How ADP's Former CPO Built a Category-Defining Ecosystem

(00:00)

Don Weinstein: Here are my three phases of, you know, how new technology frameworks get adapted into the enterprise. Step One is you take the technology and you try and force fit it on your existing model, right, my existing platform, and that never works.

Step Two is then the hybrid, it’s, well, let me run the both in parallel. So Step One is let me take my desktop interface and put it on mobile. Step Two is, let me create a separate interface for mobile. But now I've got two interfaces. And Step Three is finally, you know what? Let me just rethink the whole thing from the ground up here and create a mobile native, which becomes almost minimalist. And then I can kind of work backwards, back to the desktop.

(00:53)

Ansel Parikh: Welcome to the inaugural episode of The Current, where we explore the intersection of people, finance, and data. I'm Ansel Parikh, co-founder of Finch, the connectivity platform for the employment ecosystem. Today, I'm joined by Don Weinstein, a longtime enterprise tech operator who spent 17 years at ADP, where he acted as Global Chief Product and Technology Officer and the Chief Strategy Officer. Don, welcome to the show.

Don: Well, thank you, Ansel. Appreciate you having me.

(01:23)

Ansel: I'm really excited for our conversation today. A lot of folks in the employment ecosystem will already be familiar with you, your background, but I'd love to take a minute to give you the floor. Can you tell us a little bit more about yourself, your career, and kind of what you're up to now?

(01:38)

Don: Sure. Well, I think you hit the highlights. You know, 17 years at ADP was a good run. You know, before that, I was always in the enterprise technology space in a variety of different companies. Since then, I retired from ADP a couple of years ago. And since then, I've been very active in both the venture capital and private equity landscape, advising both portfolios in terms of investment decisions, as well as working closely with a handful of their portfolio companies on various topics of strategy, business, you name it. So staying very active.

(02:14)

Ansel: Nice, little bit of combo of operator turned investor, kind of still helping people with operations, which is pretty exciting. And it's really cool to hear about how you're giving back to the community to build this ecosystem to something that empowers kind of everyone involved. But I would like to take a moment — 17 years, you can't just glaze over that. And so I know that you led product and technology for a massive organization for quite some time.

Don: Yeah.

(02:40)

Ansel: You were really key, as part of the launch process to really make it something big — ADP Marketplace, which was an innovative model as well as approach to, again, building an ecosystem. So maybe, can you turn back the clock a little bit? Tell us a little bit about, how did that come to be? What were the problems you all were solving for?

(02:58)

Don: Yeah, no, for sure. And it was a good time. Sometimes I think back on my career and the handful of things I was most proud of that I did, and that's definitely on the short list. Look, we know that payroll and HRIS data is incredibly powerful. It's incredibly important. It powers a huge amount of use cases and transactions. And yet, stepping back from it, you know, I find in general connectivity in the HCM industry, it's just, it's always been a challenge. There's just no commonality. There's really no agreed-to standards. You know, there's the concept of HR XML has been around for decades but just never really broadly went anywhere. And as I mentioned, you know, a good background in ADP and human capital, but prior to that more enterprise technology. And if you look at an enterprise technology and high tech in general, I think that's an industry that has aligned around a number of standards, communication standards like SMS, like TCP IP, like 5G, you name it. Having those connectivity standards has really helped enable the whole industry to grow and to flourish. And then, for somebody like me, comes to human capital management and say, there's just like, there's no standards here. Instead, what you have is just this hodgepodge of interplatform kind of communication. I won't even use the term protocols, I'll just call methods. Things like SFTP, flat files, let's be honest, email. And as somebody who has a background in technology, this is not good. It's not safe from a security standpoint. We've got a lot of really important data, a lot of very sensitive data, and the fact that you've got people, you know, sending reports and flat files and emails all over the place is just, you know, that was the problem we were trying to solve. Let's put it that way. And I think we had good reference models from other industries that had come together.

(04:59)

Ansel: Yeah. No, and then you mentioned a bunch of different, I like to call them methods as well. I don't think they're always a set of technology.

You know, looking back, it's been over a decade, I think, since Marketplace launched. Is that integration landscape different today versus back then? Have some of these things changed? What are the things you think that Marketplace has really helped kind of push forward with relation to all those, those methods?

