Better Signals, Better Hires: rebuilding trust in AI-era hiring

Hi, everybody. We're just waiting for a few more people to join us, what I wanted to mention is we've got a really good webinar coming up today. We've got some really awesome data. From Jody, who's joining me today from Greenhouse. Few housekeeping stuff, so we do have a Q&A section kind of planned for the end of this call, so if you look on your screens, you'll find a Q&A part in there where you can submit your questions, and we'll make sure that we have some time towards the end to go through that.

But yeah, we're just gonna wait a minute or so, and then what we'll do is we'll start off today. And yeah, speak to you guys in a second when we kick off.

Thanks, Nathan. And thank you, everyone, for joining us during this heat wave. Hopefully, everybody is keeping cool and keeping hydrated. As, temperatures are starting to… to soar across Europe this week.

It's too hot. We'll shout out that yes, I am wearing a tropical shirt, and yes, that's because it is hot today. This is not how I normally dress. But, I'm not genetically built for this heat, as it is pretty evident to everybody on this call today, so…

Similar, similar here. Luckily, I live beside the coast, so I've just heard a bunch of seagulls passing the window, and we have… The coastal breeze, which definitely helps. Manage the soaring temperatures.

Amazing. Give it another few seconds, then we'll start. People can obviously join and attend it throughout the call. We are also recording the call as well for anyone who runs a bit late, and wants to hear our little delightful intros at the beginning, so… Awesome. We'll get started. So, good morning, everybody. My name is Nathan. I lead up our Scede advisory, our talent advisory function here at Scede. For those of you that don't know Scede and are joining from seeing Greenhouse's Socials, Scede are an embedded talent consultancy, and we partner

With companies to help build and scale their talent functions for… Some of the fastest growing tech and product companies in the world. I'm gonna hand it over now to Jody to do a little bit of an introduction to herself and Greenhouse, and then we'll crack on with the fun stuff.

Perfect, thanks, Nathan. So my name's Jody Bartley. I lead our partnerships here at EMEA at Greenhouse. I'm based in Dublin, Ireland, and have been with Greenhouse coming up to 7 years. I've been working with Nathan and the Scede team now for about 18 months. As we partner together on the

the delivery and talent services of shared customers of Greenhouse and Scede. And for those of you who don't know, Greenhouse is the hiring platform that helps organizations make better hiring decisions with structured workflows, trusted data. And a candidate experience that strengthens employer brands. And in 2026, we're really helping teams adopt AI in hiring, with confidence, and trust.

So, our goal is to really make hiring great for everybody involved in the process, and hopefully that comes across today. Okay, so our title for our webinar today is Better Signals, Better Hires, and it sounds simple, right? But what we're about to discuss, you'll see that getting there is anything but. we… let's start with the landscape. Like, over the past few years, from kind of 2022 to the end of last year, as we can all attest, there's been massive world change, a massive industry change across

TA and H… and HR. And what we've really seen here at Greenhouse is three forces colliding at the same time. So, global economic shifts are creating wild swings in hiring demand. Teams that were scaling fast are now frozen hiring, and vice versa.

And at the same time, in tandem with this, we're seeing an explosion in HR tech complexity. More and more new products are launching onto the market. There's hundreds of tools promising to solve the problems, that TA teams are experiencing today, but yet the more tools we have, sometimes the problem gets worse. It's really an arms race. And then, on top of all this, ChatGPT arrived on the scene in 2022 and disrupted everything.

We're all probably using AI in our work. Life day-to-day, at the moment, as well as our personal lives, but it's not always for the better. So this is just creating a perfect storm for talent teams out there in the market.

What have you seen, Nathan, over at Scede?

Essentially what you've mentioned today, I think one of the things that I noticed, so one of the things that we do in advisory is we do a process audit where clients is an option for them to do, and one of the things that kind of concerns me about the landscape that we're seeing at the moment I know we'll go through other parts of it throughout the webinar today. It is the variety and volume of tools being used. So, people are utilising multiple different sourcing tools at the same time, not knowing which one is actually working well for them, which ones may have forms of bias in it. There are automation tools for screening, note-taking.

I've seen situations with companies recently where even the process of GDPR is becoming more complicated because of AI tooling. Like, are we allowed to actually send video recordings that our AI note-taker is taking? Because that opens up problems. A lot of people would think. Yes, you can, because the candidate's data is there, it's them speaking. But from a legal standpoint, you also have to protect the information privacy of the interviewer or the recruiter that's on there as well, so it doesn't actually become a black and white situation, it's a grey situation. So, one of the sort of bigger concerns to me is the variety of tools out there, and especially in this sort of AI world that we live in now, there's massive amounts of new products coming out.

without really looking as well at the legislation that's starting to come through from the EU, from the ICO, legislations across the world, right? But that's sort of the main kind of concern for AI for me, is also having too many tools and not one platform that automates all of the things that are going on.

Definitely. And I suppose another thing that we get asked about time and time again from our customers, and just generally from the market is, like, transparency around your AI policy, and about the AI that is powering those tools and products. Because, as you said, people are innovating and shipping products to try and keep up with the demand, but it's not always thoughtful, and sometimes can actually create more problems down the line as well. So… being able to stand behind the AI that you have in a product as well, I think is a massive part of creating the trust, and building the trust in the process for the people that interact with those tools.

