#HRTechChat: John Sumser on Compensation, Pay Transparency, and the Future of Work

Posted on by

John Sumser, VP of Marketing at Salary.com and a seasoned industry analyst with over 30 years of experience, joins Dylan Teggart for this episode of #HRTechChat. Together, they explore the shifting landscape of compensation and the profound ways pay transparency and AI are shaping the future of work. From his beginnings analyzing the job board industry at the dawn of the internet, John offers a wealth of knowledge on HR technology and the critical role compensation plays in business strategy.

In this episode, John shares his unique perspective on key trends transforming the workforce, from navigating an aging labor market to the challenges of integrating vast amounts of data into actionable insights. With a thoughtful approach to ethics, technology, and innovation, John provides listeners with a deeper understanding of how organizations can adapt to thrive in an evolving workplace.

Our #HRTechChat Series is also available as a podcast on the following platforms:

See a service missing that you use? Let our team know by emailing research@3SixtyInsights.com.

Transcript:

Dylan Teggart 00:00
Hey everyone. This is Dylan Teggart, reporting in from 3Sixty Insights. I’m joined today by John Sumser. John’s joining me for our latest #HRTechChat. John is an industry analyst. He’s got over 30 years of experience, and he’s currently working as VP of Marketing for Salary.com. John, thanks for joining me.

John Sumser 00:20
Yeah, thanks, Dylan, it’s nice to be here

Dylan Teggart 00:23
before we get into our conversation today. Do you mind telling people a little bit about yourself and what you’ve been doing the last 30 years?

John Sumser 00:29
Sure, sure. I started as an industry analyst before I knew what an industry analyst was. In those days, three years ago, there were six or seven job boards, and I started watching them and getting to know the players in the job board business. That was just as the web was beginning, and that evolved to become the site called intervisit.com where I wrote about digital recruiting and the job board industry every day for 15 years. At the end of 15 years, I thought, well, maybe I’ve covered everything that there is to cover here. And so I so I broadened out and spent the next 15 looking into the broader HR universe and and so some of the kinds of things that I did over that time is, there’s, there’s a shortage of good, objective research about fundamental things in the business. And so I spent a couple of years building some of those. I spent about five years helping people design ethics boards for AI governance, and that was an extraordinary dive. And these days i i like to say that I was a food critic for all those years and got a job in the kitchen, so I’m working in a software company these days.

Dylan Teggart 02:06
Interesting. So on. What did ethics and AI governance kind of look like in the early forms?

John Sumser 02:12
And just, I know this is a little off topic, but just to someone early, early on, no, it’s good. I’ve been following it for a long, long time. The stuff that, the stuff that is getting all of the press these days, really has its roots almost 15 years ago, and that then it was called big data. And big data has evolved. The underlying things have evolved, and now you have this bubble. I think it’s a bubble with large language models. And along the way, the question became, how do you account for bias inside of large language models and other forms of AI? And that was the heart of the question that ethics boards were dealing with. And so I designed a variety of things. Part of the problem when you build software is that nobody can afford to have all of the necessary points of view to eliminate bias from software design. And so a function of the ethics board is to be a supplement to the design process so that you can catch bias related errors before they hit the streets. And that looks like on one case, I built a team of 20 industry experts from all walks of life who met quarterly to guide the ethics of a company, to a couple of others that were as small as me, somebody from the board and a couple of people from inside of the company as the ethics thing. So we were looking, I think, for a model for ethics. And I think that my read is that the interest in ethics, ethics is a funny thing. It doesn’t pay the bills and so, so it’s hard to get funding inside of organizations, and the need for ethics functions in a seems to have diminished. It’s more important than ever, but there’s less emphasis on it.

Dylan Teggart 04:26
I think it’s interesting that it’s diminished, given that, wouldn’t it open people up to legal liability, if not handled correctly, because they the lines are kind of blurred with law and what could be future law on AI, and also where ethics tie into that.

John Sumser 04:45
So the thing about the thing about ethics, on the one hand, is that it is a compliance issue, right? The discrimination is against the law across the board in the United States. And so the way that you test whether. Or not a system is biased. As you just look at its results, the outcomes are not biased. The point of an ethics board is to catch that stuff before it turns into a liability problem. Now, one of the things that’s interesting about liability in this stuff, if you read the contracts that most organizations sign software providers in general, distance themselves from liability, and the government holds the final decision maker as responsible for the quality of decisions, and so the liability falls squarely on the customer or the employer, depending on how you want to characterize it, and that creates an interesting model, because I don’t think that most employers are in a good position to figure out what to do with that interesting Well,

Dylan Teggart 05:58
it’s interesting to see where It goes in the future, because I think a lot of people, it’s something a lot of people are worried about, I think, and a little bit of an unknown, what will happen as these models become more and more intelligent and more and more effective?