(05:24)

Don: Sure. Well, look, I think by many measures, the ADP Marketplace has been a big success. Probably the best measure I can give you is that a number of our competitors copied us. What's the saying, imitation is the sincerest form of flattery. So that definitely tells you something there. And look, we built the Marketplace to be an open, it was an open marketplace, API-based connectivity. So that's better, safer, it's more secure, it's more repeatable. And so I think it was very successful by many measures. If you look at it today, it's the most broadly adopted and utilized marketplace platform in the HCM industry, I'm talking about like a vendor-led marketplace platform with literally hundreds of third-party partners on there. Okay, so that sounds great, right? But here's the challenge, okay? 

The HCM ecosystem, as you know, is vast. I've seen different reports out there, different numbers, one that I aligned on was, I think from IDG had it pegged at like 20,000 participants in the HCM ecosystem. And that doesn't even include other non-HCM points of connectivity, right? Like financials, obviously a very common one. And with the advent we see of AI coding agents, we think that the number of startups is growing faster by the day. And so you can see the problem here is that even as I think the marketplace and marketplaces in general have grown and they've been successful to encompass hundreds of third-party partners with these standard integrations, that's just a fraction of the ecosystem.  And the challenge of inter-platform connectivity is just so much broader than what any one vendor can probably grow and support on their own.

(07:12)
Ansel: Yeah, no, I mean, you're preaching to the choir here, obviously. At Finch, our whole job is to be kind of that intersection and really help the payroll providers that frankly aren't ADP, that didn't catch up quickly to the expansion of this ecosystem and make sure that at the end of the day, the employers are getting the experience they expect, right? They want to make sure that, especially for mission critical information about their organization, it’s kind of synced across the board.

I'm curious if when you were kind of spending time on ‘What other things we can build here?’ what was one thing that you — again, what you can share — that you were like, ‘We thought it was a good idea, but it's actually not as interesting as people thought?’ I'm curious just to learn from people out there who are like, maybe I can build some things with this and some of the pitfalls when you make an assumption around the value of a data set and then realize that it's just not how certain parts of the industry work?

(08:05)

Don: I will say there's one that still surprises me a little bit where I was just convinced it was going to transform the industry, but I just say the adoption just wasn't quite there. And this is all around the area of performance evaluations.

And we had a team at ADP, they're still there and they do great work called the ADP Research Institute. And you're talking about PhD level psychometricians, econometricians doing real research and analysis. And this was again in the area of performance evaluations where they were able to kind of statistically prove that there's bias in performance ratings based on, I'll call it the “rating footprint” of the rater. And I think we've experienced this in our entire lives. 

So examples I'll give you is, know, somebody's a hard grader, a tough rater versus an easy rater. So how do you make comparisons across people who are being rated based on different raters? And so we're able to A, identify everybody's kind of rating footprint and then factor that back into the performance evaluation to try and remove the bias. And when I'm using a term like bias, again, I'm not using like sort of an evil bias. It's just like, you can always think, you can think back to your time in school, like, ‘you don't want to take that class, that teacher's a really hard grader.’ They just grade everybody the same, but to a tough standard or to a softer kind of standard. But if you're doing things like making pay or promotion decisions, or in some cases, restructuring decisions, and you're factoring in performance evaluations into those decisions, well, the performance evaluations are influenced by not just the — there's the ratee, but we actually found more than 50% of an individual's rating said something about the rater than the ratee.

(09:58)

Don: And so we created a system where we could evaluate everybody's rating pattern. And I'm like, this is brilliant. Like we know ratings are, are biased. Like, and we've got a statistical method for eliminating bias out of performance rate. It's like, this is going to kill, right?

(10:16)

Ansel: Yeah.

(10:19)

Don: Well, turned out, you know, explaining the concept of statistical bias to some of our buyers, you know, it was a more mathematical heavy concept than many of them, you know, were interested in having a conversation about. So as a result, you know, I thought it was going to be a grand slam and instead it just didn't achieve that outcome.

(10:27)

Ansel: Hmm. Yeah, I think it's to your point, I think the challenge with having access to all this data is that you — and maybe more having an engineering kind of mindset — is like you're thinking about the science of it all. But some of these things are art, right? And being a people organization is largely an art, right? You're creating, yes, there's some skills, things like that, that you're kind of building. But a lot of it's heuristics of just knowing the it factor and then also understanding what motivates people. And that won't show up in numbers sometimes.

And so maybe taking a step back more generally as someone who's managed massive organizations and is now helping startups that are probably managing a team equivalent on your end, what is advice you give to startups building in this ecosystem on kind of either how to build good technology teams or finding the right talent? Because this market right now, especially for engineering talent, technical talent, it's challenging and it can be something where everyone's going after the same people. So I’m curious how you've taken some of the things that you've learned about building a massive organization that attracts top talent at scale to some of the people that you're working with that are going, ‘hey, I'm trying to hire my first 20 people.’

