A lot of the way recruiting works makes little sense when you step back and look at it. Economists, psychologists and operations researchers have been studying hiring for decades, and much of what they have found never reaches talent acquisition teams. So what can that thinking tell TA leaders about the problems they are dealing with right now?
My guest this week is economist Andrew Marritt. Andrew has been writing some fascinating essays about recruiting on his Substack, Working Ideas, and I wanted to get him on the show to explore them in more detail. In our conversation, he explains what AI has done to the signals in job applications, why cost per hire can lead to the wrong budget decisions, what the evidence says about hiring managers overruling the process, and what hiring might look like if it were rebuilt from first principles.
In the interview, we discuss:
- What economics, psychology and operations research can teach talent acquisition
- Why many of recruiting’s problems are nobody’s fault
- How generative AI has weakened the signal in job applications
- Ghost jobs and employer brands that all sound the same
- Why cost per hire is the wrong measure for a recruiting budget
- The real cost of cutting structured interviews
- What happens when hiring managers overrule the process
- The evidence on AI interviews
- The questions to ask AI vendors about training data and fairness
- What would hiring look like if it were rebuilt from first principles?
Follow this podcast on Apple Podcasts.
Follow this podcast on Spotify.
Key takeaways
- An economic reading of recruiting suggests that many of its problems come from candidates, employers and recruiters each responding sensibly to their own incentives, not from anyone behaving badly.
- A signal only helps an employer if it is costly to fake, and generative AI has removed most of the cost of producing a tailored, well-written application.
- One study of a freelance marketplace estimated that when AI removed the cost of writing a tailored proposal, workers in the top fifth for ability were hired 19% less often and those in the bottom fifth 14% more often.
- Cost per hire is an average that captures neither the value of the people hired nor how often a role has to be refilled, so reducing it can cost a business more than it saves.
- Raising average tenure can cut the recruiting budget without lowering the quality of hires, but tenure sits outside the cost per hire calculation and largely outside the recruiting team’s control.
- Decades of studies show that a consistent process usually predicts better than expert judgment, so hiring manager overrides should be allowed but recorded and reviewed.
Transcript
Matt Alder 0:00
Many of the problems in recruiting are actually nobody’s fault. Candidates, employers and recruiters are each making reasonable decisions for themselves. But together, those decisions can sometimes produce a hiring system that serves none of them well. Research in economics and psychology has a lot to say about why this happens. So what would it take to fix it? Keep listening to find out.
Advert 0:28
Support for this podcast comes from Ashby. Ashby All-in-One is a fully integrated solution combining ATS, CRM and sourcing, scheduling and analytics into a single source of truth, giving recruiting teams precise control over every stage of the hiring journey. With embedded AI assistance and automation at its core, Ashby empowers talent acquisition teams at companies like Snowflake, Ramp, Deal, Zapier and Deliveroo to reach the highest talent density so they can better serve customers, close more deals and ship products faster. Visit ashbyhq.com slash recruiting future today to see how it all works. That’s ashbyhq.com slash recruiting future.
Matt Alder 1:38
Hi there, welcome to episode 839 of Recruiting Future with me, Matt Alder. A lot of the way that recruiting works makes little sense when you step back and look at it. Economists, psychologists and operations researchers have been studying hiring for decades, but much of what they’ve found never reaches talent acquisition teams. So what can this type of thinking tell TA leaders about the problems they’re dealing with right now? My guest this week is economist Andrew Marritt. Andrew has been writing some fascinating essays about recruiting on his Substack Working Ideas, and I wanted to get him on the show to explore things in more detail. In our conversation, he explains what AI has done to the signals in job applications, why cost per hire can lead to the wrong budget decisions, what the evidence says about hiring managers overruling the process, and what hiring might look like if it was rebuilt from first principles.
Matt Alder 2:37
Hi Andrew and welcome to the podcast. Just start by introducing yourself and telling everyone what you do.