John Sumser 06:15
But well, that’s an interesting question. I’m not sure that’s going to happen. I’m not sure that they’re going to get more intelligent and more effective. I think, I think that’s certainly the hype. That’s certainly the hype. But if you watch the news over the last couple of days, all of the large language model companies are acknowledging the fact that scaling doesn’t work at this point in the process, and when scaling doesn’t work, this starts to look like flying cars again, or self driving cars, or something like that. And that’s happened before in the history of AI. So I think, I think one would be wise to because cautious about what you believe about the future of AI.

Dylan Teggart 07:10
Well, that’s a interesting point. I’m sure there’s going to be plateaus along the way, just like with any technology, you know, we’re gonna have a big leaps, and then stuff will kind of slow down and plateau for a bit, and then maybe, I can only imagine sometime in the future, they’re gonna advance again. But how long that takes is anyone’s guess, I suppose, like you said, yeah,

John Sumser 07:32
yeah, yeah. So, so I think we’re headed in a direction, but, but I don’t think we know whether that’s five or 100 years away.

Dylan Teggart 07:42
Yeah, for sure, some changing, changing gears a little bit here, and moving into paying compensation. I know that’s something you look at quite a lot. And speaking of the future, I guess we could start, start right there, based, you know, given where we are right now with paying compensation, where are you seeing the trends that are happening right now, kind of branching off into the future?

John Sumser 08:08
Well, there’s a there’s an explosion of data sources for compensation information, and there are some serious questions about how good they are. You know, anytime there’s a disruption, it always starts out cheap and low quality. That’s what disruption looks like. Anytime it happens is it’s cheap and crummy, like, remember? Do you remember what the first digital photos looked like? They were heavily pixelated and nowhere near as good as you can produce with the camera, and it took a decade to get to the point where you could start to think about doing digital photography, and it took another five or six years before the transformation was complete, and that was mostly egged on by the integration of the camera into the telephone, but it took, it took a long time to go from yes, it’s possible, to Oh, it’s everywhere, right? And so, so that same kind of thing applies to what’s happening with data associated with compensation. So the other thing to think about as as things evolve, is the pay the pay Transparency Regulations have had an enormous impact on the marketplace. Before 2023 there were no pay Transparency Regulations today, just before the start of 2025, 50% of all American workers are covered by pay transparency laws of some kind. And so what that means is that a whole lot of stuff that looks like compensation. Data is all over the marketplace, and as pay transparency progresses, that data is going to get better and better and better and better. So if you want to find out the pay for a job, that data will become prolific, and there’ll be some interesting questions about how you tell quality in data. Be some real interesting questions about how you tell quality in data. At the same time, having market pricing for a single job doesn’t get at the heart of compensation. You know, compensation is the organization’s tool for governing the single largest item on the balance sheet, labor cost. And the reason that you have a compensation function is so that you can introduce predictability into labor costs, because it’s a really hard thing to pin down, and it tends to be where profitability of companies live or die. And so data explosion does not get your data structure. And one of the things that that I think will start to be a conversation is whether or not data structure is the larger part of what constitutes quality information. And so I think we’ll see that shake out.

Dylan Teggart 11:31
So you’re saying essentially that even though all this data is out in the market, which is a good thing, especially for, you know, transparency for workers, if you don’t, for businesses, if they don’t necessarily know what to do with it. It’s kind of just data for data’s sake. Well, just like

John Sumser 11:48
if you had all of the books in a library in a great big pile, you can say you had a lot of books, but being able to find out what’s in that pile takes some sort of disciplined structure, so you know where to put the books, and you know how to think about this category versus that category. And that’s not part of a single metric for what should you pay for this job or that job?

Dylan Teggart 12:17
So what do you what do you see as viable solutions going forward with that in mind?

John Sumser 12:24
Well, it’s interesting, because compensation professionals currently take every shred of data they can get from wherever they can get it to arrive at the conclusion that they draw about about pricing for jobs. And so an explosion for data is probably heaven for compensation professionals, because you get, you know, it’s like your spice cabinet opens up and you get more spices to play with when you’re making something. And so I imagine that the compensation function will evolve, and that that it might spread. Currently, compensation exists as kind of a hard silo. It’s it’s all about data and analysis, unless it’s involved in approving a job offer, but the rest of it is fundamentally data and analysis, and so as the data explodes, you might imagine that compensation will spread out into the rest of the organization in an interesting way. There’s a case that that I’m not sure how much emphasis to put on it, but there’s a case to be made that everything that HR does is an expression of compensation. You know, so why do you have disciplinary conversations or performance management conversations, or deliver benefits or deliver payroll or deliver variable pay, or deliver training? All of those things are all about forms of compensation and reasons for paying and so. So you might imagine that any of those functions can be better informed if they have $1 figure sitting in front of them,

Dylan Teggart 14:21
and how, in terms of how this kind of impacts employees, obviously it gives them more for workers, I should say it gives workers more insight into what they could be making. But do you think this will have a positive or negative effect on their wages in the long term? I Yeah.