Yeah, couldn't agree more.

So, before we move on to our next piece. We want to do a little poll for everyone, love a bit of audience participation. So, if you… would like to click, how many applications have you seen for a single role this year? Is it in the…

300 and under bucket, 500 and under, or 750 plus. We'd love to just get, a quick survey from the group to see what kind of experiences you're having at the moment.

Amazing. Should we go through to the next bit, maybe, while we wait for people to respond, and then we can go back through the responses during the end of, like, the next slide?

Okay, well, we see, we've got it so far. 36% at 750 plus. That is huge. That's over a third of the audience. And that… unfortunately, those were the kind of numbers that I was hoping to see to back up the data that we're going to present here today. Okay, 31% of 500. Huge. And then… okay, so…

And then the rest of the group still with lower numbers. Yeah, this is what we're hearing more and more here at Greenhouse, that TA teams are dealing with volumes of anywhere 700, 1,000 plus per role. So… We can attest that the job market moves in pendulum swings, as we've all witnessed over the past few years. Like, this is the first time in Greenhouse's 14 years that we can see both sides of the hiring process are really unhappy. At the moment, nobody's winning.

Candidates have turned to AI. Like, your numbers are proving. with the spray-and-pray approach, mass applications for jobs, and then tailoring their CVs to match job descriptions.

So then, the recruitment teams that receive these applications could receive 1,000 applications that mostly look the same. So what do they do? They respond with more AI to rank and filter the flood. Applicants never get looked at, and then they apply for more jobs. Rinse and repeat. This is what we have termed, internally here at Greenhouse, the AI Doom Loop.

And… Over the past year, we took a survey of over 2,000, job seekers. in the US, the UK and Ireland, and published our workforce and hiring report, and I can post that here afterwards if people are interested, on responses, and it came up that more and more Job seekers we're using.

AI across their whole application process. And in turn, then, volume per recruiter has grown approximately over 400% since 2022. So we're seeing the average of about 150 applications per role reaching to 746 per role. And thanks again for your participation, because your responses are totally backing up this data today.

And we're also seeing that per job opening, it's up about 250%. So what do companies do? They're using AI, as we said, to screen those applications, which means the candidates can't get through, they use more AI, the loop keeps spinning faster, more noise, and less signal.

Everyone's working harder, and just getting worse outcomes at the end of the day. So, how are you solving for that, Nathan, or what are you… what are you seeing?

Yeah, I'll go through what we see first, and I'll go through how we've kind of solved… well, one of the many approaches, what I find is the most productive way to solve it, so… what we're seeing right now is similar to sort of the pendulum swing analogy, I've always, throughout my 16 years in TA, referred to recruitment as a roller coaster, right? We see ups and down troughs, and that goes both in terms of market velocity and not, but we also see it in trends. Trends typically tend to repeat within a throughout my experience, sort of a two-year cycle between that. And historically, what we would see is it would shift from a candidate-driven market to a client-driven market, and I'm sure everyone in the chat today knows what that means, but a candidate-driven market is where candidates are harder to find. We found this more during COVID, where recruitment teams exploded in size, companies were scaling at massive rates, and a lot of the hires, and we've seen this in our data here at Scede, were actually sourced

hires compared to application hires, because we were having to actually proactively go out and source these amazing sort of candidates. What we've seen, sort of somewhat semi-recently.

towards the dreaded meta layoffs of 2022, where the market made a massive shift, is we've gone to the other side of the rollercoaster, now we're in a trough, where it's now actually a client-driven market. And what we're seeing is astronomical numbers of… Applications come through, and an old, sort of.

story to go through is back in my, sort of, early days in recruitment, I remember having a QA role, and I went to post the QA role, and someone tapped me on the shoulder and went, don't you ever post a QA role? You're gonna get, like, 250 applications, you're never going to speak to all of them. I've seen a client of ours do that recently, within the last year, and rather than the 250, which we thought was a large number 10 plus years ago, we're now seeing up to 1,000 applications for a QA person come through. And I think one of the biggest shifts is what we're seeing now is the rollercoaster's cadence has changed, right? It doesn't seem to be following, sort of, two-year peaks and troughs that I'd witnessed throughout the majority of my career. We've been in this trough, or this client-driven market now for nearly four years, and I think one of the main things that I've noticed

in terms of optics in the market is a lot of companies are very hesitant to use the word scaling these days. Like, scaling kind of implies what we witnessed. the post-COVID, pre-meta layoff times with loads of hires and loads of hires, and that kind of predated the mass rounds of redundancies we saw in businesses towards the end of 22 and 23. So companies are still scaling, but the market For whatever reason, be it macroeconomic, be it…

the state of the world going on at the moment, where we're not seeing that uptick back to a candidate-driven market as quick as we would normally see. From my logic, we're about 4 years into this sort of trough that we're seeing now. And especially with the state of, like, global affairs going on at the moment, I don't… I don't necessarily see that changing anytime soon. And I don't mean that doom and gloomy, I don't mean to say, I don't think we're going to go back to a candidate-driven market. We will. The roller coaster has to go up at some point before it goes back down again. But I think what we need to do as an industry is kind of prepare for that, right, and reassess how we actually

Go… bring candidates through the candidate journey, and how we actually look at our hiring process as a whole.

