54

I've been participating in several multi-day hiring events as an interviewer and I noticed that candidates who got interviewed during the last couple of days had a higher chance of being hired.

For example 50 candidates are distributed on 4 days. The management set a target to hire 6-8 developers. During the first 2 days the bar is really high. Both the hiring managers and event organizers are very picky. We end up rejecting lots of the candidates.

As we approach the end of the event I noticed that the technical bar has been really lowered to the point that we started accepting candidates that wouldn't have passed a phone screening.

Now with the targeted number of candidates being hired upper management can brag about how the event was successful. This happened several times during the course of several years in different companies.

Is this normal? Do candidates being interviewed near the end of the event have a higher chance of being hired?

  • 38
    This video about mathematically choosing the right toilet on a festival is very relevant. – Sascha Sep 28 '16 at 19:49
  • 28
    Now here is 'my' question. What would it take to reconsider a qualified developers that was rejected in the first day? – AleX_ Sep 28 '16 at 19:53
  • 15
    Why would you make offers day by day instead of after the full 4 days you might might get a better candidate on day 4 when you have used up all your slots - sounds very unprofessional – Pepone Sep 28 '16 at 20:10
  • 30
    @Long that's insane which company's do this so I can put them on my list of ones never to work with or for – Pepone Sep 28 '16 at 20:38
  • 30
    "multi-day hiring events" What? What kind of "event" is this? I assumed you were talking about interviews being spread across multiple days but from your description these companies (plural? really?) have managed to turn hiring into some sort of farcical game show. Where on earth are you located that you can get away with a hiring process ripped straight out of The Voice? And for developers no less? – Lilienthal Sep 28 '16 at 22:03
69

Is this normal?

If this is the way it's happening repeatedly, then your company is doing it wrong.

If you really are forced to make on-the-spot decisions (rather than talking to everyone and moving some selected few on to the next step), then you need to go into the event with a target number of hires, and a clear understanding of the threshold for candidates.

If you find your 6-8 fully-qualified candidates the first day, then just pack up and go home. If by the end of the event you have only found 3 candidates that qualify, then again just pack up and go home. Accept that either can (and eventually will) happen.

Changing the threshold (because it's the first hour or the last hour) makes no sense. You miss out on perfectly good candidates, and accept less-than-good candidates that way.

Perfect is the enemy of good. Find good enough candidates, rather than holding out for perfection and then being forced to accept mediocre.

Do candidates being interviewed near the end of the event have a higher chance of being hired?

In your company's case, apparently they do have a higher chance of being hired nearer the end of the event. It shouldn't be this way. Candidates are either good enough, or they aren't. Being last to the event doesn't make you better.

  • 15
    I would add that they're also looking for the wrong things. Don't hire people who can do everything but rather hire people with a history of adaptation as well as aptitude and attitude. Too many companies think if they hire based on quantifiable experience primarily they'll get good people. In my own experience, adaptability and aptitude are much greater indications of success. – Chris E Sep 28 '16 at 19:59
  • 13
    @ChristopherEstep Right on target. I had a boss who didn't want the perfect technical fit. He would always say "I can always teach you a new technology, I can't teach you how to be a good worker." – Richard U Sep 28 '16 at 20:02
  • @JoeStrazzere I mentioned that because if you're eliminating many people at first and then hiring many later, your hiring process is almost certainly based on hard data over aptitude. If you're hiring for aptitude, you really don't have a "bar" where people are great vs those who wouldn't get past the screen. You're more looking for the people that fit without so much consideration for where they fall on the checklist. My point is not dissimilar from yours I think. First determine "good enough" then choose on attitude from that much larger pool. – Chris E Sep 28 '16 at 20:08
  • @ChristopherEstep it entirely depends on the tests/questions and criteria. Without going into too much detail, this is pretty much how selection for officer candidates works in the British military - there is a bar set and almost all of the criteria are "soft" skills like adaptability, leadership potential, teamwork. However, there is no quota to fill and no limit, either. I think the OPs problem is that they have a bar and a quota/limit, and these work at odds. – HorusKol Sep 28 '16 at 23:42
  • 2
    "Perfect is the ennemy of good". That's a… good answer. – Pierre Arlaud Sep 29 '16 at 8:11
43

This sounds somewhat akin to The Secretary Problem, and your problem runs deeper than you think.

The secretary problem is a series of n interviews, with the aim of choosing the best secretary. It has the limitation that candidates are interviewed in random order, and an answer must be given to each candidate after their interview. It makes the assumption that the interviewers are perfect, in that they can tell exactly how good that candidate is from the interview.