John Sumser 14:45
Well, I think there’s a couple of things. The first benefit of pay transparency is that it chases favoritism out of the organization. So with pay transparency, every employee will know how. They’re paid and why they’re paid that much. That’s what paycheck. That’s what the future pay transparency looks like. So it’s not a mystery any longer why you’re paid what you’re paid. There’s some chance that that leads to a structure that has jobs and levels in it, and those levels get hard to move through. And so there’s some chance that that this starts to restrict movement, but generally that that might be seen as a good thing. It’s how compensation works in Europe, and it’s how compensation works in the government. It’s only in private industry in the United States the compensation works differently,

Dylan Teggart 15:48
so almost like pay grades in the military or pay grades in the government, exactly. Interesting and in terms of, I guess this ties into another question I want to ask you is how skills tie into this, and how Job replacement, replacement, if any, will play into this change in compensation. It sounds like they could result in more rigidity, but when people are now trying to compensate people for skills more. It seems like that’s the talk on the street. How does that play into it? You’re then tying certain skills a certain compensation, and compensation can adjust that if you have skills. I guess that already happens in effect, essentially. But how? Because there needs to be this heightened transparency, does that mean that you’re going to have price markers for specific skills that, okay, your job pays 70k a year if you have, you know, good skills, but if you have great skills, if you have a master’s or whatever PhD, that adds $8,000 a year to your compensation, or something like that. Is that? Is that? Is that going to be something we could see this transparency trend leading to like, almost an a la carte style menu of skills tied to compensation.

John Sumser 17:06
Well, the interesting thing is, skills don’t implicitly have anything to do with what the job is. So you can have a bunch of skills, but if I dump you into a work environment and your job is to produce x result. Finding the relationship between those two things is the thing that’s slowing down hiring and managing on a skills basis, right? So, so I’m unaware of any, and my awareness is limited, but I’m unaware of any fully executed skills based compensation program, and I am at a loss for telling you how that would work. I think, I think it’s liable to becoming, and it’s liable to be coming, because whether or not AI is the answer, there are some radical shifts going on in the way that work is structured, the way that work means, what work means, and the differences are, some skills are decreasingly important, and some skills are increasingly important, and jobs change as skills change inside of those jobs, what are we going to see? Well, I have been experimenting with large language models for a couple of years now, and it seems to me that some fairly rudimentary things are able to be done that that didn’t used to be able to be done. And so it’s, it’s analogous to when I came to work every every work group had a secretary, and you don’t have secretaries anymore, or very, very few of them, because word processing replaced them, that kind of thing. So what were the basic skills that word processing replaced document creation? The idea that you would create your own documents was not something that nobody heard of in 1980 but today, the desktop is where all document creation happens. The change in skills suggests that there’ll be a change in the way that people think about compensation. So if you’ve got a narrow skill that’s highly valued and all of a sudden it can become automated. It won’t be. Long before you’re not paid for that thing that you used to be paid for. And so the question is, what replaces that right? This is, this is in the great unknown about the impact of technological change over the next 10 years, say, about how it’s really going to affect white collar work, and what parts of white collar work are going to get automated and what gets replaced? And that’s it’s been the topic of year end HR trend forecast for about five years now. It’ll start to pick up momentum in the next couple of years? Yeah,

Dylan Teggart 20:44
I feel like everyone’s kind of at the edge of their seats wondering if they’re Next, if they’re the next secretary. I guess you could say because human skills are obviously going to be something you can’t necessarily replace at the moment, but when it comes to other functions. We don’t really know what that’s going to be, obviously, physical tasks or something you can’t quite replace, you know, one to one all the time. But for Yeah, white collar jobs, it is a it is a bit of a wonder. Which leads me to one question. I it sounds like, you know, there’s, there’s been a lot of talk about labor shortages and certain fields in certain industries. I’ve always been a little skeptical of that, just because I feel like it’s largely a pay issue. A lot of the time people, there’s a demand for certain skills, but the incentive isn’t necessarily there to compensate them properly. So there becomes a shortage of skills, but it ultimately all kind of ties back to compensation or a lack of funding people to be educated on those skills, like you would in other economies where certain things are subsidized by the government, certain types of education are subsidized by the government because they need those skills. Do you, how do you, how do you view that in terms of the compensation function, and do you think labor shortages are going to really affect things that greatly for businesses or