(11:55)

Don: Yeah, it's funny you mentioned this. I'm working with a startup right now and I was having this exact conversation with the CTO just the other day and it's probably got a 20-person Engineering team, and they have a good problem right there. They're doing well. They're growing fast, he needs to hire fast, and what I really talked to him about was, think about the standards that you set for yourself when you made those first couple of hires, and the bar that you held them to. And now that you're moving to scale, don't lower that bar. The time you spend undoing a bad hire isn't worth the speed.

(12:40)

Don: So quality over speed every time. The second thing I'll offer though, it's that what I also hear a lot of times is I'm so busy, I don't have time to recruit. And my response would always be, the time that you're too busy to recruit is the time that recruiting should be your number one job.

Okay, I'm really busy. I don't have time to recruit. Okay, I really need people, but I didn't spend the time upfront. So now I gotta rush the hire and maybe I'm not bringing, you know, hire number 20 is not the same quality as hire number two. And the way you avoid that is you're always recruiting. And whether you're running a big organization or a small organization, that is the most important job.

(13:27)

Ansel: You always seem to be kind of like, ahead of the curve on what's happening in this ecosystem. So maybe if you could peel behind the curtains a little bit, like how do you keep track of all of this? How are you getting your information? And I think maybe the main question is how are you filtering through all the noise that is happening in the ecosystem? Are there certain things that you would suggest people that are listening in to, kind of do to get a little bit more firsthand knowledge of what's going on?

(13:44)

Don: Yeah. Yeah, no, well, thank you. That was a generous assessment there. But the first is you just really want to be aware of what's happening around you. Look, we're all, I was an engineer, you know, a lot of engineers out there. You just get very heads down and focused and you have to make time. It's just about compartmentalizing your day. And so I would, you know, I just think about the areas where I'd spend time. 

I'd spend time looking at what competitors were doing. You don't want to just copy your competitors because that's not a recipe for leadership, but you also have to be mindful of making sure you know what they're up to and not falling behind. The second thing is, of course, you want to spend a lot of time talking to your customers. But the thing that I did that I think was maybe a little bit different than others that was helpful was I would always look for these parallels in other industries, and I'd say, well, what could be applicable here? 

So as an example, we talked about the ADP Marketplace, which, yes, was it a breakthrough for the human capital management industry? It was. But the idea goes all the way back to the Apple App Store. So you'd say, well, coming from the world of technology, people ran on ecosystems. They built platforms, had ecosystems of ISVs wrapped around their platforms. So was like, OK, well, let me take that model and how might I apply it over here in this context? And so looking for these examples in other industries and other companies can be a really helpful way of sort of anticipating what's happening. 

I'd say the other case like that would be for folks who are in a B2B context, being very closely tuned into what's happening in the consumer tech world. Because while your customer might be some enterprise buyer, your end user is a consumer. And they interact with their consumer tech 10, 100, 1,000 times more than your tech. And whatever you think the experience and the workflow should look like, they've already been trained by the consumer tech world.

(16:00)

Don: That you can understand where that is and where that's going. It usually moves faster, consumer, than enterprise. So watching what's happening in the world of consumer tech, I think, can be very useful as a playback to enterprise.

Now, you know, the flip side of it was there's some areas where enterprise is a little bit further out in front. Like I will talk about cybersecurity and in particular, two factor authentication. And we tried pushing two factor authentication early. Again, ADP, a lot of money, lot of data, takes security very, very seriously. And we were pushing two factor authentication. And there were a lot of people complaining, this is friction. This is hard. Why are you doing it? And then once all the commercial consumer banks mandated it, you know, and that's like, you know, I just have to do it everywhere. Then people got used to it. And now nobody complains about two factor. So it's sort of like the consumer world when in terms of user behavior, they're the ones who really kind of led the way in that regard.

(16:59)

Ansel: I do like that you're taking a lot of learnings from consumer companies. Is there one consumer company that maybe when you were at ADP that you looked at and you were like, ‘wow, these guys get it, and I want to take some learnings from how they approach that?’

(17:14)

Don: Here are my three phases of how new technology frameworks get adapted into the enterprise. Step one is you take the technology and you try and force fit it on your existing model, right? My existing platform. And that never works.

