Holger Mueller – Why Has AI Not Arrived in HR?

Connect with Holger:

• 

Connect with Dwight:

• 
• dwight.brown@salary.com

Connect with David:

• 
• 
• david.turetsky@salary.com

Resources:

• Constellationr
• HR Data Labs Season 3 Episode 16
• HR Data Labs Season 3 Episode 14

• “Machines can do a lot of more things, and the big thing that fuels it [is that] a human needs time to get this done. Then it's the question, “How smart is he or she and how much time do they need and how committed are they to find out?” A very different equation to a machine because all it needs is for someone to program it, thinking of possible solutions . . . and then [all you have to do is] compute. So if there's enough compute cycles, ultimately the machine will keep trying, trying, trying and find it—something that a human will never get to.”
• “Work with vendors; see what’s possible; get educated; know what can be done; become competent; suggest what should be done. [Or] build something yourself, right? Every company has the data scientists, potentially AI people, who could make something.”

Announcer: 0:02

Here's an experiment for you. Take passionate experts in human resource technology. Invite cross industry experts from inside and outside HR. Mix in what's happening in people analytics today. Give them the technology to connect, hit record, pour their discussions into a beaker, mix thoroughly. And voila, you get the HR Data Labs podcast, where we explore the impact of data and analytics to your business. We may get passionate and even irreverent, that count on each episode challenging and enhancing your understanding of the way people data can be used to solve real world problems. Now, here's your host, David Turetsky.

David Turetsky: 0:46

Hello, and welcome to the HR Data Labs podcast. I'm your host, David Turetsky. Like always, we try and find the most fascinating people inside and outside the world of human resources to talk to you about what's going on in HR data, analytics and technology. Today, we have another repeat guest, our friend Holger Mueller from Constellation. Holger, how are you?

Holger Mueller: 1:07

I'm great. It's great to be here again, David.

David Turetsky: 1:10

Thank you for joining again. And like always, we have our best friend on the podcast co host Dwight Brown. Hey, Dwight.

Dwight Brown: 1:18

Hey, David. Hey, Holger great to have you here.

David Turetsky: 1:21

So Holger, why don't you give us a little bit of background about who you are and what you do at Constellation for those people who hadn't heard about you? I can't believe anybody who hasn't heard about Holger but give us a little bit of background on who you are and what you do at Constellation.

Holger Mueller: 1:35

You are too gracious, David. Great, great to be here. Right. So yeah, Holger Mueller, Constellation Research spent the last eight and a half years covering high wise, what I call enterprise acceleration. Companies have to move faster and become more agile. And they do that with my two sub research areas was about technology and people were talking about the people side here a lot. The technology side is about when can enterprise build software, what are the tools platforms best practice to do so. And obviously a large part of that is influenced by the new technologies which enable new things like AI, big data or the cloud and so on.

David Turetsky: 2:12

We'll provide the links for Holger inside the the bottom so you can find out more about our friend Holger. But One fun fact that we don't know that no one knows about Holger

Holger Mueller: 2:24

No, almost no one knows it's almost

David Turetsky: 2:27

necessity. No one knows that. We want the exclusive Holger

Holger Mueller: 2:32

no one knows Okay, the one that no one knows is that I was ranked in the top 10 in beach volleyball in Germany way down last century. So yeah, I used to play volleyball. It was very hard to play beach volleyball because there's not much of a coast and that the coast in northern Germany. There's always ugly and bad weather. Fairly selective for someone like me, like miles away from it. So you're thinking like, oh, I might play in the tournament. So you take Thursday or Friday off you drive like for eight hours after work you get there next morning on Thursday. Bad weather. Everything canceled. Wait, we need to get a break. Right. Right then you you you stay till Friday bad weather again. Sorry no qualification. No playtime for you, burn two days of holiday 1000 bucks so German likes to drive up there and gamble, hotel stays and so on. Yeah. Wow. I live in San Diego right now and it would be so nice would be so good to start my beach volleyball career out there. Right so these guys don't know how good they have it.

David Turetsky: 3:32

But you don't do you play beach volleyball now.

Holger Mueller: 3:35

Don't have the time and don't have the body anymore. Right? So too much. I played lots of indoor volleyball. 15 years or semi professional professional indoor volleyball takes a lot of toll so I'm doing mostly endurance stuff right now. We're just for another secret podcast. At all this jumping and fast things makes you big and bulky. And big and bulky. doesn't learn live doesn't live long compared the from the age of the Neanderthal as Homo sapiens was like a fraction of what we are today. So

David Turetsky: 4:06

Wow, that's fascinating.

