview and give feedback. If we use the author’s 5x, that means you hired 25-50 people. You’re comparing a measly $10k to the cost and value of more than 25 people. Spending $10k to hire 25 people is nothing. Companies pay more than $10k for a single hire, all the time.

> O that it were only a dozen people.

A dozen people per job is what I was talking about, just to be extremely conservative... more than the 5x the author mentioned. And with a dozen people per job, I’m saying the cost of interviewing is far less than the cost (or value) of employing them, so it doesn’t hurt to interview that many.

I would think of hires as compound interest, where interviews are linear cost. Comparing them already isn’t apples to apples.

> > A great engineer would be worth $1M

> Where does this number come from?

In 2017, Apple's revenues were $229B. Apple had 123,000 employees. This means the average employee at Apple pulled in $1.86M, not in salary, but in value to the company.

Check out Paul Graham's wit and wisdom with regard to only seeing the positives of interviewing en masse.

http://www.paulgraham.com/artistsship.html

Every check has a cost.

> The bottom end here is not a small but positive win, it’s an enormous loss.

You used the term "not-so-great", not "bad". You're not going to spot all the bad ones. And you can fire bad hires, limiting your downside.

Consider two possibilities:

•Hiring managers genuinely do select the top talent, which means that now the only people they have left to interview aren't qualified to hold the jobs, and thus they're wasting their time.

•Hiring managers don't know everything, try to make the best decisions, but often hire poor performers, and this means that their role in hiring is not as some kind of highly skilled "nerd whisperer" who can divine good candidates from bad candidates, but merely a utilitarian cog in the corporate machine, that tends to overestimate both its skill and its importance. If this were medicine, it would be institutional quackery.

I don't know he did mention there are phone-screens ahead of time. Also mentioned that the reason they are bringing in more is to figure out who they are before simply disqualifying the candidate because they don't meet the specs. If you simply don't agree with that premise, suggest that, rather than you would be embarrassed to admit those metrics. That's more of an insult than describing how the method he describes specifically and the reasons behind them are wrong.

Also those 400 interviews happened in 7 years, which is one per week. I am currently in a hiring phase and we do 1 or 2 per week.

The process is usually one person doing one 20min behavioral phone screen, and maybe one technical person doing one 60min technical phone screen, give or take. Maybe a homework. Then maybe 3-6 one-on-one interviews for a 5 hour onsite. Fiddle with the numbers if you need, but this covers 90% of companies interviewing today. That's way too few phone screens and way too many onsites if you don't get rid of enough candidates at the phone screen stage.

Simply disqualifying the candidate because they don't meet the specs happens anyway- the question is, are you a highly skilled hiring manager by coming to this realization quickly, or only after you have the candidates come in and you waste an afternoon introducing them to everybody?

I personally feel this industry could use more insults going the other way. Until hiring managers realize they have been trying to source talent with their middle finger out the whole time, they're not going to realize they need to change.

Not in my experience. I used to try to screen resumes and to pre-screen on calls. And it mostly doesn't work. Some people ace the pre-screen and are assholes in person. Some people totally fail the pre-screen because it's the first ever call they had in weeks and they get super nervous.

I will use a 15m pre-screen to verify some basic tech skills but I also then offer candidates to follow up in an email if they think they could have done better.

> If anything your engineers did, anything, had a 20% success rate,

They don't have a 20% success rate. The success rate is 100% if they hire the number of developers they need and to the standard they want.

Saying it's a 20% success rate is like saying interviewing 100 people to determine the best among them is a 1% success rate. But it's a 100% success rate if you did actually figure out who the best programmer is among the 100.

So you pre-screen on calls, but do you get any of your co-workers to do it? There's no hard and fast rule that you have to spend 30 minutes to an hour per whack, either. You get most of your signal in the first 5 minutes (first impressions and all that).

Offering a "grievance process" where nervous candidates can plead their case for a mulligan telegraphs a (probably deserved) lack of confidence in your own company's hiring practices. The only thing worse than a first date that went bad, where they want to submit to you in writing how they could do better if given a second chance, is where they want you to submit that to them in writing.

Your last paragraph totally makes my point. The purpose of an interviewing process for any serious business is not to find the best candidate, but to fill the position with a qualified hire who works well with others.

The 98th percentile score should not be a failing score. You're not picking Miss America, and most of the other 49 candidates are pretty enough, especially when pretty is so subjective.

> You get most of your signal in the first 5 minutes (first impressions and all that).

One of the best things I ever learned from an experienced interviewer is to spend the rest of the interview trying to disprove your first impression. Otherwise, you're just making yourself a victim of confirmation bias.

> Offering a "grievance process" where nervous candidates can plead their case for a mulligan telegraphs a (probably deserved) lack of confidence in your own company's hiring practices.