Step two is then the hybrid. Is, well, let me run both in parallel. So Step one is let me take my desktop interface and put it on mobile. Step two is, let me create a separate interface for mobile. But now I've got two interfaces. 

And Step three is finally, you know what? Let me just rethink the whole thing from the ground up here and create a mobile native, which becomes almost minimalist. And then I can kind of work backwards, back to the desktop.

Google was one that really forced us to rethink the importance of search in terms of user behavior. Again, we spent a lot of time with expert users. Expert users hate search. Why do I need to search for it? I know where I'm going, right? It's a waste of time. It's a sidestep. But for casual users, Google trained the world to just go and type what I want to into a search bar.

I would say the third one that maybe will sound a little out there, but I think this will be the wave of the future as well, was Meta, or at the time Facebook, in terms of building a social interaction model around people and groups. And so then at ADP, we brought that back into the enterprise with this notion of teams and how does work get done in an enterprise? It gets done in teams, but guess what? Teams aren't always reflected in your org chart, right? 

And I think about my team when I was at ADP, yes, I had all my senior development leaders, but I also had my HR — they would call them my HR partner, and my finance partner, right? And the HR and the finance partner weren't necessarily like, if you looked at the org chart, they didn't report to me.

But if you think about, what did I need to succeed in running a large organization, in addition to, obviously, a highly talented team of development leaders? I needed a great HR partner and then I needed a great finance partner. And so what we did, and this is in ADP's large enterprise product called Lyric, it's a flexible org design that's built on more of a graph database, which really Facebook were the early ones who pioneered the graph. 

So as opposed to your traditional relational database, Don reports to Ansel, Ansel reports to whomever, and it's all very linear and one-to-one, we built a graph database that says, Don reports to Ansel, but he's also part of this team over here, and he's part of this affinity group over here, and you could paint a much more social picture of the individual. And so that was the learning we brought in from Facebook and really the technology, now Meta, to say, let's really rethink relationships in an organizational construct from being linear and hierarchical to being social and multi-dimensional, which is really, it's a cool concept if you think about it. Then, and then related to that, the other thing that I really got jazzed about was bringing in the contractor workforce.

(20:28)

Don: Because if you think about some organizations out there, and I remember talking to the head of HR of one very large tech company, I'm sure you've heard of it, half their company was contractors. So now, but think about it, like every report that you run, every metric that you calculate, everything, like I'm gonna run an engagement survey, like what is it like to work here? But I'm only capturing 50% of the people who actually show up in my door. It hit home for me one day.

We were in New York, so we had location services and there was something that was happening in New York. So we all get this little text, like, you know, it pops up, everybody's phone starts buzzing. It's like, ‘Hey, there's a security incident somewhere in the city. We can see that you're there. Text one, if you're okay, text two, if you need help,’ which is by the way, it's a really cool thing. We built that, we put it in mobile, it's a caring company. 

So everybody's doing this except there's one lady at the end of the table who's looking around like what's going on? And, you know, we've said, ‘well, you know, we've got these wonderful messages from ADP security, A, warning us that there's an issue B, asking if we're okay or if we need help.’ And of course, what's wrong with this picture is, you know, she's a contractor.

What's human capital management? It's human, it's not employee capital management. And so how do we get all the humans in the picture here? And so that was another thing that we really built into the Lyric platform with the graph database that's human centric, that gets all the people, not just the employees. And I would just say some of the foundations of that, both the modeling but also the underlying tech, right? Because I think Facebook was really the ones who pioneered the graph database in terms of people modeling, came from finding a good parallel in the consumer tech.

(22:23)

Ansel: I’m curious if you have any thoughts — I've got to ask this one about AI agents, right? — if we're effectively going to start having people that are not people, but rather bots that are running processes or shared resources. I'm curious, like, how does that fit into how you think about the shape of work? 

Again, this is super early. Obviously there's a lot of claims that, you know, 80% of white collar labor will be disappearing. I really hope not. That's my job too. But at the same time, I'm curious how you've been thinking about like, as you've seen these things firsthand kind of evolve, how does that change organizations? Do you think there's certain parts of organizations that they're going to have to just rethink in terms of how that work is done or how we even interact with these kinds of third-party agents?

(22:50)

Don: Well, 100% they're going to have to rethink how work is done. And some of the startups that I'm working with right now are on both sides of that equation. There's some who I'm working with who are actually doing process mapping and modeling and coming in and helping to redesign it for a more hybrid agent human workforce. And then some of the other startups I'm working with as well as more established companies are like, OK, how would I be on the customer side of that and take advantage of that and redesign my workforce? And I think that's the part we're in right now, the phase we're in, where folks are trying to figure that out.