Andrew Marritt 2:42
I’m Andrew Marritt. I’m based in Graubünden in the southeast of Switzerland. At the moment I’m spending most of my time teaching master’s level data science and analytics at a hospitality school and also working and advising HR teams as well. Before that and probably what you and where you and I met the first time, I spent 12, 13 years running one of the first people analytics firms here in Europe. Before that I spent a 14, 15 years working in recruitment or recruitment-related roles, doing things such as employment marketing, data-led candidate research, choice modelling, building the first global career site for UBS. I’m an economist by training. I also studied operational research, and both those shape how I look at problems. And these days, as well as my teaching, I write a Substack called Working Ideas. And earlier this year, I also wrote a textbook called Bayesian Thinking for People Analytics.
Matt Alder 3:49
Fantastic. Now, I check back, and you have been on the podcast before, you were on in 2016, which is 10 years ago, on episode 53. So now, you mentioned your Substack there, and you’ve been doing some really, really great writing, writing some really thought-provoking long-form essays in a world where no one’s writing long-form stuff anymore. So it’s really kind of refreshing to see. Tell us a little bit about what you’ve been writing about, why you’ve been doing it, and what the sort of the key theme running through everything’s been.
Andrew Marritt 4:22
Yeah, sure. So I started the Substack. I’ve been writing back since 2003 on and off. Recruiting is where I started my career. And it’s something from a HR analytics, people analytics, however you want to call it, perspective is I’ve always felt it’s completely undervalued. It’s where I started learning all my people analytics stuff. And in terms of hiring we’ve got five academic fields that have been sort of looking at hiring and they just don’t read each other’s work. So you’ve got personnel economics which is my area, you’ve got IO psychology which is probably the area that most people analytics teams seem to have come towards in the last four or five years, there’s sociology, there’s operational research, and there’s computer science. So, you know, as an example, this guy called Michael Spence found signalling results, which we’ll talk a bit about, I’m sure, in 1973 in economics. Zahavi found the same result with peacocks in 1975, and employer branding, you know, Simon Barrow’s original works, they started in 1996 and nobody cites each other. So we see time and time again that type of thing. What I like pointing out as well as this sort of multidisciplinary approach is parts which I call individually rational and collectively daft. So most recruitment problems aren’t caused by anybody behaving badly, they’re really just what a system produces when everybody acts sensibly on their own incentives.
Andrew Marritt 6:03
And in terms of my writing I start from the decision, and for me analytics always is about how can we make better decisions and then I go back to first principles, I look at the research, and the value usually turns up as a connection that somebody’s missed.
Matt Alder 6:20
I love the way that you write because, first of all, I think the multidisciplinary approach is fantastic and it’s a great way of really kind of understanding how things really work and the things that we’ve missed and why so many things about recruitment make no sense when you step back and look at them. And I think also the fact that you’re asking those questions, why do we do it this way, and going all the way back to how that formed. So, there’s so much we could talk about. So, I thought that perhaps we could just sort of look at a few common scenarios and sort of reflect on some of the things that you’ve written. So, let’s start with the problem that comes up in almost every interview at the moment, which is employers are getting more applications than ever, the applications all look the same, candidates complain they’re not getting seen. It’s kind of a mess out there in terms of the basics of how recruiting is working or not working at the moment. Looking at this through your economist perspective, what’s going on and who’s fault is it?
Andrew Marritt 7:21
Yeah, I think it’s a great question. So I think the easiest way of framing this from an economist’s perspective is to think of signalling, which I just mentioned about a few moments ago. And what we’ve got here is a complete collapse of this value of information. So a signal only works if it’s costly to fake, right? That’s the important part. Signals are when some people find it genuinely easier to show something than others. So a good academic student finds it easier to get a degree than somebody who’s got to labour and work every evening into the late hours. So although we don’t necessarily need that particular degree, we can then use that signal as a way of understanding the quality of the individual. Generative AI has made this sort of polishing application almost free, right? The marginal cost has dropped down to zero. Now, there’s a great instance that’s helped a lot of the way we’re thinking about this, and this is about Freelancer.com. I can’t remember how long ago it is, but they added one-click proposals. And the link between how tailored a proposal was and getting a reply fell 51%. So we economists love looking for these things. We call them natural experiments. And there’s a huge host of papers, at least from the economics, because everybody looks for natural experiments. So we’ve got a paper by Goldin and Stüdio that said that following this, the top fifth workers were hired 19% less. And the low and often bottom fifth workers went up 14%. So this is showing this sort of signal collapse that we talked about. But employers also have the same problem for the other side. So, a recent paper suggested that up to 21% of job postings might be ghost jobs, right? When the employer costs almost nothing to post, they post more and more often. So candidates take the rational approach, they discount all of the job postings. And the people who most pay for this are the ones with real job postings, right? Because it can’t work out which is which. So we’ve also got a stage where employer branding has converged. So no firm can stop making claims without losing applicants. So they claim everything. And in terms of the branding perspective, it isn’t additive. You don’t add up the claims. If you make more claims, you just spread the value over a weaker set of claims.
Matt Alder 10:12
So from an employer branding perspective, all the employer brands are sounding the same as well, because everyone’s claiming everything, yeah.
Andrew Marritt 10:18
Precisely. Everybody’s saying exactly the same. And I remember at UBS, I convinced my Swiss recruiting team, which we were the biggest, to employ a copywriter, because the recruiters wrote recruitment copy and they weren’t doing advertising. And these days, you’ve got generative AI obviously writing quite a lot of parts of ads, and it’s just converged into a nothingness. There’s nobody really at fault for this. These are all rational decisions. So everybody works rationally, but collectively, it’s just daft. We’ve got this daft situation. The thing that survives are signals that can be costly and can be checked. And this has to some degree always been the case. So things like work samples, structured interviews, verified work histories. The employees on Freelancer shifted weight onto verified histories when this happened. And it’s not all bad. I mean, we all think about the bad things because this is the stuff we have to deal with. But AI has improved some things like the ability for non-native language people to be treated equally. And also, I think we were looking at signal in terms of application as good at writing applications rather than necessarily good at doing the job. And to some degree, it’s flattened out that curve. But ultimately, from an economics perspective, what we’ve got now is the rational thing of a system where everybody’s acting in their own interests and collectively, we’ve got the problem we have at the moment.
Matt Alder 11:59
That’s fascinating, actually. And it’s like, AI is just like a catalyst for this. It’s not the kind of the root cause. I think that’s really interesting. To move on to something else that you wrote about, which is recruitment budgets, for example. So lots of pressure on recruitment budgets at the moment. Theoretical scenario. CFO is asking for a 20% cut in the recruitment budget. What comes back is, you know, TA teams will use things like cost per hire and lots of metrics that don’t perhaps resonate with a CFO. What’s the kind of the answer to this? What’s the challenge to this around recruitment budget?
Andrew Marritt 12:39
Yeah, so cost per hire, and I think it’s good that we’ve got a sort of a standardised approach of doing cost per hire. That’s one of the changes since I started doing people analytics. But it’s an average, and a budget depends not on the average, but on the last dollar spent, so to speak. It also contains, cost per hire contains the value of the hire, so how much those people are going to bring and contribute to the firm, nor how often the role has to be filled. And one of the things I’ve always looked at is that averages especially can be gamed. And if you give somebody a target, they’ll game targets, right? This is a rational behaviour of every individual. So given that we’ve got a target that doesn’t capture all of the work being done, it doesn’t come to capture the value and it doesn’t capture the replacement costs, you’ve got something called suboptimisation. And there’s a whole different area. Every area has got a different view of this. So it’s either in operations research, suboptimisation, there’s a chap called Goodhart who used to be on the Bank of England’s committee and was LSE. It’s in Goodhart’s law. So if you focus on one aspect and only on one aspect, then everybody goes towards that one aspect. What I said in the article is that the recruitment budget doesn’t depend on the recruitment manager because there’s a bunch of things that aren’t in the recruitment manager’s brief that are really important for improving the value that the recruitment function does. And therefore, I said in the article, actually, the only person who has this full remit is the CHRO, because they’re the only one who owns recruiting cost, replacement cost, and the performance aspects of it.
