HRx Radio – Executive Conversations: On Friday mornings, John Sumser interviews key executives from around the industry. The conversation covers what makes the executive tick and what makes their company great.

HRx Radio – Executive Conversations

Guest: Shon Burton, CEO, HiringSolved
Episode: 369
Air Date: June 12, 2020

Transcript

Important: Our transcripts at HRExaminer are AI-powered (and fairly accurate) but there are still instances where the robots get confused and make errors. Please expect some inaccuracies as you read through the text of this conversation. Thank you for your understanding.

Full Transcript with timecode
 

John Sumser: [00:00:00] Good morning and welcome to HR Examiner’s Executive Conversations. I’m your host John Sumser. Today, we’re going to be talking with Shon Burton, who is CEO and founder of HiringSolved. Hi, Shon.

Shon Burton: [00:00:25] Hey, John, how are you?

John Sumser: [00:00:26] Yeah. So welcome to the show. Would you take a moment and introduce yourself?

Shon Burton: [00:00:31] Sure. You did a great job.

I’m Shon Burton. I’m one of the founders and the CEO of HiringSolved and been in the industry. Didn’t know anything about HR Tech when I got here, I guess that’s, I’ll share that about myself. Didn’t know anything about HR in general. And so it’s been a steep seven or almost eight year learning curve.

John Sumser: [00:00:47] And it’s been full of twisty turns as you’ve gotten to know the space. What led you to be here?

Shon Burton: [00:00:55] So I’ve been doing startups for a long time. Did my first one in the year, 2000. It’s crazy as that is. And in 2009, after doing a bunch of tech stuff, engineering stuff for a long time, I’m sitting in San Francisco and it’s the very beginning of 2009.

Things were upside down and a friend of mine who was a great recruiter talked me into starting a recruiting company. And I had only ever been recruited at that point. And I thought, wow, this is going to be a piece of cake. I’m an engineer. We’re sitting in San Francisco, I know this great recruiter and let’s do it.

And it was one of the hardest jobs I’ve ever done. We did have some early success and we scaled that business into a multimillion dollar business, but the amount of work that we had to do and just the type of work and how grueling it was quite a shock to me. And that’s what gave me the idea, at least at HiringSolved and put us on the path of figuring out how do we make this a little less soul crushing in terms of the workload?

John Sumser: [00:01:43] Well, that’s interesting. So tell me about HiringSolved. What is it? What does it do?

Shon Burton: [00:01:49] HiringSolved, you know, like the name says, we were trying to solve hiring for the last eight years. What we do is we look at the data. We are very good at structuring data. So I think nowadays they’re calling us early. It’s early in the morning for this question, Don.

I gotta say, I think we’re being called a talent intelligence company now. What we do, like I said, has been the story of how we got here is try to figure out how to make hiring easier materially. What that means is there are, there are things in the recruiting workflow, in the hiring workflow. They’re extremely labor intensive, for example, today, you know, maybe it’s sorting through and figuring out who in a thousand or 2000 applicants that you just got, who are you going to call first.

And to do that, the human has to look at every resume, for example, because every application, so we do some of that work for you, do a first task for you automatically, and then other functions like that. If you think about it, I always think about what we do is we try to solve the data problem and recruiting the data problem is we’re just too much.

Data and humans are really, really good at pattern matching and recruiters are amazing at figuring out the pattern and understanding who’s going to fit. But when you look at it from perspective of now, you have to do that on a thousand or 2000 or 10,000 applicants. All of a sudden it goes the other way.

And it gets too hard to do that function. So that’s where we step in. We do a lot of stuff that’s like that with data and deriving intelligence from data to arm the recruiter. So they can quickly sort through it, our missions a lot like autopilot in the Tesla, right. We’re not trying to replace the human.

We’re trying to give the human a bunch of information and sort through it very, very quickly and say, here’s what we think you should do.

John Sumser: [00:03:15] So this means that you are trying to make sense of the increasingly monstrous piles of job applications. So you get this 10,000 applications say for a job, let’s say that you’re a Caesar.

Some of those applications are good enough to hire. How do you show that the question of how you visualize this kind of data is central to your work? And so let’s talk a little bit about what it used to be looking for a needle in a haystack. And now it’s looking for a needle in a haystack for a needle.

How do you help there?

Shon Burton: [00:03:47] Well, you said the right word, which is visualized. You know, I always look at it as the type of math that we use to do this function is all, has been around since the fifties, you know, and got a little better in the eighties, but it’s all old open source math. Right? Everybody has it.

That’s not really the secret sauce. The secret sauce is how do you instrument that and how do you interface that, visualize that to the human so that they can do something with it. And that’s, you know, when we think about how does that work, we look at something like 10,000 applicants. We have a technology, we call transparent scoring.

What it does is it shows the recruiter exactly what it’s looking at. And that’s one of the big problems in automation today is it’s very hard to get something that works well and does something for you, but also is very, very explainable and understandable by the person. We spent a ton of time working on that and getting a happy medium.

But at the end of the day, it comes out as one to five stars. So all 10,000 applicants are going to land somewhere between one and five stars, five being the best. So that’s how to visualize. And then beyond that, you know, it’s very important for the interface to constantly be reminding the person that’s using it.

This is what this means. This is why this person’s five stars. And you had a hand in deciding what was five stars and what was two and three and four, you know, and that’s just one of the features, but that is material kind of what that allows you to do is there’s a lot of complexity underneath that. But what it allows the user to do is say in real time, very quickly work on that scoring algorithm to get to something that feels comfortable.

And then once we got, you know, maybe like you said, maybe it’s a hundred people from that 10,000 or maybe it’s 50 or whatever it is, we can continue to filter it down by things, you know, the normal things people do like location or. You know, additional components that maybe we didn’t put in the requirements initially down to even smaller list.

And then we can operate on that. So that’s a lot of today what people are thinking about. We’ve had that function for over a year. It’s in our newest version of the software, and then it’s becoming quite handy right now.

John Sumser: [00:05:33] So you’ve been at this eight years. I understand that there are six versions of the software on the streets. Now is version six. What’s the evolution been like? That’s a long time to be working on a problem.

Shon Burton: [00:05:47] It’s a really hard problem. It turns out I always joke with my cofounder, having known how complex HR tech was. We would’ve just started video game company. Yeah. Probably we would have eight games by now.

No, it’s a good question. So the evolution has been, and by the way, the reason that that is true. And I always, I love this industry now because it is so hard and I, and there’s so many people now rushing into it, but you know, software is hard to build no matter what to make it good. HR tech software. It’s not enough for it to just work well, it’s in a heavily regulated industry and it’s working on people.

So it’s quite a bit different. So when we look at that version, one was honestly, version one was really just search. It was, things were hard to search in the very first version had the people aggregator, I guess. You know, most people know, say, look back far enough in our history, we got around version four, we’re now at version six, like you said, around version four, we got sued by LinkedIn.

And what we were doing was back then, we were looking at all of this data, some of which was LinkedIn and doing things like figuring out if a person had a LinkedIn account and a GitHub account, for example. And that was one of those things that at that time it took a lot of human effort to do. And it was one of the things that in that little recruiting company would do manually.

Using Google and all kinds of big Boolean strings and things like that. So that’s been this evolution in the very beginning, we have been thinking about things like matching and scoring from the very beginning and the very, very first website that we ever built, talked about that stuff. And the first thing we ever built back in 2011 was a matching system.

But what we found and how things have changed in the industry is that that was not very, when we went to try to sell that back in 2012, people didn’t want it. People didn’t like the idea of matching. And that’s been a really interesting thing to see change in the industry is that back in 2012, what we could sell and what we really became known for was search incredibly fast, faster than anyone had ever seen and good to use, nice to use and smart search because what we got feedback on early homeless, I don’t want your matching.

I don’t want a machine to tell me who to hire or who to look at or who to call. That’s completely changed. It’s funny. Cause the technology really hasn’t for us changed a ton and they go back to the 1950s math anyway. So now everybody not only wants things like matching and automated ranking stack ranking wouldn’t even call it, but they expect it to sort of work a certain way.

So that’s been an interesting evolution and I see it as still today. I think of a lot of what we do is making sense of information, structuring information at a very high scale, and then presenting it to the user in a very easy to access way. And some of that still is search, but now this other stuff like ranking and matching and analytics and other functions are now being put in the system.

Thankfully, we’ve just gotten to a place where we’re very viable now, post COVID world. And that’s something we’ve always been, trying to understand is what will happen when the market changes. And, you know, you have what you have today, which is a lot of people aren’t employed. So we’ve been working on that problem for lawsuit, but that’s sort of the evolution originally search and now something quite a bit smarter than that.

John Sumser: [00:08:44] That’s intersrting. So, the market is catching up with you. What do you suppose is driving the move from wanting to do complex, and precise Boolean search into trusting the machine to do that for you basically. Isn’t that the transition?

Shon Burton: [00:09:00] It sort of is. And to me, I see it as I see this exact same thing has happened in other areas of technology.

For example, I was a security networking security engineer, and back in the nineties, it was very complex and everything was, you know, these, these textual interfaces and a lot in a lot of ways, we all made our living that way. And Cisco had very hard certifications, incredibly hard, you know, CCIE. At some point, it started to change over.

I remember seeing in my career, someone else bring out of machine network gear, it sort of had a graphical user interface and kind of did a lot of this stuff for you. This stuff that we would do, will you type commands into these things to do. And it’s interesting to watch that evolution still today.

There are people that make a tremendous amount of money knowing that stuff, they know how to do it manually. And they are the top one or 10% or something, something like that. I see that still today with people that know Boolean is you’ve got this very top layer and they’re very, very good at it. And they can still do amazing things even with our software.

And we have a really verbose Boolean syntax and probably one of the best. But the problem is you can’t really get that many of those people. And that was a problem for Cisco. That’s actually why they started make it easier back in the day in network engineering stuff, Cisco realized, gosh, we can’t sell enough.

Cisco boxes, cause not enough people know how to use them and work on them to sell more of it. So they started to try to make them easier because it made it the technology more accessible. And I see that exact same thing here is Boolean is amazing. If you really truly understand it. And all of the operators and the language to it, you know, our search has probably 35 operators beyond the original Boolean and we’re not right.

You can do all kinds of stuff, but the amount of people that have an appetite to actually do that and to learn that and to do it really well is very limited. And that hasn’t changed. So, what I see is now that the interfaces have gotten better, there’s a trust problem with automation. You don’t trust it.

The human doesn’t trust. I always think about, would you buy a car without a steering wheel? We’ve had these automated systems for awhile. You know, I personally had a Tesla for almost four years and it’s always had self-driving, but I wouldn’t buy it without a steering wheel. Cause I still don’t trust.

And for good reason, it’s not, it’s tried to kill me a few times. So, I just see this evolution. But I do use it. It is good enough to use and sitting in traffic in Southern California, I’m very happy to have it right. Cause it’s taking that load off of me. I don’t want to do that driving. That’s less frustrating to use it. So I think that’s the evolution that I see is in 2020, the systems have gotten better and they’ve gotten better at explaining what they’re doing.

And so people kinda like, yeah, this isn’t so bad. Kind of like the spam filter, same thing with that originally we didn’t trust it. I don’t want a machine telling me what email I should read, but it got better and better and better. And eventually today that’s actually something we’re quite willing to offload.

So I think that’s a lot of what’s happened.

John Sumser: [00:11:35] So, how do you help your customers trust the machine? Right. Cause that’s basically what you’re saying is over time, people have started to trust that the machine will do an adequate job, but that’s almost a conversion process. It’s kind of nightmarish to build a business over eight years of that premise when you’re that far ahead of the market. So what’s that been like?

Shon Burton: [00:12:00] I mean, fortunately, we’ve had really smart people that are at our customers or our advisors that have helped guide us along the way. And what we’ve learned is that those original systems that we built were completely okay. If they would make a decision and to say, this is the answer, this is the one like a calculator does.

And you didn’t get to see any of the inputs or what went into that. And that was very scary to people, especially if you’re taking a list of hundreds or thousands people down to 10, that is very jarring to people. And you see the same thing in most automation systems, like the autopilot type systems is another one.

So what we’ve found is that there’s a great bit of subtlety and importance around. The interface of these systems and how, how they communicate and how well they instrument their process and what they’re doing to the user and let the user, you know, I always think of it as iron man, iron man is a person inside of a suit.

And inside of that suit, they have Jarvis, which is the AI that helps control Ironman and Jarvis doesn’t control Ironman, Jarvis doesn’t pilot the suit at all, but it does the suits very complex and it has all these crazy functions. So what Jarvis does is translate what that human wants to do. Tony stark 40 wants to do, and also give helpful insight, right?

Rather than piloting the suit. Jarvis says, Hey, there’s 16 inbound missiles. They’re closing at this rate. I suggest you click left and it’s displaying at the same time. That’s, what’s so cool about, you know, sort of looking at this example of displaying great infographic on the screen and you see what it’s doing and what it’s tracking and why it’s recommending that.

And so that was an early inspiration for what we do. And we’ve been guided along the way by people in the industry that actually were explaining to us, you know, what it needs to be. If you have a system that’s learning automatically, for example, one of the people that was a customer that we hired, Heather Thomas said to me a long time ago, Hey, this is great, but it learns right over time.

And I said, yes. She said, well, how do I know when it’s done learning? Like how do I know when it’s had enough to make something good? Because it gives recommendations right away. So, you know, stuff like that, you know, we’ve had a lot of great input from users telling us what this instrumentation should be.

And originally there was not literally, it was just like, here’s the answer. And so now it’s evolved to a place where we’re, we’re getting to that more of that Ironman sort of scenario, where the system is ingesting and processing and understanding a bunch of data. And then it’s visualizing that for the user and saying, I think this is a five star candidate because of these things.
And then the user has the opportunity to say, well, here’s where you’re wrong, or let’s change that or influence that behavior and see, and see in real time how that affects everything else. So there’s more of a give and take between the user and the system. And I think that’s know that’s across the board with these automated interfaces, same, same thing.