Dwight Brown: 4:08

Now that's dedication. You're when you're willing to drive all that distance knowing very well that the game could get canceled.

Holger Mueller: 4:15

Well, it was the only way to do it.

David Turetsky: 4:26

So Holger, what we wanted to speak to you about today is the use or the lack of use of artificial intelligence in the world of human resources. And so, the first question I have for you is where have you actually seen AI being successfully used or being used with use cases that are, one could say adding value to the business?

Holger Mueller: 4:51

Yeah, well, obviously in the most competitive are part of HR, right even talent management talent acquisition. Recruiting plays a huge role when which candidate appear of which candidate to contact. And that really has paid off has been working. So there's no question about that. Another interesting area has been in the area also on talent management and performance management, how to help a manager which topics which techniques to use for performance management, one to one meetings and so on. Very interesting use case from that part. Last but not least, also an old industry problem tackled right, before they even talked about AI was flight risk. What kind of compensation do I need? And small combination? Not? So not so terribly complex AI? But what do I need to keep that person on staff? How do I pay a competitive salary, and then the distribution of the salary dollars and modeling that? Not necessarily AI, But getting closer to that and helping.

David Turetsky: 5:45

So the one that you started with, though, which I really want to pick on a little bit, you started with talking about recruiting. First, let's talk about recruiting, because I think a lot of us have experienced the recruiting part of AI, where they're filtering out 1000s and 1000s of resumes, or filtering through resumes, to try and find the needle in the haystack. And and I think one of the things that we're that a lot of people are trying to do is to figure out, what's the gaming strategy there? And is there a gaming strategy? Is the AI smart enough or mature enough to be able to find the best candidate? Or is it finding good enough candidates? Through all of that noise?

Holger Mueller: 6:30

Well, the question is always who the best candidate is, right? Because you only can hire one person need to have done a higher number two, number three, and don't know how the development where it's really interesting is if you think about like the big consulting companies who hire a whole class, right? And you know, all these people were ranked at the hiring moment, are they ranked at the end of onboarding? And after one year, right. So that's really interesting part. And interesting enough, many of the interesting recruiting AI pieces are coming from those white collar mass hires, like consulting companies, some tech companies, for the developers, and so on. So that's where you can do the comparison. Most cases, you don't have the data to do the comparison, if you hired number two, or number three, and number four, would they have been even better? Right? That's the big problem. But I think I strongly think that if you have the right data, which means you have to have the position of the skills needed for the job that you're posting, and you understand and those are faithful, right, right, everybody was LinkedIn out always Tripoli. colington, the network of liars, right? I mean, a true story here, right? I got endorsed by my dentist for enterprise software. He's nice guy. He's smart guy, but

David Turetsky: 7:34

You have good conversations with him?

Holger Mueller: 7:35

Yeah, no, I don't know if he puts me under for some form of dental work if I start fantasizing about enterprise software, but then even if I'm talking about it, can he judge it as an expert, right, so it kind of takes away from my enterprise software reputation if someone finds the next person who endorsed me as my dentist, right. So what a fake profile that is, right? So yeah, it's a challenge. Of course, with the quality of data and AI, of course, machines can only be as good as data they have.

David Turetsky: 7:59

And to that point, one of the problems that I have with AI is the training set, right? It's about what is it telling you, because it's using past history as the tool to train itself on where it should be hiring in the future? And we all know that there are lots of problems with the way recruitment has happened in the past.

Holger Mueller: 8:19

Of course, of course, no, no question about it. But it the question is, how much can the perfect and complete view of the past, help the manager make a better decision, right? If you hire for a future job, and you don't make that investment, you don't make that product? And so on? Then, of course, everybody said, Yeah, I'll be hired that person for that. And then that didn't happen, right. So on the projection side, I think equal AI and humans are bad. Because if you knew what the future holds, we will play the lottery, right? Yeah. But yeah, I would play it.

Dwight Brown: 8:49

And the model is only as good as the data that feeds it. So if you don't have that good data there, you know, if your organization's aren't diligent about collecting the data, or governing their data, or whatever it might be, your models only gonna be that good.

Holger Mueller: 9:05

Right. And then even let's say, if you have the perfect data, you have stuff, which is not in the data, because something like COVID happens, but my point is always on my statement, he has its block, I have to write like how COVID killed AI or paused the AI for at least a year from two years, right? Because even if we have the perfect AI for everything in HR, I know who should work there. I know how much money they should make off and I should pay them. I know how to motivate them. I know when they should work in workforce management. If you had everything perfectly done, all of a sudden, everything changed. And there's no pandemic dataset, right? Because even if we had the data set from 1918, or whatever, and the Spanish flu was out there, the economy was working differently, right? So we couldn't even say, hey, let's get the last flu data set and let's apply that right. So anybody would pass the emergency button to certain point to see let's check again if this perfectly working AI is working to perfectly during the pandemic.

David Turetsky: 9:57

I think you'd agree though Holger, that business cycles happen. And there are problems that happen, whether it's inflation, whether it's political issues, whether it's instability in certain global markets. And so the pandemic was a gigantic application of a big blip in the market. And so you have to be able to or not have to, but you should be able to train your AI to be able to deal with market cycles, and be able to, if not you, because obviously nobody can predict the pandemic, but predict the down downturns that actually happen every three, five, ten years.

Holger Mueller: 10:35

That's an absolute valid point, a very good point where it says you have to have your models for downtown, you have to protect the models for the bottom and for the top and so on when over the top is absolutely right, very good point. when the pandemic can seem like this very fair point. And no excuse actually, but I'm just thinking that the hit the pause for that chakra says our priorities too, because approving applications to go into System is constantly where the HR disciplines like super, super cautious, right? Anecdotally, I can't tell you, the vendor, and medium sized bank, this was a success story. After three years, medium sized bank, it was about the performance recommendations to be given to help a manager and out of the box, the AI from that vendor was only helping 50% of the performance reviews. So HR vetoed for getting it out. Right, which is like fundamentally flawed, because even if he only helped 50% That that is half of your company getting a better performance review, right. So so they wasted one and a half years to get in the low 80s When HR finally said, Oh, that is good enough, right? So do the math with the one half years, how many performance reviews would have been better? How many people might have left or feel better understood? Right? So there there's a wrong understanding of and unfortunately, they are professionals, not the biggest risk taker across the executive table. Right. And that manifests of all technology, but especially heard something. So mistake, non tangible, unexplainable, like AI.

David Turetsky: 11:54

Yeah, well. So one of the things I wanted to bring up in one of the use cases that I actually have a friend who does assessments on sales teams. And what he does is he measures the skills of the sales teams and tries to predict based on performance based on actual sales, who has the skills needed in this organization? And how do we model those? And how can we predict who's going to be a good salesperson and a bad. His name is Chris Kunze. And he does a really good job. And he uses AI, to be able to find those skills to be able to predict those skills, and to be able to train those other people inside the sales group. So it's not just the model itself, it's finding the skills based on an assessment, and then being able to utilize all that data to be able to help fill in the gaps and give people what they need to actually be successful. And then they measure again, to make sure it worked. So it's kind of completing the cycle.

Holger Mueller: 12:49

And great use case, right? companies spend inordinate amounts of money to train the salespeople and move the needle, your salespeople by and by only one or two percentage points is so much more revenue. And so, right, very interesting use case and good to see that that works.

David Turetsky: 13:02

Now, that's, that's AI in HR. But if you could give an analogy to where AI is being successful in the business side, what are the examples that we could learn from in HR to say, that's the kind of thing that will move the needle in HR? Other examples? You know, so can you give us examples on the other on the enterprise technology side, where, where AI is being used successfully?

Holger Mueller: 13:27

Yeah, I mean, you see in beginning So across across all the big tectonic automation plates of the company, right? So see it and finance and finding out which assets to sell or not to sell, how to valuate them, how to close the books, how to do financial planning and analysis, from any scenario planning side and so on. That's the popular things and finance. If you move on to classic manufacturing, the question there is like, how big should your stock levels be? Right? So again, make right yeah. As an example, right, the emergency button would have been hit there. Or too late or late even without much of the I wasn't the culprit, but they could have claimed the AI for it. Optimizing manufacturing floors, moving things manufacturing, understanding what other downtimes were to move things right. And the fascinating thing is, companies have moved from one or two manufacturing plants, say for the month to 1000s being analyzed by AI and proposing and finding the best one, right? So very interesting season CRM, right? When we're basically still lose a fight. The big thing is next best action. What do I need to do next for sales? Rep. What is the right thing, but there's some very interesting better models in there. And kind of like later to it right? I mean, the showcase for all of this, the the trigger for many of these technologies to take for granted today comes on a lot of the advertising space, right. So it's increasingly difficult situation to say which display ad to serve you right, and we typically ignore them and don't look at them right. But if you look at last year of the supposedly Tik Tok acquisition and forcing them to disclose the algorithms right The advertisement, there's much, much better, right? And interesting enough, like, I'm not a big for instance, Instagram user, I only use it to keep tabs on what my kids are doing. Right? No, no digital exhaust there, right. So if the Instagram ads are really good, so I'm buying more from Instagram than on any other Insert tab, advertising things which are coming to me when it's like, it feels like they've figured something out if it's AI or not, it really works well. Right. So don't don't forgetting like we talked about finance, manufacturing, sales, HR CRM, suppliers, right? Purchasing, right, similar, big part when to buy things where to put them in stock levels, estimating expenses expenditures, right. But again, all those models would have been wrong, figure out how much the sales are from travel, right. So

Announcer: 15:47

like what you hear so far, make sure you never miss a show by clicking subscribe. This podcast is made possible by Salary.com. Now back to the show.

David Turetsky: 15:57

Just so. So let's go to one of those examples. Let's take the manufacturing example. And actually, it's also the supply chain example. There's a component in there, which are people, right. So if if we have plans for a certain amount of revenue in a quarter, and we know how much supply we need, then in that case, we certainly know how much demand of HR we need. We know what the workforce planning model would look like. And so that can inform how many hours how many staff in what skills and what positions, and then also be able to tell us how much regular time and overtime we need to be able to complete? What's the profitability of that particular revenue that we want to get? Yes, great couscous. Yeah. So the question is, how does that not flow? And maybe, maybe this sounds stupid, but how does that flow into HR and HR then be able to get what's our demand on the on the H on the human capital? How many people do I need to hire in what skills and what positions and what locations?

Holger Mueller: 16:58

Well, flows kind of Vish over in the business plan, right of doing workforce planning, but usually the form of gelling is why don't we have enough forklift drivers? Right? Exactly. I don't really have enough welders left, right? He's a certified how could we run out of them? What happened? Right? So so it's really that's, that's part of the challenge of HR. Of course, it's not like a single function stovepipe, right? That manufacturing men are in charge of a shift. So Tyler was certain production unit has to make sure that he she has the capacity and the skills of people then then they open the opposition, right? Or they say, we need to have better training, we have to certify people, right. So this is the decentralized nature of it, which makes it harder to because disenfranchise the HR people from the real process, right? If you compare it to our people, to finance people, they grow up in their function, but the finance people deliver finance and their function, the HR people liaise with the business, I don't understand what's happening the business. I know what happened in the business five years ago, manufacturing, it certainly will help me in 2022, but it's very company very different 2022. So those those those learning lessons stale, and get older, that's a challenge for the job people to really serve the business units they're working with?

Dwight Brown: 18:07

Well, I would imagine that they're, you know, one of the challenges is that you almost need specialized use case scenarios, because staffing or forklift driver can be very different from staffing for an IT program or for example, you know, hospitals, hospitals have been doing this for years, where they're trying to guess what sort of patient care staff they needed at any given time based on the overall patient acuity within the hospital. And it's it's fluid all the time. Now, I would argue you can't necessarily take that model, and superimpose it on IT programmers or forklift drivers, for example. And so I would imagine that any, any change with the AI ends up being for incremental as you work on the use cases.

David Turetsky: 18:55

But But I would argue that what Holger was talking about would be that HR should have flex the muscle or should have muscle memory, and be able to remember that they should be a part of the conversations, or maybe I'm misinterpreting you, but they should have the skills built up in HR to know if we're going to have this kind of revenue and this kind of business plan and work it out with the managers, then we're going to need this number of people at this time. And here's how much we can pay them, not about how much we need to pay them. But here's how much we can pay them. And here's where we're going to go find them and solving for the where we're going to go find them becomes the problem for HR to get them there at that date trained ready to go on the line to make whatever the widgets are, right.

Holger Mueller: 19:43

It's a big difference. Absolutely. Why that's the big difference between finance and HR. Because the finance people yell at the business and say something's wrong there, right. Figure out what's wrong there, right, fix this, right. It's not comparable and so on. The HR people usually are the one way street and finding the forklift drivers to stay with the example. They get yelled at for that right but they don't have the turnaround part of understanding, why do we need the forklift drivers and so on. But the challenge of all, of course, to be fair to the enterprises, all the ERP vendors who are supposed to do this right, lack badly on the overall planning side, or there was a big movement of strategic enterprise management and strategic planning, it has not been delivered, but it's very much still art and enterprises and usually optimized in the stovepipe, and piling on a great tool from a company out of Redmond, Washington, right? Excel could be prioritized. All of these problems solved, right?

David Turetsky: 20:35

Well, people have tried. I mean, people have tried to take things that get done in Excel, and they, they try and make it a product. And then what they realize is there's so much customization and configuration necessary that people go, I'm just gonna do it over here. It costs less, it costs nothing. They say cost nothing. And then you realize with all the errors, people make an Excel? No, it doesn't cost nothing, necessarily.

Holger Mueller: 20:57

Well, it doesn't cost anything, because I'm doing it in my free time in my weekend. So so that's the the deal. Yeah, people don't do to be in control, but they want to do have to do right. So it's it's a catch 22 level of the pieces. But really, the planning needs to get better. So we have the records for machines to learn for AI to learn. This was the plan, what went havoc, right, if you can't look at 20,000, spreadsheets on the enterprise to figure out tell me nurses you need.

David Turetsky: 21:22

And that's the that's an interesting thing. Because if you could collect all of that data from all those spreadsheets, and if they actually had headers, and if they actually had a really good field structure and a row structure without multiple rows, and without lots of sums and other things, then the AI could read it. But because we make Excel spreadsheets, like we look at a piece of paper, based on the old forms that we used to create, you know, which have a sum, they look like invoices, they have a sum on them, they do this, they do that. So machines could read it, but they probably fail miserably at being able to tell what the hell people were thinking about when they were looking at it. Right.

Holger Mueller: 22:02

So but, David, you just unveiled what, if ever the Excel product manager was listen to this. So it should really be doing as an evil plan, you can still make it look like your form and your invoice. But in the background, the evil plan all the good planners, we can try everything and send us format to figure out what you've done there and connect the whole thing up.

David Turetsky: 22:23

Wouldn't that be just awesome? I mean, that is evil. But wouldn't it be awesome if there was a way for us to unstructure data, and still have the understanding of the structure of it so that we could read into it and be able to understand the concepts that were being portrayed there?

Holger Mueller: 22:39

Right? No, no, absolutely. Why it's a dream of any machine learning software automation guru, right. And I think there's progress on that, you see that when systems get integrated from A to B, that basically basically was still a white collar job, right? Somebody has to say, Oh, the postal code in the UK corresponds to the zip code in the US, right? But But by now machines have enough data to figure out I don't know what those nine digits are. Oh, that's my postal code. Let's see if there's a city field or place field. Let's see if I can find like a mic example. I know by heart 92130, my zip code and see if San Diego Wow, bingo, I found this combination. And now you can think about a trial and error combinations could that be on social security numbers or serial numbers or English use zip code postal codes without six digits and so on? Right? So the machines can do a lot of more things? And the big thing that fuels it, right? A human needs time to get this done? Right? And then it's a question How smart is he or she and how much time do they need now committed other to find out. a very different equation to a machine because the machine all needs someone programming, right? And thinking of possible solutions may be out of the excels? Why? Because if you will take all the excels out of the world, we would right away know what the postal code is a zip code, right? And then it just means cheap compute. Right? So if there is enough compute cycles, ultimately the machines will keep trying to I'm trying and find something where the human will never get to, right this is the the innovation which you see in healthcare for ideology, right? I mean, yes, a good ideology, initial immutable geologists will see something on the screen because he or she has seen this before. But it's completely unfair to go against the machine who sees millions and kind of compute 1000s per second, right? So so I come to believe that we're still leaving it to the human ingenuity to find out what's on your X ray. Alright, well, regulatory thing while we have the other day to do ties machine could be such a better job,

Dwight Brown: 24:27

the machines have become a lot less literal than what they have been. Yeah.

David Turetsky: 24:33

You know, and kind of taking on that, you know, think about or watch something like one of those programs on Netflix like that talk about the future, like Lost in Space, where they have a machine where someone could lay down or this is the Star Trek Next Generation, someone can lay down on a bed, and a scanner can come over them. Right. And the computer then scans the body for whatever is going on there. It says the good stuff and the bad stuff and then it tells somebody in plain language what's next actually happening with this person would they're human or an alien and they have all that data. I know that science fiction, but

Holger Mueller: 25:08

Hollywood can do it. 20 years later, we can do it.

David Turetsky: 25:10

exactly! but then watch the Jetsons and realize we don't have a car that flies,

Holger Mueller: 25:18

artists to figure out which 50% happens and 50% will not happen, right.

David Turetsky: 25:23

But what's fascinating, Holger is that we think about what the future is. And this is a good place to start the end of the of the episode of the future of AI for HR, we've always thought about AI helping HR by telling managers and practitioners and department leaders, what could they do? Within the boundaries of the law within the boundaries of the budgets that we have? What can you do that will help you that could be successful inside the culture of the company? Like you were mentioning before about turnover probability, which basically says, what we've seen, and we built that model at ADP, what we've seen leads us to believe these five people are at high risk of leaving and these 15 people are medium risk, and these 100 people are low risk. What could what should HR focus on as the future to get AI into the business to help the business succeed? What would the future be? What would the future if you could go into last in space next episode, and be and be the HR director? What would they be telling the HR director?

Holger Mueller: 26:33

Great, great, great question. Right. So but I would tell the HR director to look for the vendor use cases often coming from startups push the existing windows on the bigger data to help him and those enterprise or her to be more successful, right? And, and it's less mean, of course, you can say, I would like to have help here. I'd like to have help here. Why can't you do it here. But for the first one or two years, if you just look what the vendors think is working, and it's ready to do and could be done, that will educate any HR director starting at zero, or nothing very quickly, and to knowing what works, what does not work, and then make them more competent for after one or two years to say, I really need this. And I think you can do this, right? Because we have the data for that. And we could do this right? And then all of a sudden, suddenly things are happening, right? So the 2021 moment for AI and HR Celeste to happen, like, what's happened for space travel that we had a record number of people in space, right? If somebody today says next year, we'll have the record number in space, everyone's on die, I just have to figure out how many flights will originate when there's gonna be more people in 2022. Right. But the idea was to figure out, hey, in 2021, there's going to be a record number of spaces, and they're not highly trained astronauts, but could be people like you and me are ready to spend half a million or whatever it is, yeah. To be in space for a few minutes, right? And so so that moment will come to HR. And the question is only because there's not like a, Hey, I'm going to be in space moment, right? This is going to be really making a difference for a company. And if one company makes a difference compared to other ones, the other ones will notice, but if they can't do it fast enough, these business cycles are always moving faster. Hence my research enterprise acceleration, right? And you really can't miss the space moment of AI happening. And maybe it's happening 2022 As everything restarts to notice, right? So work with the vendors, see what's possible, get educated know what can be done, become competent, suggests what should be done, build some things yourself, right? Every company has the data scientists potentially AI people can do click to something there.

David Turetsky: 28:42

Holger, thank you so much for joining. You're awesome. We always learned so much when you're here. And we really appreciate your support.

Holger Mueller: 28:49

Well, thanks for having me. And you can like you can only be brilliant if you got brilliant questions, David. So thank you made me get the cold sweat on my back asking me these tough questions. So

David Turetsky: 29:00

yeah, especially at 10:30 at night, so I apologize, Holger.

Holger Mueller: 10 past 11: 29:05

00 Okay,

David Turetsky: 29:07

okay, well, you're not going to bed for a while anyways, then

Holger Mueller: 29:09

Easy time. No, I'm jet lagged, so don't worry about it.

David Turetsky: 29:13

Oh, Holger, have a wonderful holidays and enjoy yourself while you're there and really relaxed.

Holger Mueller: 29:18

It's up to you Happy Holidays to help you. And thanks for having me.

David Turetsky: 29:21

All right. Thank you, Dwight. Thank you.

Dwight Brown: 29:23

Yeah, thank you. Appreciate you being here. Holger, especially this late at night.

Holger Mueller: 29:28

Anytime. Thank you, Dwight.

David Turetsky: 29:30

And thank you for listening. And if you appreciate the episode, please click subscribe. And if you know somebody who might find it fascinating, please send it on to them. Thank you very much. Take care and stay safe.

Announcer: 29:41

That was the HR Data Labs podcast. If you liked the episode, please subscribe. And if you know anyone that might like to hear it, please send it their way. Thank you for joining us this week and stay tuned for our next episode. Stay safe