It originally had other names, such as the Marriage Problem, which in my mind is a more apt (although perhaps less PC) name, due to the requirement to respond to candidate x before talking to candidate x+1.

By taking the limitation that responses are given to the candidates immediately in The Secretary Problem, the chance of finding the best candidate drops from 100% to 40% for a group of 10, dropping down to 36.8% when interviewing many people.

So, the 'technical bar getting lowered in the last day' is not your problem. It is a minor, and probably inevitable symptom of your problem.

You should not reject any candidates on the first day, nor the 4th day. Do not reject any candidates until you have filled all available positions and you have the signed contracts to prove it. Unless a candidate is so bad that it would be worse to hire them than leave the position vacant.

Instead of treating each decision as a binary 'yes/no', you should be aiming to build up an ordered list, well past the number of candidates you are looking to hire. As each interview goes on, find somewhere in the ordered list to insert that person. When all interviews are finished, go through your list from number 1 until you have hired as many people as you want, or decided that it's probably worth re-posting the add and looking for more.

  • On the "Marriage Problem", I've also seen this talked about (by a comedian at maths gig) as how to optimise how many dates to go on before picking a partner given that you are only going to go on a specified number of dates. – Kvothe Sep 28 '16 at 22:22
  • 1
    Possibly you have the CVs of all candidates and you don't need the interviews to be in a random order. Maybe you can change a few things so that the order of the candidates is not so biaised? In any case it doesn't change the fact that the bar should be set and the expectations of the candidates fixed until the end of the interviewing process. – Pierre Arlaud Sep 29 '16 at 8:16
26

What you're describing makes intuitive sense.

The risk with going late is that they've hired 8 people and your chance is now 0.

The risk with going early is that they'll think "Wow. Great candidate, but this is candidate 3 of 50, think what better candidates are in that 47!"


For a weak or middling candidate, going later is clearly superior.

For a strong candidate, it shouldn't matter. And going early-mid may be superior. You don't want to be the diamond they can't hire because they've filled the spots.


The event you're describing reminds me of the NFL draft. Anyone can come out of the event claiming success. It's not until 3-5 years later that you can look back at performance and see how well you actually did.

7

I'm wracking my brain for a case where in a span of days or even 2 weeks I had to give final answers to a candidate before I'd seen a sampling of other candidates. I'm not sure what the structure of your events are, or if this is just a norm that I've never seen before.

I'll say this:

  • you want a clear set of "good enough" criteria
  • there are cases where the pool is so varied that "good enough" is tricky until you've seen enough of the pool to know what your trade-offs are - I find that the more rarely I hire, the harder it is for me to know what's realistic.
  • you want to avoid getting pressured into a situation where you have to make hasty decisions. If you have a 4 day cycle of 50 candidates, that all gets done in a week, it should be utterly rational to have a rating system, and get back to ALL candidates after you've seen the whole set. If your system is slower, you should be able to get a representative enough sampling that you can assume that more days of interviewing will not produce wildly different candidates.
  • realize that if you are working with a recruitment team, you have a case where your teams are working towards tangential goals - the hiring manager wants the best person within cost parameters, the recruiting team wants to close the deal on recruits quickly with a minimum of effort. That doesn't mean they don't care about quality - they care about the quality they are told to care about, and they know how to efficiently find it. But they don't have the headache of a poor quality choice - the hiring team does. So there is a bit of a push/pull - even in a healthy recruiting/hiring partnership of defining exactly what good enough, fast to hire enough, and cheap enough is.
  • generally when I do hiring, I have to be aware of the challenges between recruitment vs. hiring - and be careful of how I set up the parameters of the process. I have certainly had many cases where I worked with recruiters to figure out how long I could reasonably delay a decision to get as much information as possible before making a decision. The information could be "interviewing other candidates" or "digging more into the qualifications of this one" - but either way, that DOES mean more work for the recruiting team, as they have to keep communication going with all potential candidates.
  • "So there is a bit of a push/pull - even in a healthy recruiting/hiring partnership of defining exactly what good enough, fast to hire enough, and cheap enough is." Excellent summary of what the dynamic between managers and recruiters (even external ones) looks like, I'll have to remember this. And this is all good advice for when you need to quickly process dozens of candidates but can I assume from your third bullet point that you wouldn't advocate actually making a hiring decision in this kind of time frame? – Lilienthal Sep 28 '16 at 22:13
  • 1
    In the 3rd bullet, I'm trying to say that if you have a huge chunk of data coming at you - wait and make final decisions when the end of the known input set is reached. If you have a slower hiring stream, you don't have that luxury - candidates generally expect a certain rate of speed on responses or they will assume that nothing = "No, thanks" - so regardless of your data, there is an implied deadline. And... in my experience... if you are taking longer at this than a week - look at whether you are clear on your requirements, or if you have an excruciating process for deciding. – bethlakshmi Oct 3 '16 at 15:07
  • True, but hiring is a slow process in many organisations though, even if they're well-run. You'll typically want multiple rounds of interviews and cut people at each step. At the bare minimum you're looking at 2 in person interviews, probably more and that all takes time to set-up. I get that each interview round should ideally take less than a week but a hire/no-hire decision in one week or after only one interview seems very much outside the norm, even for entry-level positions. – Lilienthal Oct 4 '16 at 8:18
  • 2
    IMO - that assessment is very specific to the role and company for which you are hiring and also the expectations in the market. I usually try to write assuming as little as possible about those elements. My expectation, personally, when hiring SW developers, is 1 on site with 3-4 interviewers. More than that in my regions has proved over kill for the value of the extra information, and also likely to drive away candidates, but mileage varies too hugely on this for a regular rule of thumb. – bethlakshmi Oct 4 '16 at 18:09
4