That’s the thing that you have to get right. It’s not so much the math part. It’s not so much that can you match people to a job description and some of that stuff, it’s really, how do you do that? Because that is always going to be imperfect. And that’s the other thing that’s always going to be imperfect.

So how do you do that in a way that is helpful and gets the user some assistance, but also let them understand what you’re looking at and influence it and therefore trusted.

John Sumser: [00:15:02] So, part of what you’re suggesting, and I imagine that this functionality is part of the product, but part of what you’re suggesting is that the real job description exists when you find the right person, rather than that, you start with an approximation of what that is, and you get results associated with what that is.

And then the recruiter tweaks those results because they’re not quite right. And so what they’re doing in fact is modifying the job descriptions on the fly. I wonder if you’ve thought about how you would document that as a backup. So you said you are six years of underwater basket weaving. When it turns out that you’ll take two, you might want to notice this job description

Shon Burton: [00:15:48] Exactly, I mean that’s the, you kind of hit that on the head.

This is part of the reason why it’s so important is that job descriptions are inherently terrible. They just are. And we’ve, we’ve seen more of them than almost any other company. Cause we, we look at every us job description. That’s published everyone in the United States. We look at it every week. So we have a system do that to your point.

They’re on average, pretty bad. So that is part of what the human equation does here is translate that into what the real world is. Yes, they are having these conversations with the hiring manager as they are doing the things humans do that the machine can’t know about, right. The machine has to operate in this world of that’s the job description, and this is the resume in that.

That’s what I have to work on. So bringing, bringing that in exactly as, as the job description, in fact, the suite one, one thing that we’ve started to do. Just this year is also instrumenting that using that data to instrument that for the, for the user as well, which by that, I mean, Hey, you know, here’s this job description.

It’s this title, it’s these requirements, did you know that the last five people that were hired for this, here’s what they look like. And so, you know, we find this incredible wealth of data, for example, in our, in our customers were exactly, as you said, job description says it requires a, you know, a PhD, but the last three people hired had a bachelor’s degree for the same role or for something that the system thinks is very, very similar.

So bringing the power of that information forward. Cause we all know that that is a lot of. The intricacy here and that’s part of the game is built. That’s why HR tech is complex by the way, is so complex is aside from the regulation and all the stuff you’re working, always on the system is always working on imperfect set of data.

And there are always a side channel conversation going on in the human world. The system is not going to be privy to, as you sat down and had coffee with that hiring manager and said, Sarah. Do you really need 10 years of JavaScript? No, actually that makes sense. Great. And maybe they update the job description, maybe they don’t, usually not because it’s published.

John Sumser: [00:17:42] Right. I could talk to you all day. We’re running towards the end of the time. Is there anything you want to let new functionality covers down the road or the final release of the new version? Anything you want to add?

Shon Burton: [00:17:59] Oh, my gosh, I’m not going to leak too much. I will say that the next piece that we’re building right now is really this incredibly powerful analytics package.

And some of the stuff I just alluded to is, you know, it gives you the ability to visualize things like success profiles. And who’s really worked here in this position before. And where have we hired and what does the system consider more relevant? We’re taking this idea of instrumentation and. In analytics and information, a large step forward and building that as something that could be a standalone product in that sense that you can just explore your own data and understand what you’re actually doing and what you’re hiring and what your talent pool is and how to shape this.

So, a lot of that’s already baked into our system in different areas, but we are expanding that greatly in the next release.

John Sumser: [00:18:40] I think the one thing we might have missed in this conversation that would be useful for people to know is that HiringSolved straddles the ATS and the CRM and other pipeline sources so that you get a comprehensive, single point picture of everybody who’s in your system.

And I believe you have an operating theory about sourcing. That’s something like: if you’re a sizable company and you’ve been recruiting for 10 years, you have everybody you’re gonna hire in your database already. That a fair way to go out?

Shon Burton: [00:19:15] Absolutely. We look at it as you are sitting on the gold mine of data, but the gold line is a mine. We are the mining tools. The mine never said it was easy to get the gold out of it. And sort of that’s your ATS, CRM, HRIS. They’re a gold mine. But gold mines are actually quite hard to get the gold out of. Right? We’re the mining tools.

John Sumser: [00:19:32] Got it. Thanks Shon. I really appreciate the fact that you took the time to do this today and tell people how they can find out more about HiringSolved.

Shon Burton: [00:19:43] I’m Sean Burton. I’m the CEO of HiringSolved. You can find out more information by going to HiringSolved.com. Going on Twitter slash HiringSolved and then I’m also on Twitter just slash, S-H-O-N. Twitter.com/shon.

John Sumser: [00:19:56] Thanks. Appreciate you taking the time to do this. You’ve been listening to HR Examiner’s Executive Conversations, and we’ve been talking with Shon Burton, who is the CEO of HiringSolved. Thanks for tuning in, and we’ll see you back here next week. Bye bye now.