Yeah, yeah.

Amazing. So I've got something to share with people today. Some people may have heard of it. It's… I have a horrible tendency to take things from other industries and approach them back to recruitment. We will probably do a webinar on this example at some point in the future, but we actually assess our recruiters on something called level of effort, rather than, like, number of hires and stuff like that, and that's a project management measurement. And MVA is, kind of a spin on MVP, which is Minimal Viable Product. So, MVA stands for Minimal Viable Assessment, and essentially what it is, is the leanest version of something we can use that still delivers maximum value, right? So, in terms of sort of, like, an operational definition of it.

for… in relation to recruitment, it's the absolute smallest number of stages, touch points and questions, essentially required to make an incredibly accurate, high-quality hiring decision. So it's essentially like a… an architectural blueprint of a scalable talent system. So, one of the main things about it is kind of moving from talent as a craft Which I've got some bits to say on in a moment, but talent as a craft is more… talent as a craft is more defined as where everyone trusts their gut.

asks different questions and uses their individual, sort of, like, secret sauce, if we want to call it that. And then it's actually moving to something as talent as a science. So, this is where, essentially, the process is standardized, it is data-driven. And I guess more importantly, especially with the volumes we've kind of been discussing through the last slides and last bit of the conversations, it's… it's entirely repeatable. That's the main thing, right? So if we go to the next slide, I can tell you a little bit more about it.

So, what… Going back to what I mentioned before, when we have a client-driven market. And we have the ability to Essentially pick who we want to hire.

the older guard form of recruitment really stands true in for that, and… I'm go back on what I say in a moment. One of the first things is every recruiter has their sort of secret source, right? Me, personally, I'm a huge fan of that. I think we don't hire recruiters to be robots, especially in this time in the world where we're actually using robots. But… In regards to that, I think there has to be a way to make it a repeatable session, so recruiters can still have their sort of secret sauce, their personality, and bring through their true self to work, if you want to refer to it as that. But in a scaling environment with application numbers the way that they are.

if everything is different, we're not being fair to people, right? And that's one of the most important things for me. These sort of processes as well do have a horrible tendency to burn out once volume hits, so if we go back to the scaling phase, we may have had high number of roles, but we had minimal numbers of candidates. We times that based on the data. You've shown, like, 140-something applications and 22, up to over 700 now.

is kind of untenable, it's unscalable with that sort of framework in mind. And with this as well, you tend to find hiring managers go rogue. It becomes those, I want an extra process. or I'm gonna validate something a previous interviewer has already validated, and then we're gonna argue during the debrief. I'm pretty sure most people in chat have had that during a debrief call, where your…

CMO is saying they got them for marketing performance when someone in the marketing team has already assessed them on that and put them as a plus for a right. And then what we actually moved to is more of a scalable process. So, an MVA is the minimal viable number of stages that are necessary to actually make a hire, and these are structured and agreed typically during the kickoff call. I'm assuming, again, most of us recruiters do a kickoff call, but I think the level of kickoff call needs to be a lot higher from that, and you can use AI. This is one of the things about automating tasks and where AI is useful.

But it's basically a standardised MVA for every role. It doesn't mean you need to do it for every single role when a role comes through. If you have high volume, you can repeat it multiple times for the same MVA. It doesn't need to be entirely bespoke for every single version of the same role. But it means that one, the process is streamlined, so that we don't have to have increased time to hire, which, again, is what we're seeing in the market at the moment. But it also means, from a candidate experience point of view, everyone's being assessed by the same framework, right? So there is no going rogue from hiring managers, there is no…

elongated stages, there is no unfair bias towards people, and I think as we're going through this phase at the moment of higher applications coming through, I think bias is intrinsically kind of creeps into that in some way, be it via robots, or be it via hiring managers, or people involved in the interview panel, right? But one of the most important parts of MVA as well is leading by metrics. So, I am a humongous fan of, like, time to hire and numbers. I think anyone who's worked with Scede or has worked with me in the past knows I'm a massive quote-unquote, data head. Time to hire is a retrospective analytic. It doesn't…

I mean, you can actually make improvements as you go through. So one of the main aspects of MVA, and we will release a white paper on this for the next few days, so people can actually build out a framework of their own, but it's actually looking at timing stages while the stages are going on. It's looking through, NPS scores throughout the process, like, repeated NPS, rather than just one when a candidate leaves. You need to get feedback from people But it's also actually getting reviews from the hiring managers as well. So, what's your thoughts on that before I kind of go through a little bit more about the process step, Jody?

Right.

Greenhouse has tools for that sort of, like, repeatable frameworks, doesn't it?

Well, I think… in so many ways, Greenhouse was built for minimal viable assessment. You know, if there are people on the call who've used Greenhouse previously, or if you're not familiar with the system. It was built on the foundation of

And goal to create structured hiring for everybody, and that kind of embodies this whole idea of minimum viable assessment, so… it really speaks to talent as a science, and making clearer, data-driven decisions, but I think, the fact that when you create a job and an interview plan in Greenhouse.

It's set up on the framework of scorecards that you add standard attributes to, you bring the kickoff process into the job setup as well, and bring in the hiring manager, and have it all in one central place. I think… the whole… overarching mantra here as well, to your point, Nathan, about fairness is, like, to create an even playing field for everybody, so they're scored on the same attributes for the same role, or the…

M… And then, again, it's all about feedback, and, like, bringing that feedback into one place, making it transparent, making it accessible for the team that are interacting with that process. And then, even when a candidate needs to go one step further, if they're unsuccessful and they come back to the hiring team and ask for feedback and ask to see, there… you can actually download a candidate packet and share their data under GDPR, especially here in Europe. So I think, you know, we're completely…

We are the system that was built on this type of thinking as well.

I think you raised a good point there, especially about feedback, and one of the things you tend to find with what I quote-unquote refer to as a fragile, recruitment process is it's very hard to actually relay feedback back to a candidate, right? If there isn't a structure to the interviews, and you're then having to go through… I know this is a very dramatic example, but a 12-stage process or something like.

Wow. Gold. Yeah.

a 12-stage process, but amalgamating that all into actual tangible feedback for a candidate, then to actually go and improve their performance in interviews for future opportunities becomes incredibly difficult, right? It's… I can't think of a 12-stage process where at least the maturity of it wouldn't actually be opinion-based rather than, like, factual assessment that we can today, right? And that's what we want to kind of move away from, especially in terms of, like, speeding up. the last thing to go through is, like, process debt, right? We're… we're seeing

Joanna, jumped the gun. At the moment, so what we're seeing is a number of, like, 4 hours of hiring manager time, which is a high ceiling. We're seeing time to offer drastically increase, right? What we've seen as a market is time to hire has actually got worse over the last 4 years compared to where it was.

That's right.

But, like, COVID was a very, very fast time to hire. I've… hires I made in 2022, where I did it in 5 days, and I know for a fact I would not be able to repeat that now, with the way that most companies are running their TA functions, right? And especially even though there are high numbers of applications, we're seeing much higher than I personally can remember ever seeing over the last 16 years. the number of the best talent's off the market in 10 days is still true. The best talent won't be around, they will be snapped up by companies applying something like… like an MVA, right? As I said before, we'll make sure that we make something up and send it to people on the MVA, because we've seen it work. I've applied it to every customer that I've worked with throughout my time here at Scede.

I think it's something that needs to definitely go out to the wider public, because we've seen very tangible results from it, so I'll make sure I write something up and we'll share it on LinkedIn or wherever afterwards. But I'll hand back over to Jody to go through the next bit, which is my favourite part of this.

But, well, just before we move on to that, yeah, I suppose one thing that I didn't respond to there is just the whole idea of… you know, streamlining the process, and for scale and repeatability, you know, Nathan, like… more complex isn't always better, to your point, and I think…

over… in a previous life here at Greenhouse, I… would consult with TA teams on the best ways to To design and mirror their process in our system, and again, trying to cut that down to just the stages that were absolutely essential, so then it was easy to bring that out to other departments across the business, or even just for future roles.

And having those attributes that made the most sense. for one hiring manager, so again, he could just go to his TA team. And then, in turn, we would see reduction in time to hire, better talent, better retention, better employee experience. So there's, like, investing in this type of thought at the start, and maybe changing processes can seem a lot, but it will

Like, really yield great results and benefits for your teams in the future as well.

Excellent.

Yeah. But I… I suppose… We're talking as well here today about, you know, a lot of volume. and then a lot of negative experience for candidates and TA teams alike. And I think…

It isn't just a volume problem, but this is also creating a trust crisis, and just a distrust in the system, for the people that are interacting with these applications and companies. What we've seen over the past few months is Over 90% of recruiters have caught candidates misrepresenting themselves with the support of AI.

Then, on the flip side of that. Nearly 70%, 69% of candidates have encountered fake job postings, like, for data harvesting. And then… This leads to over 40%, nearly half.

Just saying that their trust in hiring is really decreasing. And I have no doubt that these numbers are even a little bit higher since we put this content together. And just to… pick out, because we are… we're talking here today to quite a large UKI audience.

we're seeing 42% ghosting rate for candidates in the UK, and that's the worst of any market surveyed, by us over the past 6 months, which is just shocking, really. Do these numbers resonate with you, Nathan, and what you're hearing from your customers?

Yeah, there's a few concerning things in it, obviously, right? But… we, we've… We've definitely experienced increases across the board. I think I'll start on trust, as the first one. I think anybody who goes on LinkedIn or anybody goes on Reddit can see that there's been a marginal decrease in trust. I will always say that it is obviously usually the vocal minority that tends to be the most outspoken on things, so obviously take what you read with a pinch of salt, but…

if you equate it back to raw numbers, again, data guy here, we're seeing higher numbers of applications, which kind of intrinsically means we're going to see a higher number of people that have been rejected. It's just… pure numbers, right? And it tends to be people that are dissatisfied with processes that are quite vocal about it, and so that kind of naturally relates in a higher number of trust in hiring has decreased, but I think it also relates back to a shift that we've also seen of

moving away from transparency and becoming more transactional. I think we got to a very good point a number of years ago as an industry about being very transparent with people, and I think, especially with the rise of AI tools, companies are quite resistant to being transparent now. They don't necessarily want candidates to know all the tools they're using in TA. I know this won't be an issue moving forwards because of regs coming in, people will have to disclose what it is, but I think there's a number of factors at play into… into trust. In regards to the ghosting.