John Sumser 22:16
Well, so, so the labor shortage thing another, another term for labor shortage is aging workforce, right? So the labor shortage is a material fact, and the material fact is, for 40 years, family size has been shrinking, and so at the startup, end of the work life cycle, there are fewer replacement workers. Simple, it’s simple math, when, when, when I was born. Average family size was almost four kids. Today, that’s well under two kids. And that means that half of the replacement workers that you would expect are not there. And so what you see is people getting older in their jobs, and older people hanging on to their jobs for longer. And that’s what the labor shortage looks like most places. That labor shortage, another way of talking about your view that that that it’s really a pricing problem, that you that, in other words, you can’t get any Java developers for $50,000 a year. And so if you went out looking for Java developers at $50,000 a year, you wouldn’t find any right that’s that’s one way of thinking about it. But the other way of thinking about it is the labor shortage creates the pressure that make wages go up. And wages go up for a variety of reasons, including tenure and jobs. And as the workforce gets older, people stay in their jobs longer and so, so you’re starting to see this bubble up pressure about what people get paid, that that will be the source of inflation over the next four or five years.

Dylan Teggart 24:24
And for how do you how do you predict, or what is the best way for companies to handle that by investing further into optimizing their workforce? Or are you going to be seeing companies letting go of older people just because they’re more expensive to keep on.

John Sumser 24:46
Oh, you might see some of that, but it’s against the law, and so doing that results in huge settlements. And so that’s not likely to be a common thing.

Dylan Teggart 24:59
Well. Not, not explicitly, but there could be ways to circumvent that, like just by anecdotal means, of saying that someone’s not doing their job properly, or there’s restructuring. Oh,

John Sumser 25:11
you know, this is another place where what matters is not the rationale. What matters is the numbers. And so if you have, if you’ve got attrition rates across the board, and all of a sudden, your attrition rate for older workers spikes up. That’s a foundation for a lawsuit. And the great thing, or terrible thing, about pay transparency and contemporary data flows is that in order to figure out your internal attrition rates by demographic, you have to create something that is known in courtrooms as evidence. And so there’s a lot of discoverable evidence inside of companies that if there’s a hint of age discrimination, if there’s a hint of a whole bunch of different kinds of discrimination, it becomes the foundation of lawsuits. So there’s, there’s a governance mechanism in there. It’s imperfect, but it but, but it kinds of works. Kind of works. Interesting.

Dylan Teggart 26:27
Just before we wrap up, is there anything else you feel like people should be aware of? There’s one final point, one final trend you’ve been noticing, or one, one big trend you’ve been noticing that we’ve been touched on today. Well, I’m

John Sumser 26:42
going to come back to you about, how do you tell if your job is likely to be automated? And so while we were talking, I went to chat GPT, and I said, Could you monitor software companies and they’re changing product lines, drawing conclusions about trends on a continuous basis that sound like your job, and it gives me 10 items about what it could do to do that job. And so, so if you’re listening to this and you’re wondering whether or not you can be replaced, go ask one of the large language models. It’ll give you some idea of, of of how much time you have before you need to figure something out, or how much time you have before you need to redefine the value that you create. What’s, what’s the

Dylan Teggart 27:39
timeline I got over there?

John Sumser 27:42
Well, it says it can do regular analysis of product announcements, monitoring financial statements. It can observe mergers, acquisitions and partnerships, track patent filings, utilize data analytics and AI tools, and engage with industry conferences and webinars. So I mean, this is a joke, you know, but, but, but, but I’d look over my shoulder a little bit. I’d look over my shoulder a little bit. Now, you know, the the analyst business is all about point of view. And one of the things that you can’t really get in a large language model these days is some kind of focused, consistent point of view, and that’s really what makes one analyst for a different from another. And I don’t think that’s probably replaceable. You may find that you need fewer people to do the work. It’s

Dylan Teggart 28:40
interesting. You say that, because I do. I think ultimately it comes back to your you being yourself as your best brand. And while that’s not always going to be mean it’s desirable or in demand, but you, as your like you said, your consistent point of view is likely the best thing to make you continue to be relevant, because you’re bringing something that is not such a homogenized product that aligns with a lot of other people.

John Sumser 29:17
I think that’s right, as long as you continue to learn and work to keep your point of view fresh, right? A dead point of view is not very interesting, for sure, absolutely well,

Dylan Teggart 29:35
it’s been a very interesting conversation. John, thank you so much, and hopefully we can do it again sometime. If people, if you’re open to people getting in touch with you, what’s the best place for them to reach out to you?

John Sumser 29:48
My email address is john.sumser@salary.com

Dylan Teggart 29:55
Great. Well, thank you everyone for listening. And John Always a pleasure.

John Sumser 29:59
Alright, Thank you very much. Great to do this.

Share your comments: