Scaling HR Technology w/Terry Baker

[00:00:04] Welcome to the Recruitment Flex with Surgeon, Shelley. I'm Serge. And I'm Shelley and we talk all things recruitment starting right now. Balls you and welcome to the Recruitment Flex. Shelley, we've got the big man not only in

[00:00:24] stature but also in influence in this industry so I'm really excited to bring on this guest. We've seen him a lot recently. Yes, I think he must be feeling like we're his shadows or something

[00:00:36] all the way around. I am very pleased to welcome back to the show is Terry Baker who's the CEO of Daxra. Terry, welcome back. Thank you. It's great to be here and yes we have been bumping into each

[00:00:48] other a lot recently. More specifically you were one of the speakers and what an inspirational presentation you did at TA Tech so we had the privilege of introducing you on stage and getting to

[00:01:03] sit and listen to your presentation. It was quite inspiring. So what we're I'd like to start is for those in the audience that maybe didn't have an opportunity to hear you the first time you were on

[00:01:14] the show. Can you just share a little bit about your career journey and what led you to your current role as CEO of Daxra? Sure, I happy to do it. You probably know last time I came

[00:01:27] in the show I was with Pandalogic and Pandalogic went through a phenomenal growth stage. Really had some significant growth during the pandemic which is unique. We pivoted from real match to Pandalogic

[00:01:44] about five years ago and we weren't the first to do programmatic job advertising but we had a unique value proposition in a unique angle because when we were real match we had all this job performance

[00:01:58] we powered hundreds of job boards on a private label based system. We had a programmatic ad that ran across all the sites so we had performance data on almost every conceivable job type and

[00:02:11] that was a huge advantage when we went into the game of programmatic job advertising because we could do predictive analytics on what a job type should have in terms of its campaign budget and what

[00:02:26] they could expect as an output and we would go into competition with appcast or critics in the like who were using rules based systems in a rules based system means somebody has to write the rules.

[00:02:41] So you got a campaign manager, they look at all these job types and they write some rules and they run a campaign and then they'd like oh gee we got the price point wrong let's change

[00:02:51] it running again and see what happens. Well with predictive analytics you get it right the first time and that saves people money it provides confidence for buyers and we had a 90% accuracy rate on

[00:03:05] our predictive analytics by virtue of the data and we like to say AI is only as good as the data that you're using if you have great data, accurate data you get great outcomes if you have

[00:03:17] inaccurate data like a rules based system you're not going to get great outcomes until you find tune it and it takes time and effort to do that so automation was great prediction analytics was great

[00:03:31] and so that led us to in exit and we were private equity adventure capital backed we had a 10 year run which for some venture interest is a long time. We had their backing when we pivoted shifted to our business models and within five years of

[00:03:53] shift we wrote the significant results huge growth and that enabled us to get a great valuation in the market. Before you go a little bit deeper of a couple of questions on that as well because

[00:04:07] you mentioned programmatic and how they came in play and obviously there was a recent acquisition of appcast acquiring bared as a CEO in the industry and having a background in programmatic and competing against appcast what was your take on that acquisition? First thing that came into my

[00:04:24] mind is margins because agencies operate on the standard margin and they publish their margins. And 15% is the typical margin. If you're a programmatic software company and I don't want to get into the secrets of margin analysis but if you can predict in advance what the outcome will be

[00:04:46] what the optimal campaign budget would be you can build in your margin and you don't have to disclose it. If you're a software vendor, if you're an agency you're disclosing your margin

[00:04:57] you're telling them what the margin is that they're applying and it shifted the game a little bit to a race to the bottom. And programmatic has become far less lucrative by virtue of the agency involvement.

[00:05:12] And if you talk to appcast and work critics, the same thing. They merged with an agency and now they have fixed margins on their software product. If you're a software vendor and

[00:05:26] you can predict in advance and you can optimize that campaign you deserve to generate a little additional margin. Absolutely and it was interesting because we had to add them to CEO of recruit eggs on

[00:05:37] stage and that was a question that I asked him. And I think one of the things that I was always curious is how are you going to alleviate concerns from other agencies, right? And you're now potentially

[00:05:48] leveraging your data. Does you work to a lot of agencies at Pandologic? Was that ever a consideration? I know it's revisionists history but did you ever think of acquiring an agency and doing the same thing? We actually competed when indeed made an acquisition of a programmatic vendor.

[00:06:06] We pitched indeed. We discussed valuations with indeed and we decided we didn't want to be affiliated with one site or one agency. So you go down one of two paths. You remain independent

[00:06:23] which you have to be as a programmatic vendor. You can't have one site operating as a buyer and investor and another site not because there's a conflict of interest there. And the same is

[00:06:37] true for agencies. So I'm sure that Appcast sought for it a little bit by virtue of some agencies, shape or recruitment probably a good example was utilizing them and decided we don't want to be on

[00:06:51] this same path as Bayer, right? We want to compete. Good enough after you bit Terry, you brought us up to the point in time where Pandologic had grown quite significantly as you said

[00:07:05] and got a great valuation. So what led you to Daxtra? How did you find your way there? Did they just come looking for you? Well, Stratum Capital came looking for me and it was a year ago

[00:07:17] in June that Stratum invested in Daxtra. Yeah, Stratum is a very unique private equity firm. Okay. Oh, my work with a lot of different private equity firms and they do something unique. They buy founder based SaaS companies. When you're a founder and you've built a unique

[00:07:39] business and you get it to a point, it's your baby. It's valuable and it's not something you just want to hand to somebody and say, okay, good luck and take care of my baby and off you go. So they

[00:07:51] come in and they evaluate the business and they come up with five things they're going to do and five things they won't do. And they get agreement before you get a purchase, they got agreement with the founders. Here's the five strategies we're going to implement.

[00:08:08] Here's the five things we're not going to change that will give your company, your culture, your employees, some assurance that we're not going to come in and change the whole nature of the game.

[00:08:19] And so when I evaluated what the five point plan was with the extra it was spot on and I think it helps the founders in giving up a little bit of the latitude to a new majority owner

[00:08:34] and get some onboard with the growth plan. So I looked at that five point plan, I looked at the your diligence that they did and my raised my hand and said, this is a company that can scale

[00:08:47] significant. Talk a little bit about what Daxra does. Sure, Daxra's about around 22 years, the founders were linguistics. They did a DARPA contract 22 years ago on a list of linguistics analysis and so they converted that into a parser that parses resumes and parses

[00:09:09] job descriptions, pulls out skills. They now have that in 50 different languages, which gives us the ability to support really large global companies. So for example, Google has licensed our parser, Amazon has licensed Apple has licensed meta Ernst and Young some very large companies because

[00:09:31] they do global hiring at scale and they need a parser that can take a resume and pull out 200 variables out of that resume and use it in a semantics search environment. Most recruiters today

[00:09:45] still operate on Boolean. Then construct a Boolean search and what is Boolean? It's just keyword analysis. So if I'm looking for a CNet developer and I use in my Boolean search, CNet and the resume

[00:10:00] only says C++ is not a match but we all know C++ is a CNet development language, right? It should correct. And so semantics search and linguistics enables us to use machine learning, build a taxonomy, and to extract skill based information in enables recruiters to do skill based hiring.

[00:10:23] So we do parsing parsings a little bit of a commodity now in the janitor of AI world. Yeah, give you an example that we had a staffing agency that came to us and said, we want to go into the solar engineering market. How is your taxonomy for solar

[00:10:40] Boo. We don't really have. So we went and scraped about 10,000 jobs off the internet for solar engineers dumped them into eight different generative AI models from bar delama to unayment and we looked at the taxonomy that came out of that in within two weeks built

[00:10:59] of linguistics, semantics search based upon a federated approach to generative AI and built our own large language model for solar engineering. To do it publicly with a chat GPT has risk. Yeah, you're putting that out of the public marketplace so to doing it with a

[00:11:18] parser company that can provide semantic capabilities enables us to build them the search interface for recruiters. So ultimately what we do is recruit our automation. We make search easy for recruiters. We enhance it with semantic search and we grade score and rank every candidate that comes into

[00:11:39] the platform. And so we enable this search to run across job boards. If you have your own account on indeed or any other site, you can credential us to run that search on your map.

[00:11:54] So we'll take the job description. We'll take the resume and we'll say here's the optimal search criteria to run on indeed. Let's bring those candidates back into posit them into the

[00:12:06] ATS. So data integrity is part of what we do to a lot of recruiters look at a lot of resumes and a lot of those resumes never go into the ATS. In fact 40% or so across the industry never get

[00:12:23] into the ATS. They get stuck in an email inbox, they get stuck in running the search on indeed or monster wherever and they don't take the time and effort if it's not a highly qualified resume

[00:12:36] doesn't go in the ATS. And the whole goal is to build and augment the data in the ATS so that your recruiting becomes a more effective over time, right? And you can update resumes based upon

[00:12:50] some of these change of careers to teach job category, big setter. And so you don't have to go out every single time and run a program at a CAD, can they run a program at a search

[00:13:00] or cost different job boards which is expensive? Right. You're really expensive. Yeah. So Terry I think you really like transactions because recently you're a year in a required pivot CX. What's the strategic play there for you? Yeah the strategic play with pivot CX

[00:13:20] is to move down the funnel and provide more of the hiring capability in an automated fashion for recruiters. So we got up to the point where we were grading scoring and ranking candidates and then

[00:13:32] said, that's up to you. Go get them. And there's lots of means to engage and I acquired that Pandologic Waiten Wendy which was contextual bots or engaging candidates. And candidates know now when a bot is interfacing, they can tell the difference between human engagement and bonding

[00:13:55] engagement. And if you can grade scoring rank the best candidates, why send them to a bot? So what pivot CX had that was really instrumental and fit well with our product is the humanization approach

[00:14:08] to quickly engaging a highest quality candidate and getting them to a hiring manager. And doing it through SMS or WhatsApp, doing it on a one-to-many approach. So recruiters could ensure that candidates were engaging quickly in getting through the application process at a higher conversion rate.

[00:14:27] So those are the two heat criteria we deliver. The average SMS that goes to a high quality candidate is responded to within 30 minutes. How many people did you add to the team as far as

[00:14:40] pivot the transaction? Pivot was not in the staffing marketplace, there in the kind of enterprise marketplace and they sold through a lot of partners. So it was a fairly small team but we're taking

[00:14:52] them out to the staffing marketplace where Daxstra has quite a few clients. Daxstra has about 2500 clients. Wow, that's impressive. Part of that is because we do OEM, Isem cells are product, eightfold cells. We do support those customers direct. I'm putting on my TA leader, my tele-naquisition

[00:15:15] leader hat and I'm thinking like you name some really big companies. So if I'm not this Fortune 100 that you named I'm not Google, I'm not Amazon, I'm not Apple. I'm trying to understand like where does

[00:15:29] Daxstra fit into my talent acquisition infrastructure? Is it sitting on top of an Isem's or a workday or a success factors or is it underneath? Is it beside? It's baked in and it's on top.

[00:15:45] So a parsing is fundamental to every ATS. Every ATS has got a license parsing because that's where you derive the data from. And again, the more accurate the data, the better outcomes are going to have.

[00:15:56] So one of the things that we're known for is really accurate parsing and multiple languages. And then what we do on top of that is provide a search interface for recruiters that sits on top of the

[00:16:09] ATS. It connects with different job boards. We support over 100 different job boards that we can run searches through. And we collect all that data and ensure that it gets into the ATS. And when a candidate comes into the ATS it's graded scored and right. So that the worker

[00:16:27] or knows immediately based on that skill set capability, what the quality is for that candidate relative to a particular job description. So we work with 100 different ATS. So it's not that you have to be a top 10 global recruiting company. You can be on greenhouse and used AXDRA, right?

[00:16:50] You can be on top, right? And use AXDRA. Okay. So when you talk about parsing and I just think about the T8 leaders and my circles, when we talk about parsing, they immediately think how

[00:17:01] when somebody uploads their resume onto say workday. And it's supposed to extract the information and put it into their little form. It's never right. That's not what you're referring to though.

[00:17:13] Because when you say it's on top of the ATS, is it running out in front before you get to that register to apply process where it parses your information and puts it all in the wrong field?

[00:17:25] Yeah. If a candidate comes to a corporate oversight and deposits their resume and starts the application process, that comes through DaxDRA. And we grade the score and rank that candidate relative to the job description and what was in their resume. Now parsing's probably not the best

[00:17:42] term to use for talking to a team. I think skills based hiring is more appropriate. Yeah. Okay. Because if we can collect out of a resume, all of the skills that are relevant to the job description and identify them and put length of time and performance relative

[00:18:00] those skills, then we can produce a highly qualified ranked candidate. Got it. You get the hard skills out of the resume and the soft skills is the harder part. And that's where PIVITCX comes into play as well because it's the conversational approach

[00:18:19] with a human that can either run a campaign or do a direct outreach to a candidate. We can collect all that information and start doing skill analysis on the conversations, which gives you more of the soft skill qualification than the hard skill. Brilliant. Brilliant.

[00:18:38] Thank you. Thank you. Terry, let's take on the team of T.A. leaders. One of the questions that we get a lot is I have no clue where to start off with AI. I know I need to leverage AI. I don't know who

[00:18:52] the partners who I should look at, how should I evaluate? What's your thoughts? What advice would you give to a T.A. leader when you're looking at implementing integrating AI into your text stack

[00:19:05] order process? There's lots of kinds of AI now. And Genre the AI is captured the marketplace, because Genre can produce content on your behalf. Can write a great job description. I don't

[00:19:18] know if you saw that SIA this year there was a LinkedIn presentation and they showed a recruiter using chat GPT to try to qualify candidate in real time on the phone and the candidate using chat

[00:19:32] GPT to respond. The recruiters trying to extract the qualification of the candidate and the candidates searching with chat GPT to find the right words to show them their qualified. And of course,

[00:19:45] when you have AI engaging with AI, the content you get is all artificial. It's not real. That was a good exploitation of how chat GPT can lead you down a path that isn't good. And frankly,

[00:20:02] we did performance reviews of Pandologic. And I started reading the performance reviews from my direct reports and they had all had chat GPT answers to all of the questions that were really, I'm like, okay, this is worthless throughout this out. I'm just going to have a conversation

[00:20:20] with people about what they accomplished this year and not enabled them to provide me answers. I would chat GPT. There's lots of different kinds of AI. I think the most valuable kind that

[00:20:33] operates on very large data sets. We parse a couple billion resumes a year and when you parse that many resumes, you run it across keywords and you build a taxonomy that becomes very efficient in

[00:20:47] accurate. But it takes huge amounts of data in order to do that. So machine learning large language models built into your taxonomy gives you advantages in the marketplace. So we like to talk about

[00:21:01] a federated approach to Gen AI, the solar engineering example that I gave you and building our taxonomy so that it's always present, always most accurate, and it's learning from the market close because things do change, right? When you see companies in this space that are launching

[00:21:19] AI tools and you being very knowledgeable about what it takes to build a business here and they're building over a chat GPT API or whatever the case is for the TL leader that doesn't know.

[00:21:33] How should they evaluate that company and make sure that, hey, this is legit? That's a good question and it's a hard question for TA leaders because the problem is TA leaders they have to accept

[00:21:45] the liability now that's coming out with the ATT and all these laws that are ensuring there's no bias being introduced into their platform. And so the question I would ask is what data does your

[00:21:58] platform operate on? And can you present me with an audit that shows you've taken that data and pasted through a third-party auditor that's identified whether they're not, there's any bites. We did this years ago back, I think it was HR tech 2020 and there was a, there were probably

[00:22:22] over 150 companies that announced some construct of AI. And there was a trade association of CA charos that went to HR tech and they're like oh my gosh this is a lot of AI, let's engage

[00:22:38] the Morton school and have them run an audit. So they sent letters to everybody that did press releases on AI at HR tech. Guess how many out of the 150 press releases people actually said I'll go

[00:22:51] through an audit? None. Eight out of 150. Handelogic went through that audit and we actually got badged for two proficient capabilities of AI. One was prediction and the other was location analysis because location in the programmatic world can be the biggest introduction of bias.

[00:23:21] Right, if you're trying to hire somebody for a warehouse in Brooklyn and you're only advertising within a 12-mile radius of that warehouse, you're hitting a demographic group that is unique. And by definition it's going to introduce bias. So you have to expose that.

[00:23:39] Right? But I think there's a lot of companies now that are doing audits that are demonstrating where is the potential bias in the platform and the data on which they operate. Love it. Great advice. I want to go back a little bit. You talked about the

[00:23:57] panologic exit, you had a couple of exits before that and that's what your presentation was about. And we have a fair amount of entrepreneurs in the TA tech or HR tech space that listen

[00:24:08] to this and I'm curious what advice would you give to that? If they want to set up their business to scale and eventually exit or potentially go to IPO if they get that big, what's the advice

[00:24:20] do you give to those folks being an entrepreneur? I think if you're going to be an entrepreneur, part of the motivation for an exit is the raising of cash through either venture or private equity companies because private equity and venture back companies have a timeline for investment. They

[00:24:39] have a return on their dollars that they have to justify and that motivates the team to work on that particular mall and just like with stratum they've got a five-year plan. They've got a model

[00:24:51] for building and scaling the business. They have metrics that they want to see being met and achieved over time. Those metrics drive the outcome on the exit and I talked about that and in the presentation the rule of 40, right? The lifetime value of a customer to cost a

[00:25:10] acquisition of a customer. Those metrics show whether the company has achieved a maturity point where it can continue to scale and produce break results. When you're starting those metrics are really hard because you've got to make an investment and build a model before you can scale it.

[00:25:31] Part of the conflict that comes sometimes with venture and private equity cash infusions is they just want to see the scale. They don't want to see the better before you get bigger. When you try to scale a business that doesn't have all the constructs in place, it's really

[00:25:52] painful and it doesn't work. You've got to build a strategic model where you can scale on something that is demonstrable has product market fit and has been proven in the marketplace. So I'm going to come back to the practitioner side of things that are using a tool like

[00:26:12] Daxtra and now with the added capability of pivot CX which is super cool. Something I really appreciated learning about them too was that there is a human touch, not everything is solved by

[00:26:28] are there any key metrics that you think TA leaders should be paying attention to as in now it's time to start looking at solutions like this. What would be some of maybe the trigger points

[00:26:40] that a TA leader would go, okay here's the problem. Yeah. But how would they be quantifying the problem and know that it's the right time? I think there's three metrics every vendor should answer

[00:26:52] to a talent acquisition manager and oftentimes they're hard to get to because you are not controlling the entire hiring process. But time to hire cost to hire and quality of hire or the three metrics that any talent acquisition provider ought to be able to answer.

[00:27:13] And quality of hire is the hardest, right? Yeah. Because what are you measuring that against? Are you measuring a year time frame down the road on their performance management? How long did

[00:27:23] the employee stay at the company, right? How well did they perform? And can you get a quality of hire measurement before they're even onboarded? What's the projection on the quality of this hire? And that comes down to skills based hiring. Again, hard skills and soft skills.

[00:27:44] And whether or not you can identify those. And those are hard things to produce, right? You can go to somebody's LinkedIn profile and you can look at their track record in the industry.

[00:27:55] And I've got a decent track record in HR tech but I wouldn't be the right CEO for all types of talent acquisition providers, right? I have a unique experience with entrepreneurial and

[00:28:09] building companies to scale. And if you put me in a $200 million company plus and ask me to run it, I would be frustrated. Yes. I wouldn't do very well probably. But building a company that can

[00:28:23] scale, that's something I'm quite familiar. So quality of hire is the hard one cost to hire is a combination of doing the right things to find the right candidates. Having come from the program

[00:28:36] at a rural and seeing now the search base to equivalent, I'm a far bigger advocate of search than I am of a programmatic because most job board companies have gone to a four-strain situation. So they

[00:28:53] have a database of anybody that's looking for a job, make it immediately send you resumes based upon criteria of your search. So why not just grab those resumes and qualify them as opposed to hope?

[00:29:07] They see an ad you're going to run and if you're not on first page, maybe they're not going to see you or they're going to easy apply in applied to 20 companies at the same time and the

[00:29:19] recruiter calls them up. They don't even know which companies they've applied to. So that's a messy process now on the advertising side that makes search just far more effective. And finding the right candidate, if you know the criteria under which you need to search. Very interesting perspective and

[00:29:38] you're seeing that, right? Like you're seeing even indeed launching smart sourcing, especially in a market that you're getting thousands of applicants on certain jobs as well, which is almost impossible to go through. So you've accomplished a lot in your career and now

[00:29:53] you're in a new journey with DAX, I'm just curious what's next for DAX, DAX, DAX, DAX, is going to scale from the staffing marketplace to the enterprise. And we've got some complaints in the enterprise market, right? We talked about the Google's in the Amazon's,

[00:30:10] but 70 to 80% of our revenue comes from the staffing market. And if you can leverage what we do uniquely in the staffing market to not just the largest hiring companies on the planet, but to the rest of the world, then there's significant opportunity. And the talent acquisition

[00:30:30] world is four to five times the term that it is in the staffing. So that's the growth opportunity for us. And for us that means doing some unique things that we don't do on staffing. For example, if you're

[00:30:45] a talent acquisition manager and you've got a new job description, your current employees, probably have never even seen that job description. Might be on the corporate curricide, but they got their head down, they're working, and they're trying to perform. And they're not every

[00:31:01] day going to the corporate curricide saying is their job for me in terms of a career path. So internal mobility is a far more effective approach than going external and trying to hire somebody

[00:31:13] and determining the quality of your hire. And then hope they come in and they fit the culture and they perform. There's a lot of risk there. So if you can do a search based upon not just what's

[00:31:26] out there in the marketplace for candidates, but also compare them to internal employees and the ability to upscale a current employee give them some mobility within the organization. That's going to be a far more effective approach than going out and running job advertising.

[00:31:44] So the skilling model that we have applies to employees as it does an external candidate and being able to show a talent acquisition rep, here's the employees that just need a little bit of upscaling within the organization, giving them upward mobility versus the spend and the

[00:32:02] time on the risk associated with hiring an external candidate is a unique capability. Hmm. Should it full be worried, Terry? Eightfold uses our software. Oh, today they're taking advantage. Perfect. Terry this was fantastic. It's always a pleasure to talk to you. It's always great to

[00:32:23] get your insight. If anyone's listening and wants to reach out to you, what's easiest way? Yeah, LinkedIn is probably easiest way to reach me. Terry Baker very easy and daxra is a daxra.com. It's daxra technology's limited and the website is daxra.com.

[00:32:41] Okay, so I had it. You nailed it. I nailed it. Finally, they got something right. This was fantastic. Again, thank you so much for joining us, Terry. Yeah, thank you. There's always a pleasure talking to you guys. Thank you, Terry, Arvwa.

[00:32:56] Shelley, let's face it. Taxing candidates is the easiest way to hire quicker today. But your cell phone doesn't connect to your ATS. You're sharing your personal number with strangers. It's pretty scary, right? Shelley and it's not even legally compliant. This is where our friends at

[00:33:20] Rectx come in. They've created simple yet powerful tax recruiting software that works with your Plus, it's designed by recruiters for recruiters. So you know what works. To learn more and book a demo, visit www.rctxt.com. Mention the recruitment flex and get 10% off annual plans.

[00:33:44] Do you love news about LinkedIn, indeed, Google and just about every other recruitment tech company out there? Hell yeah, I'm Chad. I'm cheese. We're the Chad and cheese podcast. All the latest recruiting news and insights are on our show.