It's always been a problem with our industry, but… I think… I don't think I know. I know that this is going to become a problem for industry moving forward, so me and Jody actually spoke about this a couple of days ago, as we were going through a little dry run-through from this, but… what we tend to see with regulations, and regulations come through, is it usually comes from a small sample, or a small country, or a small state, or a small county setting certain rules, and then typically a larger organisation will watch and see how the results of that have played out before they start making it a larger rule, right? We see it nearly with every form of regulation that's possible.

And the one that always comes back to me from ghosting was one we all kind of found out about towards the end of last year, which was the Ontario anti-ghosting law that came through. as much as I'm not gonna sit here and say everything needs to be regulated. I do think for the job that we do as recruiters, it's quite an important job. It doesn't mean always that companies realise that, that we work for, or whatever, but

We are the people that are responsible for getting people who may have been made redundant, or people who have lost their jobs. We give them the opportunity to find those jobs to provide for their families, or their pets, or whatever it is, their personal circumstances, right? So, a varying amount of regulation is important for that. And the one that's coming through from Ontario it's quite a big one. It's a very shift towards being a candidate law more so than a company law, which is not what we normally see, right? And for those of you on the call that don't know, Ontario passed a law semi-recently where

For every case of ghosting that's over 45 days, so if an employer doesn't notify a candidate if they're successful or unsuccessful 45 days after the last act in that process of your final interview, or whatever it is, it's somewhat of an instantaneous 100,000 Canadian dollar fine.

Amazing.

If they see if the Ontario government sees repeated offenders of this, it can actually go up to a $750,000 Canadian dollar fine. Part of that as well. is one of the first laws put in place. Obviously, we've got the EU AI Act that's been delayed, and we've got the ICO here in the UK delaying certain things at the moment, but Ontario was also one of the first, sort of governing bodies that have made it, so if you use AI at any point in your recruitment process, it could be anywhere, it could even be just meeting notes internally if it impacts the recruitment process in any way.

it has to be disclosed on the job advert. And again, if it's not, it's an even bigger fine. So… what we need to do as an industry is realise one thing, bigger places are looking at these sort of small regulations. We're already seeing stuff come through the EU AI Act, and any company that works in the EU, even if you're UK-based, you have to adhere from it. But we're also seeing with the ICO changes here in the UK, is… what I would like to see is more people being aware of what these legislations are that are coming through. Like, we've had a pretty good ride.

Challenge.

That's the challenge, right? And we've had a really good ride of the industry, if you go through GDPR. recruitment's kind of a grey area, like, we weren't singled out in that whole legislation, but what we're noticing now with the EU AI Act and the ICO is recruitment is now being explicitly named in things, right? And part of this is based on what you were sharing earlier. Application numbers are higher, more people are being affected, so the governments are listening towards these things now.

how to solve it. I think we need a broader conversation. We have challenges. I'm not saying it's easy to go back to every single candidate. If I was saying, oh, I did that over the last 16 years, every single one, I'd be lying. We've all done it. But I think we need to have a bit of a longer conversation about how we solve these sort of intrinsic problems.

Definitely. But I think it's… it's a real positive, you know, that Our… our countries and our continent. are starting to bring in legislation, albeit delayed, which is slightly disappointing, but I think for… for teams

That are on the call today. to Nathan's point, awareness… Of what these regulations are going to look like, starting to enable your own team, starting to be proactive and act as if

that legislation is already in, getting ready for it, like, is really what we should all be trying to do, like, create a new norm, within… within our working life. So, yeah, I love the anecdote about Canada. Somebody who lived there for nearly 6 years, they're… they're just such a… thoughtful and polite race of people, so it seems very fitting.

We lived there for a year as well, and I fell in love, so I can… I could see that happening before it did. We'll go over to the next slide now, which is…

Tight on time.

We're tight on time. Washington study. So, a lot… I kind of briefly mentioned a minute ago about the EU AI Act and the ICO and stuff like that, but the Washington study, was actually the study that kind of pre… cursed all of these laws kind of changing, right? So it was a study that was conducted by the University of Washington. It's a fascinating read, and we're going to go too far in depth for a few all today, but I'd highly recommend people Google it and see what impact it led to things. But the Washington study was a test that was run across 528 recruiters, and they incorporated an AI screening tool for these recruiters to utilise.

ran this in three separate tests with the same recruiters, and they actually incorporated bias into the AI. And one of the things that we, as an industry, aren't aware of is actually how our AI tools are built. Like, that's the, I guess, initial risk point number one, right, is we're not AI engineers, we can do vibe coding, awesome. It'd be very hard to build, like, an actual candidate screening tool via vibe coding. You need to have a very good understanding of how AI operates, right? But AI is programmed by one person or a group of people. It's intrinsic that every single person has a formal bias. That's why a lot of companies, and I agree with it. conduct, like, unconscious bias training. It's one of the things I've done, and I learned a lot about myself from it. But the test itself, so they ran three separate tests, so with that, they used an unbiased AI, which resulted in the recruiters being unbiased when they were reviewing the candidates post the AI screening.