It's human nature is what it is. You've pretty much described how many experience going to a bar but applied to hiring events.

We go into a situation determined to hold to some arbitrary standards because you're expecting the candidates to be better than they turn out to be. So as the event winds down, people begin to realize they've not had the results they wanted and so begin to get a little desperate and lower their standards.

They do this not just because of expectations too high, but also because nobody wants to make a decision and then have something better come along but they have no room for. That's also human nature. It applies to dating, buying a house, you name it. Everyone wants to get the best they can but they don't want to miss out, either.

Candidates do it too. Many will be hesitant if they start their search and get an offer in a matter of days. They start second-guessing the company as well as their expected wage.

  • I'm struggling to figure out how "going to a bar" fits into this? – Lightness Races in Orbit Sep 29 '16 at 10:37
  • @LightnessRacesinOrbit like people that goes on bars to find someone for the night (or the life who knows :p), expectations are high at start, they are quite lower at the end – Walfrat Sep 29 '16 at 12:39
  • @Walfrat: Oh, right. Could be clearer in the answer. Many of us usually go to bars/pubs just to have a good time with our friends, y'know? Not to pull. Bars are not just a dating venue. – Lightness Races in Orbit Sep 29 '16 at 12:41
3

You might be talking about Decision Fatigue, which is quite well documented. See this article for instance (emphasis mine):

The researchers examined 1,112 judicial rulings over a 10-month period. All of the rulings were made by a parole board judge, who was determining whether or not to allow the criminal to be released from prison on parole. (In some cases, the criminal was asking not for a release, but rather for a change in parole terms.)

But the researchers found exactly the opposite. The choices made by judges are impacted by all types of things that shouldn’t have an effect in the courtroom. Most notably, the time of day.

[...]

After taking a lunch break, however, the judge would return to the courtroom refreshed and the likelihood of a favorable ruling would immediately jump back up to 65 percent. And then, as the hours moved on, the percentage of favorable rulings would fall back down to zero by the end of the day.

Which might be the same unconscious thinking: "I don't want to keep deciding but we have to hire 6 people, so anyone will be approved".

-1

Have you considered that you are dealing with what is essentially the secretary problem?

The optimal solution to the secretary problem is to actually flat out reject the first N/e candidates (where N is the number of candidates and e is the base of the natural logarithm) and then select the candidate that seems better than the best candidate in the group you just rejected.

For your case, you are competing with other companies for the best candidates, but the theory still applies.

Turn up early to guage the level for the N/ecandidates and identify one of them as the best candidate in the group you will reject. When you have selected the candidate that is better than all in the rejected group, you have found the candidate who is likely to be the best candidate. Keep on selecting candidates until you have filled your quota for recruitment and you will have found the best candidates you can.

  • The problem with your algorith is that there is a 1/e chance (37%) that the best candidate will be in the gauge group, so you will either hire nobody or take last 8 candidates irrespective of their skills. Both outcomes are hardly acceptable. – Dmitry Grigoryev Sep 30 '16 at 12:11
  • @DmitryGrigoryev Afaik the algorithm guarantees 37% of the time that the first candidate that is better than everyone in the guage group is the best candidate in all of N – kolsyra Sep 30 '16 at 18:35
  • The "secretary problem" works under the assumption that you don't know how good a candidate is in absolute terms, only how good compared to all the others. It also assumes that you are the only one interviewing and hiring. At a hiring event you should know what to expect. – gnasher729 Jul 31 '18 at 22:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.