Andrew Marritt 14:40
If we think about budget, we’ve got a sort of spend per hire, the total budget of the recruitment department includes the spend per hire, the number of positions they have to hire, and also the tenure of the individual. Tenure is a lever on the budget, and at the moment, cost per hire can’t see it at all. So you can think about, you know, if you’ve got a position like I worked in call centres and they’re replacing the same seat many times in some instances a year. The more times the recruitment team have to replace the seats, the more volume they get, the average cost per hire goes down. Right. So the firms tend to optimise on the positions and the recruiting, not the sort of seats. I also talked about the value added, right? So a good employee adds more value to the company than the salary and their recruitment costs. Otherwise, we wouldn’t be hiring them. And we can start to model the expected value that we’re getting there. And what we see if you’re looking at cost per hire is often the easiest ways of reducing cost per hire destroy much more value in terms of the firm than they realise by saving budget. So the example I gave in the article was the structured interview, which all of the IO psychology suggests is the best of the ways that we do a selection. The numbers I plugged in sort of suggested if you got rid of that and just let managers have unstructured, non-trained interviews, you lose about £48 of value for every £1 saved, right? You’re dropping test stages, lose about £4.40 per every stage. You could reduce the number of people you interview, right? And that, depending on how many you’re doing, interviewing for every role loses value. At some stage, the additional interview doesn’t really create any value. But raising tenure, if you raise tenure from four years to five years on average, that would do 20% cost savings, and the value would go up, right? But that’s completely out of the cost per hire calculation. So cost per hire thinking encourages movements in the way that actually destroys company value. And the one area, which in that instance, tenure, doesn’t change cost per hire. But it’s the one that you should be following, ideally.
Andrew Marritt 17:26
In terms of going back to finance, we talk about cost per hire, but finance don’t talk about that. They talk about something called the shadow price, which is the return on the last pound spent. So we need to be focusing on those sort of mixed values and looking at the value there and optimising on that rather than cost per hire.
Matt Alder 17:50
There’s some implications on here. So you focus, you know, not every role creates as much value, right?
Andrew Marritt 17:56
Who’s the Boudreau’s A positions, those roles that are the most valuable. And that, you focus on validity. You want to try and make sure you get the best possible person ever. You know, I spent a bunch of years doing executive search. You know, that is, that’s effectively focusing on A positions, albeit it’s often just on, historically, it’s just on senior ones. But there are other roles that the variance between a good performer and a bad performer are really, really different. I mean, I would say sales possibly comes into there. Also, you need to know what your replacement cost is. And some positions have higher replacement costs than others. And when it’s about, I think in my example it was about 20, 21 percent of salary replacement cost, you were better off looking at hiring people with better tenure, predicted tenure, and when it was below then it was other factors that were best. And finally I think that what we always find is there’s never an obvious solution that’s the golden bullet. If you focus on tenure, you can only get to tenure by about 15%, the academics suggest, by selection-type levers. Most of it’s coming from managers and all of the other type of aspects. And if you focus on tenure you’ll screen out early career hiring. So there’ll be systematic problems by focusing, so you always get these balances and you have to kind of optimise on the right things, make sure you include everything that’s really valuable in there, and sort of try and pick these — there’s always trade-offs there.
Matt Alder 19:44
Just give us a quick insight into, so you say, for example, that £48 cost around structured interviews, how did you kind of land on that number?
Andrew Marritt 19:52
There are a bunch of different aspects in terms of ways of doing it. There’s a utility calculation that was created and is pretty much widely adapted. It was an economics model created by the IO psychology folks. And it looks at the value of the different aspects in there. The part where the difference is the most is the sort of standard, what’s called the standard deviation of performance, so SDy in sort of modelling terms. The difference between a really great person and an average person is enormous. If you get a high performer in the top 10% compared to an average, the financial difference is huge. So you look at what parts of the equation have the most variance, which parts are most different, and then you focus on trying to raise that. We like to think that the accuracy of selection is okay, but it’s not amazing. If somebody’s trying to sell you a test and saying this test will get you X, Y, Z, all the research suggests that we’re at 0.5 or something accuracy there, potentially a bit less. You have to realise that an awful lot of this is, there’s a lot of measurement error, but we can model this into there. The article goes through the worked example, but basically you want to be focused on performance. You want to be making sure you’re getting the best people, unless you’re at a level where you can just swap them out.
Matt Alder 21:35
Absolutely. So it’s kind of looking at the bit that actually is going to make the best difference in terms of working out who the best people are. That makes a lot of sense.
Andrew Marritt 21:47
There’s an interesting example, actually. I think it’s in one of my next couple of articles in terms of recruitment, and it’s about the replacement cost. So people often think of safe hires, right? They just want to get somebody who’s right, and especially HR, who are often risk-averse, tend to be focused on that. And one example I give is when I was working as head of European recruitment for the trading arm of JP Morgan 20 something years ago. We had a candidate and we couldn’t decide whether we were absolutely brilliant or rubbish. We just did not know. But what we knew is they weren’t a safe hire. And I remember the trading room, the trading desk head going, yeah, let’s extend them an offer. And I said, OK. And he said, well, Andrew, if they’re rubbish, we can get rid of them. But I really don’t want to take the risk of them going to one of my competitors and being brilliant. So the ability to get rid of people, either here in Switzerland, we have quite strong probation periods. And there really is an absolute decision. I remember in the UK, you have a three-month probation and nobody really takes any notice of it. Here in Switzerland, they really care about it. The ability to part ways if it’s not working is one of the most important things for raising up the quality of hires.
Matt Alder 23:13
Continuing on a similar theme around processes and people and judgement and all those sort of things, because it’s something that’s, you know, very much sort of being talked about at the moment. Cases where hiring managers are sort of overruling test scores, overruling recruitment processes, because they think they know better. You mentioned that selection tests weren’t necessarily as accurate as we’d hoped they would be. What’s going on in terms of that sort of subjective judgement versus the process and how is AI sort of shaping, shaking all of this up?
Andrew Marritt 23:49
I think that’s a really interesting one. So the headline result is that in almost all examples where having a process versus the expert, there’s an experiment there right back 70 years with Meehl, almost always having a process works. The problem with the expert is that they tend to be inconsistent. So there was a study looking at 136 experiments done. Formulas were about 10% more accurate, and the gap was widest when the interview impressions were among the inputs. We’ve got another piece of work four or five years ago where they look at the firms who overrode it. The ones who followed the results were about 15% better tenure. And the managers typically who overrode got worse results. And the candidates they passed over, because they could look at these days now with things like LinkedIn, you can sort of say, okay, we rejected these. What did they go on to do? The ones they went on to do were often hired later elsewhere by other firms and did better. So you can now see. So what we’re seeing is in most instances, there’s bias. When somebody overrides, it’s bias rather than information. There are a few cases where if you know something that’s really important and the algorithm doesn’t, in those small number of cases, you can be better. Right, but most people think that they know something that the algorithm doesn’t and they overweight it. So it’s about 90 percent of the time the override will produce less good results.
Matt Alder 25:43
And I suppose there’s also some confirmation bias there that that kind of one time that that person was right makes them think they were right all the time.
Andrew Marritt 25:54
Yeah, absolutely. We all remember certain aspects more than others. So we have a few things going on there. Algorithm aversion, right, so we all value that. But also the experts are more likely to discount advice than novices do, right. We talked about, my psychologist colleagues at UBS talked about the most dangerous thing being overconfidence, and this is this — novices tend to go and look at the data a bit more critically, right, where an expert will rely on their gut feel because they’ve done it so often before. So the best advice on here is really to just record where somebody’s done overriding. Because most of the time, and certainly from recruitment, a lot of the things that we do in recruitment processes or the selection things certainly don’t get analysed later on. It was one of my bugbears in people analytics is that most, we spend vast amounts of time, effort, and things in selection, and then that data never gets used later on in the process.
Andrew Marritt 27:06
So we also have a look at, you asked about AI. So it all depends on what the model was designed to do. So you can use AI to do exploratory work, right? To find people that the recruiter might have missed, right? So recruiter, I know what I was like when I was a recruiter. You hire four people from one firm into somebody else, and you suddenly think that everybody from that firm must — it’s those type of, we chase it and you know, if you’re doing search, you get your research. I used to get my research team to target certain firms that we knew were there and we’d miss a whole bunch of things before. And you can use AI to certainly break down those type of biases that we have. If we’re talking about AI in terms of large language models, they’re generative. They’re not replicable. The machine learning models that I spent so long of my time building and writing and cutting my teeth on, you give them the results and they always give you the same results. And AI doesn’t do that. So we’ve seen, there was a really interesting paper earlier this, I think it was earlier this year, where they looked at one of the AI selection interviews. So these online interviews that are there. Over 70,000 plus candidates, they looked at 12% more offers. And more than 18% more were employed at 30 days. So the algorithms were genuinely showing some really, really positive experimental results. However, and the candidate perspective, four out of five candidates preferred the AI approach than the human approach.
Matt Alder 29:00
That really chimes with a lot of conversations I’ve had on this podcast, actually, and doesn’t fit in with the broader narrative that’s sometimes out there.
Andrew Marritt 29:09
Yeah, but there’s a catch, and there’s almost always a catch. One automated stage halved completion rates. So you had a huge amount more dropout in the process. And who were dropping out? They were the strongest candidates and the women. So we have this thing where the AI improves the process, but it’s also having a problem in terms of dropout. The dropout was often to do with they felt there was more competition. And in terms of how do you fix this, the ones I identified, the tech companies, as you would imagine, tend to get these things right. There was an approach called Bar Raiser at Amazon. And the Google Hiring Committee. And both of those were sort of trying to take away the decision from necessarily the people who were the hiring manager and reduced the effect that one hiring manager can break it. And those things all show good work.
Matt Alder 30:14
The thing I take from this conversation is that there are no simple answers to any of this. It’s so much nuance, so much going on. And with that in mind, I’d just be really interested. So sort of taking together everything that we’ve spoken about, what would your advice be to the TA leaders that are listening in terms of what they should be doing? And if we have got the chance to sort of rebuild recruiting from the bottom up, from first principles, what’s hiring going to look like in the future?
Andrew Marritt 30:42
So I think the first thing I would do is stop reporting cost and start looking at value. And the way I look at it, I don’t think I’ve done it in the recruitment articles, but I’ve certainly done it in the book and I did it in other things, is that work with finance to get some agreed costs and returns and not just a single number because every single number has been wishful. Look at a range and there are analytical techniques to do that. Recruiters should be thinking about the value that they’re contributing to the organisation and therefore that moves the conversation to things that are adding value to the organisation. And a good selection process adds value to the organisation because the difference of performance between an average and a great performer is so good that it just outweighs everything else they’re doing. They need to focus their spend and their resources on the places where it makes the most difference. Again, it’s that variance of performance. You should allow overrides, but you should record them, right? And you can see where they’re doing. And also, when you’ve got tools and you’ve just been at the conferences and have seen, I’m sure, hundreds of different selection and recruitment tools, you need to be asking your vendor some good questions. And what I would be looking at if I was doing AI is what data were these things trained on, right? Because they’re often not a good balance. They’re not necessarily the people that you want to be hiring or in the countries that you’re operating in. Where did that data come from? So whose hiring history is it? Does it explore rather than just looking at selection? And how did they define fairness? I did an article earlier in the summer which picked up on, there is a result that basically says you can’t be fair. There’s three different definitions of fairness, and it’s mathematically impossible to focus on all three of them. So you have to work out what type of fairness you want.
Andrew Marritt 33:04
And from an employer brand perspective, which is where I spend so much of my time, you know, Spence talks about indices as well, which are things that are sort of hard numbers. So look at things that become hard to fake, like pay bands, median tenure, alumni destinations, really, really important, something that candidates can do. There was a fascinating paper that, there was in the last article by a guy Fogg and his colleagues which looked at employment advertising and they just moved one statement in the advert to the headline, to near the headline, and entry-level applicants rose 44 percent, right, and nothing else changed. They still stated the same stuff. So it’s not just about what you claim but how you claim it and whether it’s observable. And people don’t read stuff word for word and look at it. It’s skim read, so you’ve got to make sure there.
Andrew Marritt 34:05
So what would happen in three years if you kind of went back to first principles and looked at it? I think you’d look at recruitment as a flow problem rather than necessarily a sorting problem. And every ATS that I’ve worked with and every technology that I’ve seen seems to treat it as a sorting problem, right? If you’re hiring lots of people, it’s statistically easier to get it right than hiring one. And sort of, yeah, look at things like work samples, verified histories, those type of aspects. Take away the decision from the person responsible for hiring. One of the things we used to see within people analytics is that the performance reviews given by managers who hired the person were a lot higher than performance reviews. So if you look at the examples where the manager has been swapped out and can’t compare there, because for the manager, the performance review of their hire is a signal in terms of how good they are as recruiter, right, and how good they are as manager. So they bias it. They’re woefully inaccurate and they do it. So take the decisions away from necessarily, or complete decision away from just the hiring managers. And as I said, the talent acquisition folks, I think, need to talking to the HRO is, this is your issue there. They can handle one set of the levers but, you know, if the recruitment team hires somebody brilliant and the rest of it’s broken and the person leaves quite quickly because onboarding is rubbish or the new managers aren’t trained to look after new hires, then the recruiter gets blamed and it’s ultimately a full HR problem rather than a recruitment problem.
Matt Alder 36:08
Finally, I know that a lot of people listening will want to dive into some of the things that we’ve been talking about in greater detail. So tell everyone how to find your Substack again.
Andrew Marritt 36:17
It’s called Working Ideas. I think it’s andrewmarritt.substack.com. You can also go via my website. I have a personal website, as most academic folks, andrewmarritt.ch for Switzerland. In terms of, I’m always open to have conversations with folks. At the moment, I’m jumping into team meetings with recruitment to people analytics teams to talk through ideas and to challenge them. And I’m happy to do that if it works for both of us. Yeah, I think I’ve written about eight or so recruitment articles so far. There’s number four or five scheduled up till the end of the year. And hopefully some of the ones I write which aren’t about recruitment will be interesting for some of the talent acquisition team. I also speak to some technology companies as well about the processes and helping them think through the process. Often they’re a research team. I’ve got a strong background in ML. So I may be able to sit one foot in the recruitment, sort of why it works, and one foot in this is how the algorithms work. So yeah, I’m happy to have conversations as well with technology folks.
Matt Alder 37:38
Andrew, thank you very much for talking to me.
Andrew Marritt 37:40
It’s a pleasure, Matt.
Matt Alder 37:41
My thanks to Andrew. You can follow this podcast on Apple Podcasts, on Spotify, or wherever you listen to your podcasts. You can search all the past episodes at RecruitingFuture.com. On that site, you can also subscribe to our weekly newsletter, Recruiting Future Feast, and get the inside track on everything that’s coming up on the show. Thanks very much for listening. I’ll be back next time, and I hope you’ll join me.