They ran a separate test where there was moderate bias, it was about 50% or 60% bias. At this point in time, the vast majority of recruiters noticed it. I believe it was around 75-80% of recruiters were able to actually notice bias, so when they put their notes in after the AI screening tool, they were like, I don't agree with this, this doesn't seem right, I don't think the AI tool is actually reading this correctly, I don't believe it's… fairly associating certain traits with certain people. But the really alarming fact was the final test. So the final test, they…

put in a severe bias to it, so the AI was fairly judging… unfairly judging the large number of the candidates on small little things, or creating things up completely entirely about the candidates it was screening. And the very glaring fact from that was 90% of the recruiters, so 90% of 528 recruiters, actually backed up the AI.

We're good at what we do, we're good humans. But if you have something… it's the golden rule, if something projects something with enough confidence, people tend to believe it. But the alarming thing is most of the recruiters post the assignment were asked about this, and most of them were absolutely adamantly actually overruled the biases. So, it's a glaring thing of being very responsible for the tools that you do, and there's a list of questions we went through in a previous webinar, and I'm, again, happy to share it afterwards, but there's certain questions I believe all of us within TA need to actually discuss with the vendors when we discuss the vendors for these tools.

To make sure that we're able to actually fairly utilise these tools from the system, but… This study is a great one because it directly correlated into a lot of the law changes we're seeing at the moment. What's your thoughts on it, Jody?

Yeah, well, I suppose, to your point, we've spent A lot of time, and we really invested heavily in the whole… area of removing unconscious bias from the hiring process, and…

I think this just really proves that no matter how good we think an AI tool is, or how… much more efficient it can make a process. We cannot remove

The human decision maker, and… and the human review, and… Even when they say they've programmed it to replicate, like, these different levels of bias. it still needs to be questioned. And it's interesting that you said that the group

It's the herd mentality that human nature can't get away from, right? It's still agreed with it, but I think it's always good to question, even when you're working with your own Claude or so on. If it gives you results, go back and question it. Like, QA, test your own agents. Because you'd be surprised at the results you can harvest, as long as you still remain in control. But I also feel, as people use more tools, we're going to see a bit of human bias come back in, because of bringing

making sure that we can be very assertive as the decision maker at the end of process. So, will we ever get to a place where unconscious bias is totally removed? I don't know, you know.

I'm gonna spoil it there and say no, we won't. that haven't gone for unconscious bias training don't even know if they have it. I wasn't aware, I've lived all across the world. Prior to my recruitment years. I thought I was a very open-minded individual, and I went through unconscious bias, and I came away thinking I was a horrible human being. So, I think it's a… mandate, there's free courses online for people who haven't done it before in the past. You can get trainers that will come to your company and do it. But, it's a humongous thing that definitely benefited me towards the latter point in my career. We'll move over to the next slide now, because I know we're out of time, and I know we have a few little questions. So, I'll hand over to you for your closing slides, Jody.

Sure. So, I suppose the example from the Washington study was I won't use the extreme, because it's a real-life example, but it's a very strong example of that AI kind of loop as well. But let's move to, you know, how we can start to break these cycles and positive steps that us and our teams can take. In the face of all of these different tools.

what we are seeing here at Greenhouse is we feel that, like. The confusion and kind of the chaos that can be At the top of… A candidate pipeline. can be really prevented by strong signals. So, the answer isn't necessarily more AI screening, but it's better signals at the top of the funnel.

So… For example, we have, a feature in Greenhouse called Dream Job. It's basically an intent-based feature. Candidates can log into the candidate portal, My Greenhouse, and they can actually mark jobs that they really want as their dream job, one per month. And a candidate who marks that role as their dream job, and then it's flagged as a signal to the TA teams in charge of that, and the recruiter in charge of that.

They're 4 times more likely to get hired for that role. Also, applying for a role in the first 24 to 48 hours, that shows… signals high intent, again. you know, you can do that via LinkedIn, we also can do it via our candidate portal in My Greenhouse, you can have

notifications for companies and roles that you're interested in. As soon as they go live, you'll get an alert. Apply for that, and you're going to be one of the first in the door, so that's really going to help your cause. Then when we look at, like, misrepresentation and fraud and that trust crisis that we touched off of earlier, we… Are looking at unearthing real talent through…

Features like identity verification, confirming work history, and other… surfacing other signals that confirm that a candidate is who they really say they are directly within the ATS. And then, of course, good old reliable referrals still outperform cold applicants by 10x. If you are…

A trusted employee of a company, and you can vouch for somebody that you know, or have previously worked with, and bring them in. as a referral, you know that they really want to work with you, and also. Your employer can see that this is somebody that you are willing to stand behind.