The next step though will be, you know, well, what happens if my coworker is an agent? What happens if my manager is an agent? Can we prove that they're unbiased? In theory, if we've coded them correctly, they should be unbiased. On the other hand, we know from a variety of experiences that you can have kind of algorithmic creep that introduces biases into an agentic workflow. And who's watching out for this? Are the agents going to police themselves? 

My favorite will be what happens when your boss is an agent, your manager is an agent, and they give you a performance review you don't like. Who do you go complain to? 

(24:08)

Ansel: That's my nightmare. Yeah, the age of age, you know?

(24:32)

Don: So yeah, exactly. So we're in the phase right now where, you know, you mentioned kind of the art too, and I love — there's a figure skating analogy. It's the Olympics right now, right? So there's, get, you know, if you think about a figure skating routine, there's the technical portion of the program and the artistic portion of the program. And we're in the technical portion of the program right now. And everybody's thinking about the technical side of it. 

Okay. What jobs can I, what work can I shift over to agent? It doesn't have to be whole jobs, right? Jobs can be a scary thing versus work. Okay. What work can I shift over to an agent? What work is better suited with the human, what's the right interaction model? That's the hardware side of this next phase. And what you're asking about, and I don't think there's a lot of thought going into it, I don't know that I have the answers, other than just to recognize it's gonna be an issue, it's gonna be that software side of it, or the artistic portion of the program, which is, okay, I work on a team, and my team is me and seven agents.

(25:28)

Ansel: Yeah, it's definitely, I think a lot of people asking these questions, it's going to be a while before we really fully understand the implications of it. And I think, also, it's kind of like the internet too, right? Where you have this kind of upskilling, re-skilling of entire groups of people that are going, I don't need to do this anymore, but now I need to figure out what is higher leverage. And I don't think we've all been programmed — no pun intended — to like think that way yet, because it's just such a big step change.

(25:43)

Don: Sure.

(25:58)

Ansel: So I think it's optimistic.

(25:59)

Don: Yeah, it is.

I mean, we've been through technology adoption cycles. I think humans have adapted every cycle that's come. So, look, I'm a tech bull here, obviously, but I'm also pro-human. I think humans will adapt. But the challenge is the cycle. So we've been through cycles before. The cycles are definitely coming faster.

(26:21)

Ansel: Yeah.

(26:21)

Don: So it puts more stress and challenge on the humans to adapt more quickly.

(26:27)

Ansel: So I wanted to end on one final question, that kind of you teed me up for perfectly because it really is about humans. And so, can you tell me about in your career, was there one person that made a particularly positive impact that you think helped kind of shift the trajectory of kind of where you were to where you are now?

(26:46)

Don: I’ve had the fortune of working for a lot of great folks. I'll just give you a few of the great leaders I had a chance to work for at ADP. I mentioned, you know, so our prior CEO, Carlos Rodriguez, was the most ethical human I've ever seen in the workplace. You know, I just, I mean that sincerely and just to see it through his eyes. 

Like when we built the ADP Marketplace and I talked about being an open marketplace. And that included being open to solutions that competed with some of our own ADP products. And you can imagine there was a lot of internal noise of people like, why are you letting those guys in? Like, you know, I'm the GM of product X at ADP. That's my competitor, you know, and Carlos's view is, well, we're going to do what's right for the customer.

And if it's right for the customer for us to have an integration, you know, to that product then then that's what we're gonna do and you know in the short run it may not be a an easy decision but in the long run, if you do what's right for the customer, it's always gonna pay off in the long run. And I think that's something that really informed my thinking in everything that I've done since. And it took courage for him to kind of stand up to a lot of internal pushback. But at the same time, was like just watching the decision-making process that he went through, it wasn't a long, thought out, highly fraught, wrestled-with decision. It was like, that's better for the customer. That's what we're gonna do.

(28:21)

Ansel: Yeah, no, I love that. I think that's a great takeaway for anyone, whether you're early in your journey to much later in a much larger organization. I think it is like that, ‘what is right for the customer?’ And again, I think, you know, I think even Jeff Bezos had things said about that from the early days of Amazon. And it's, I think it's something that no matter what you're building, it depends on, it really matters who you're building for. 

So, Don, I really appreciate you taking the time, especially for our first kind of inaugural episode. Super excited that we got to chat and excited  to potentially host you in the future.

(28:57)

Don: Yeah, my pleasure. It was fun.