I have a great deal of confidence in my hiring practices. I've rarely had a candidate reject an offer, so I'd guess the candidates also like it. I've even had candidates tell me that they think the interviewing process I've designed is really good and they learned how to be a better interviewer.

It's not a really a grievance process. It's a simple "Here is why we don't think you're qualified (e.g you don't actually know any JS in reality) but if for some reason you think we got that wrong, please let us know." Once in a blue moon someone will come back with a legit response.

> The purpose of an interviewing process for any serious business is not to find the best candidate,

No true Scotsman? I sometimes have extremely specific hiring goals. I may for example want to hire the best JavaScript person available in town because the company needs to be a thought leader in that area. Every company is different. Don't assume your hiring needs are my hiring needs. For sure when my hiring needs are to find any qualified candidate, that goes pretty quickly.

> I've rarely had a candidate reject an offer, so I'd guess the candidates also like it.

How many failed candidates did you ask?

I follow up with rejected candidates from time to time with a phone call. Often I'll follow up with a phone call and talk about the areas I think they could improve in and recommendations about how they could do that. They seem to really appreciate it.

I'll close the call asking what they think about our interview process. They are mostly very happy with it and disappointed they didn't make the cut.

I've unfortunately never had much constructive criticism about my interview process from a rejected candidate. The only criticism has been the rare few times someone thinks they are qualified and kind of lashes out because they didn't make the cut. "Well I didn't want to work for your company anyway" type of thing. Not much valuable info to go on.

The valuable criticisms I have had are from colleagues over the years who have been in interviews with me participating themselves. Some of what I've improved over the years has been related to those criticisms.

How about yourself? Do you ever follow up with rejected candidates about your own interview process? What techniques do you use to measure your hiring effectiveness and what steps are you taking to improve your process?

I take a different approach; I've zoomed out from the problem, and looking at it from a macro level, I've concluded that the system is what's corrupt, not the candidates, recruiters, or hiring managers, and all of the individuals are victims toiling under a broken system.

It's a business opportunity, and it's one I'm going after once my current project gets to a stopping point.

Facebook's guidelines were 25% confidence they could pass an onsite and they would be given one.

I'd personally rather give the candidate AND the company a chance at hiring a potential candidate than missing out on one. The longterm impact on your company is huge.

This fails to consider the fact that with 25% confidence, 3 out of 4 candidates have no chance, and you're just wasting time, theirs and yours.

There are two kinds of hiring managers- those who think the system works and they're skillfully picking the best talent, though at the last minute, without formal education on the subject, without metrics, without reproducibility, without accountability, and with minimal prep time; and those who recognize the system is deeply flawed and go along to get along.

That is only wasted time if there was next best opportunity. People are looking for work so going for on-site interview is not waste of time.

What would you do, go to the beach, hang out in a bar? So your time would be spent better. Maybe you would go to other interview where you would be hired but you cannot know this in advance.

The same with company they don't know if other candidates would be better until they actually do on-site.

If you can predict future then yes. But most don't.

> That is only wasted time if there was next best opportunity. People are looking for work so going for on-site interview is not waste of time.

What's the unemployment rate for top-end talent in Silicon Valley? 2%? 1%? This foolish strategy of candidates-time-is-free is designed to turn off people who already have jobs and want an upgrade. Not only are you _not_ hiring the best, you're limiting your applicant pool to the worst.

<sarcasm>Why, sure, I'll blow 7 hours including 2 hours' roundtrip commute to audition for a 20% chance of a $10k raise.</sarcasm>

this! At Google, saw this and signed up for 100% phone screens, where I could have much more impact on improving the process, esp for experienced engineers with 7-10+ yrs experience. Conveniently, most people hated doing phone screens, so HR was appreciative as well.

Given standardized questions (with depth - impossible to memorize), you can quickly calibrate to the organizations expectations by following candidates that you approved for interviews.

1:5 and 1:10 do seem atrocious, but looking back on my tech career, I probably applied to about 100 companies for every 10 interviews I got, for every 1 offer, so it sounds realistically atrocious.

He says he interviewed X candidates to hire X/5 engineers. That means they probably selected more than X/5, but that’s the number that accepted offers. Maybe the problem is their offers aren’t competitive.

I totally agree that seems like an excessive amount of interviewing for engineers on the team.

Let me take you further in this direction of thought- many offers aren't accepted because the candidate, after having been through a dysfunctional interview gauntlet, realizes a different opportunity is better (sometimes after waiting weeks to get the start date).

If interviewing has been skillfully conducted, and the candidate gets an offer in 24 hours, there's substantially less likelihood they'll turn it down, because they won't have had as much time to interview elsewhere, or wonder what organizational deficit is causing the delay.

That is certainly true. Since you had already jumped to implicate the interview process as the reason it took so many candidates to make a hire, I was simply offering an alternative. In fact, it could be that both the interview process and the offers are flawed; there can be any number of reasons why a candidate doesn’t convert to an employee, and you’re solely focused on one.