That outperforms 10x each time. So, intent beats volume, every time. And then, just moving on to wrap it up…

we really feel that hiring's being reinvented. I think, you know, everything myself and Nathan have spoke about here today, over the past 4 to 5 years, the industry's just been turned on its head. So… We have a new mission and vision. We really want to make hiring work for everybody, not just the employer, but the recruiters, the candidates, and the hiring managers. This is a holistic process that touches off of everybody and impacts everyone, and we want to reimagine what hiring can be in the AI era. We're building AI into every key workflow in Greenhouse, so sometimes you don't even notice it. It's just peppered there as you go along, with major productivity gains, giving you suggestions, helping you write job descriptions, helping you with summaries, and so on.

But never removing the human decision-making or auditability. As I mentioned earlier, we're really trying to help teams fight fraud at the top of the funnel. And focus on the real candidates that are applying to join their company.

And then, a new one for us is Voice AI. When you have a thousand plus. Applications for a job are 750 plus, as a third of you here today said you've experienced.

It's not going to be viable for a recruiter to actually speak to every single one of those. By bringing in voice AI screening, they can actually have a 15-30 minute conversation That will screen them, as opposed to just rejecting, and gives everybody a fair shot. So, as I said, our mission is simple, we just want to make this work for everybody.

I think, you know, that's why it's so good to work with Nathan and the Scede team. Scede sits exactly at the intersectionality of advisory and action that TA teams need today,

So, Nathan, anything to take us home?

I think… Closing statements, I guess, yeah, before we get into, like, a quick little bit of a Q&A. I know we're overrunning a little bit, but I'm personally more than happy to stay to answer some of the questions. Is the landscape recruitment's changing?

And it's… becoming very different from a lot of what used to, and I think we have to make sure, as an industry we adapt, but we also have to make sure that we're being fair. like, I know it's a very boring statement, but I'm pretty confident everyone on this call believes what I'm saying, and they want to be fair themselves as well, but there is being intrinsically fair, but then there's also taking control of the processes and procedures we use as a business to ensure that we are fair as well, right? Like, we can be fair as people to people.

But I think, as TA, we need to be responsible for making sure that our processes are fair, and I think one of the things Greenhouse does well, and one of the things we love partnering with you guys, and having conversations with you guys, and all the interactions me and yourself have had over the last 18 months really shows the Greenhouse as much as it is a tool for TA. I think the precedent on a lot of things that you do is actually for the candidates as well, right? And that, to me, is incredibly important. Like, when things get busy, we as an industry tend to…

curl up in our shells a little bit, and we tend to be more like, okay, cool, I'm really busy, I've got… hundreds and hundreds of headcount, I've got thousands and thousands of candidates to go through, I've got no time to do this, and that is true. That is very much true, we've seen it over the last four years. But I think we need to make sure that we keep to the root of what we do as a job, which is we… we help people that can sometimes be in very unfortunate situations, and we enable them to come and join our awesome companies that we recruit for. We pay for them to do that, because we're not volunteering, we have staff, we pay them, and this then leads to better lives for people. And I think if we keep that in the forefront of what we do, a lot of what me and you have discussed today

will come naturally to a lot of TA teams. It's very hokey-pokey answer, but I'm convinced after 16 years that it's… that it's true. So we've got a bit of Q&A time here, we've got a few questions come through. Obviously, I know we're overrunning a little bit, so if people need to drop off, I know the call's being recorded anyway, so if you've asked a question and yeah, can't stay around for it, we always share the recording afterwards, so, you'll be able to see the answers to your questions. But we'll kind of go through. So we've got one question here, so… How often should teams review and update an MVA once it's in place?

it's gonna sound like a very short answer, as much as necessary. If the role changes fundamentally is usually when a review of an MVA needs to be done, I'll try to keep the answer quite short, but if you know… Let's use a software engineer as an example, a back-end engineer uses C++. If you know you're doing repeats of that exact same role, the MVA doesn't need to be reviewed unless there's a problem with the MVA.

So if you're noticing that time to hire is not decreasing, if you're noticing that candidate sentiment's still bad once you've employed the MVA, typically that means there needs to be a review of the MVA. If the MVA is successful, and it is working well, and you're noticing your time to hire drop, you're noticing your NPS scores rise, you're noticing that you are getting those candidates that are off the market, usually in 10 days, and you're getting them through your process quickly, usually means the MVA is working as intended, as according to plan. That being said, if the role slightly changes, so if you do an additional kickoff, let's say you've filled your headcount for the role, and then the hiring manager comes back to you in Australia and says, I've got another 2 headcount, I need to go back up to market.

I wouldn't go in automatically to review the MVA if it was successful the last time, but during the kickoff call, when you re-kick off that role, you'll tend to find if there's changes based on the original requisition approved, right? If there are changes, then you need to re-look at the MVA. If it is just a carbon copy of what you've done before, and you've got a very solid MVA, proceed with it as planned. But I think one of the most important things about MVA, and we'll cover it when I write something up, and we'll make sure we send it to everyone, is it's not looking past at data, it's looking at data as it's going through, right? And what I mean by that is don't look at data reactively. I've done this, I filled this role, what's my time to hire? Oh, no, it was bad.

we need to work on that in the next sprint. For an MVA to be successful, you need to look at, like, your conversion metrics, your timing stages, your pass-through rates, as it is happening, and then that will actually give you a solid frame point of knowing if your MVA is successful. So. it's kind of a simple way to explain it, but if you're being proactive with your data throughout the process, you'll know if it's working. If it's… if you're being reactive at the end, you're going to be…

two steps forward, one step forward, two steps back, essentially. But yeah, that's the answer to the question.

And I'm happy to take the next one. I think, how you balance a repeatable framework with the fact that some roles genuinely need different assessment methods, like, similar to Nathan's answer there, but this might seem very basic. templates. Like, you can create departmental-level templates, you can create role-specific templates, and then they can be Scaled and customised as necessary, as different needs come up, or as a role evolves.

Like, we have… you can create a massive repository of templates that different people can have access to within our system, and… we have seen some companies, some huge global enterprise companies, get it down to 5 or 6 job templates for their whole company, and then they're taken by different TA teams, and they're customised. as necessary. So… I suppose it's getting to the point where

You are going to either have standard attributes on a company level, or on a departmental level, or on a specific role level, and then just creating those… those templates. And different kind of frameworks around roles. Another thing to mention, the data piece, Nathan, that we have is

you can go into Greenhouse as you're creating a new role, or even copying over one of those templates, and you can look at recently filled roles that are similar, so you can get a summary of the data around that, like time to hire, what the job description looked like. You know, the offer that went out, and you can use that data to inform you as you're opening up this new role as well. So, it seems… it seems very simple in this… in this AI-driven world, but I think it can't be as basic as that.

I'm a huge fan of bringing things back to basics. I know we're running, so we'll pick one more answer. If we… one more question, I'll answer one more. If there's any that are left over, we'll make sure that we follow up in a proposed email afterwards. We want to make sure we answer everything. The next question, do you think anti-ghosting rules like Ontario as a locally is spread to other markets? Short answer, yes. Yes, yes I do.

my previous background, so prior to being at Scede and Vitality before it, I worked in investment banking for about 10 years, 9 years or so. A lot of my recruitment there was actually building for regulatory teams, so, like, MIFID too, if anyone's worked in banking and FRTB. From my experience, bearing in mind I'm not a regulation expert, I recruited for it, I read through the regs so I could recruit for it correctly, But prior experience tells me when something… some… some small region incorporates a law, and it turns out to be successful, especially when also recipient countries or countries like us are

with more left-leaning governments, they tend to listen more to these type of regulations. If you're a country that is on the other side of the spectrum, they tend to be more about abolishing regulation, but you tend to then find regulations that are mirrored and tweaked. I would also say as well, when those regulations are mirrored and tweaked, they're usually imposed more stringently as well, so they will look at the original regulation and go, cool, 100,000 Canadian dollars seems bad, how about we make it 200,000 euros? We want to be… Because we want to… it's where an ego comes in from, like, a governmental point of view, is we want to be seen as being tougher than last people who did it, because we're… we're this…

regulatory body, or whoever it is, right? Yes, so from rule of thumb, we've seen it, the EU AI Act came in before, or was drafted before, what our ICO draft was drafted. They talk, they listen, there are regulatory bodies, there is the UN, there is the EU. they typically look at these small things and then will mirror them. And I think as AI is becoming more and more and more, and Application numbers are getting more and more and more.

we're not kind of like a shadow industry anymore that's not being noticed by people. We're being specifically named now. Like, for the first time I've ever seen in 16 years, we're specifically named in these legislations and regulations that are coming forward, and I do think we'll see more, And I think… it will be difficult for an industry like ours to react to that, because we're not used to being regulated, we're not ACAS with HR, we're not…

FCA or finance, like, we've never had a regulatory body, but I do think it's something we're not going to have a choice to kind of look into moving forward. And I think the quicker that we adapt to that, and maybe as companies start adhering to these things before they become laws, like, we make a commitment as a company, we're not going to be ghosting candidates over 45 days, you can run reports in your ATS for it, and, like, you can find out what the ghosting rate is from your own ATS. I think the easier those regulations will be when they finally come in.

Definitely.

Cool. So, we do have another question, but we'll hold off for now, because obviously we've run over for a little bit. We're about 10 minutes over today. I really want to thank everyone for attending. I want to thank everyone for putting their questions in. I want to thank everyone for doing the poll as well. I'm a big fan of polls, so I love hearing the voice of the people as we do these kind of things. I want to do a humongous, humongous thank you to Jody for joining us today.

Thank you, Nathan.

absolute pleasure to work with, and I've been really impressed, and I have been fascinated by the data that you've provided today when you first sent over the draft to me a few weeks ago. I think it's amazing what you guys do, and I think we as an industry can learn a lot by actually working with the software providers we actually use, right? So thank you to everyone for attending today. We'll do a follow-up email afterwards, we'll make sure we send out the recording to those who need it, and I will work on making a little white paper for MVAs, because we had a few questions about it, and we'll make sure we share that wherever We can share it so people can start building these kind of frameworks themselves.

Amazing.

Enjoy the sunshine! Bye!