EP 55: What Good Looks Like: The Growing Role of Benchmarks in Private Markets

Josh Green: Welcome to Sustainable Intelligence, where we discuss all things sustainability for the private markets. I’m Josh Green, Co-CEO and Co-Founder of Novata, and today we’re joined by Caitlin Pentifallo, Novata’s current Head of Product for AI Data Products and former Head of Regulatory Strategy and Sustainability. Caitlin is here to discuss Novata’s latest benchmark release. Caitlin, last time you were here on the podcast, you talked to us about proxy data and benchmarks by extension. Now you’re here to discuss some exciting news, Novata’s latest benchmark release, which is the largest and most comprehensive data set we’ve produced to date.

Caitlin, welcome back to Sustainable Intelligence.

Caitlin Pentifallo: Thanks for having me, Josh. Thanks for the intro.

Josh: So let’s jump in. This year’s benchmark release is the largest Novata has produced to date. Tell us about it. What changed behind the scenes to make this scale possible?

Caitlin: I think the biggest thing that changes here is our data. So our data is the fastest-growing part of our platform. The number of data requesters that we have on platform, the number of companies providing responses to their investors has grown. And all of this means that our benchmark pool can only grow and grow.

So as new and additional metrics come on, whether they’re from a regulation or they’re from a framework, all of that is growing at a rapid clip, and so we’re seeing more benchmarks than ever, so meaning that those numbers are getting over our minimum thresholds. And with that, we’re able to enhance this set, and it’s able to become larger.

So really that data growth is what’s driving our benchmark growth.

Josh: And can we talk a little bit about when we talk about the number of benchmarks on the platform? So let me see if I’ve got this right. So there’s the number of metrics that we have benchmarks for, that number is growing because we are growing the set of people on the platform. And then for those metrics for which we have benchmarks, we can slice the data in more and more fine-grain ways if we have more data. Is that right?

Caitlin: Yes, exactly. So thankfully, I have spent some time in the stats. Seventy-four percent of our benchmarks grew their sample size. So even benchmarks that are carrying over from one session to the next, three-quarters of them, those pools are simply growing. So that’s great news because it means that we’re seeing the depth in all of those areas that we’ve come to expect.

And then in terms of the number of metrics, so even though you might hear some, you know, really high numbers like, oh, there’s tens of thousands of benchmarks on the platform, at its core, we right now have 333 metrics that have been benchmarked at least one time, and that is actually a pretty big growth from what we had last year, which was 239.

So we’re seeing that growth, nearly a hundred, so a hundred new metrics to be benchmarked, all of that kind of coming in. And then as you mentioned, with those underlying data points jumping up as well, that means that we can take additional cuts and look at everything with an additional lens. So even if we’re saying we’re generating a universal benchmark, there still isn’t ever just one universal benchmark.

We can specify that for size. We can specify it by region. And so what we’re talking about is kind of just like these benchmarks in multitudes. That means that we can filter through all of those benchmarks and help our users find the very best and most relevant benchmarks suited to their context.

Josh: Okay, so Caitlin, 333 metrics for which we have benchmarks. You and I are both Mets fans.

Caitlin: Mm-hmm.

Josh: Would we be interested in a hitter that was batting .333, adding that hitter to our lineup?

Caitlin: Would love to have one. Would love to have just one. 333 would be incredible. But you know, people are getting focused on the, on base percentage right now and the OPS. So I don’t know if the true batting percentage is what it used to carry. But yeah, it just is showing a lot of growth, and it’s showing a lot of use of our metric catalog.

And that metric library, even in the time, you know, I’ve been here three years now, it’s grown from being about 490 metrics to today over 5000. And people might say, “Oh, wow, proliferation, this is crazy. We shouldn’t have metrics like this.” But it’s actually really reflective of where the sector is, where sustainability is going.

We are just on the cusp of ESRS metrics being released for their refinement within CSRD. Things are changing. You’re going to see things like, you know, various frameworks, whether it’s related to life sciences or venture capital. These things are growing in popularity. So in the way that we’ve come to rely on baseball statistics to tell us really quickly, “Hey, how is this person doing?” These metrics, to me, tell us a lot about the quality of our library and the quality and breadth of the data that’s coming into Novata.

Josh: Wow. Okay. I thought I was going to throw you off your game by taking us down the baseball route, and you just totally turned it around on me. I respect that.

Caitlin: I can do sports all day.

Josh: So let me throw you a softball. Why does this matter? Why do benchmarks matter? It’s, you know, we get excited at Novata about the number of benchmarks, but what’s the importance? What’s the relevance of them in the private markets context?

Caitlin: Yeah. Yeah. A great, great question, and I’m glad one that we’ll spend time on. I think, you know, since the dawn of, you know, business, of capitalism, people have always wanted to put a finger in the air. How am I doing? How are the winds blowing? And if you are an investor in the private market, you wanna know what is it gonna look like relative to my exit?

What is it gonna look like to my performance? How does it impact my financial bottom line or, you know, whatever other bottom lines you wanna look at. And that has always been really, really important to core private market enterprises. How am I doing and what does good look like? And benchmarks really allow us to try to answer both of those questions kind of simultaneously.

When I look at something like our Boolean benchmark, so strictly metrics when you’re talking about, you know, a yes or no. Do you have a code of conduct policy? To the average observer, that’s not, “Okay, great. Of course we have one. Like, we’re a functioning business, yes.” A lot of people do see that question and answer it, and we know that because the answer rate for a metric like that is near 100%.

Everyone knows that they have this or not, they’re answering it, they’re moving on from it. But where we really start to learn things about where people are at maturity-wise as a sector you really can learn from that. So when we compare things like employee code of conduct, health and safety across food and beverage compared to financial services, compared to professional services, compared to oil, mining, and gas, you start to see really different patterns, and it’s in the patterns that I think we can really, really learn a lot.

So if we know that that Boolean metric, that percentage is now shifting up to 90%, it tells us a lot about where that growth is going and where all those companies are. So as an investor, you can now look at that and say, “Okay, actually, for every non-alcoholic, you know, beverage company that’s out there, whatever kind of soda, seltzer brand there is, all of them are having this right now.”

That sets my expectation as an investor that if I’m going to be investing in a business like this, I wanna know what everybody’s doing, what is that expectation, and then I wanna take the opportunity to set that for myself. So now I know as I go on to purchase, you know, equity stakes and whatever it might be in those private companies, this is what that expectation should be for how those companies perform and what we expect them to have.

So it’s one thing to kind of run through a governance checklist and be like, “Yeah, every good company should have a board, and you should be doing these things, and this is how we expect governance to roll out.” But it’s another to see our benchmarks because what your benchmarks are reflecting are your peers.

So if we know that this is kind of the trend and this is the mode of travel and this is where people are going, it gives us a lot of clues overall in maturity and then can also start to think about what that likely success looks like. So we can start to see what the impact of, you know, emissions reductions activities are looking like.

Are SBTi targets making a difference? I’m talking about some really big questions that a lot of people are asking themselves, but we do really have the building blocks here for how we can eventually answer and really put good data behind all of those really important and pressing questions.

Josh: How are these used in practice? Are people using benchmarks to brag, to spot opportunities for improvement, or something else entirely?

Caitlin: Yeah, I think a little bit of both. I think it’s great to know, okay, for companies of my size and type, this is what emissions are looking like. So it’s really, really different when you compare the emissions of Novata to the emissions of Google. So I don’t want to compare against both of these. I don’t want to put myself in one big technology sector, but it is really useful for me to take that normalized value, for me to look at it, you know, for companies between 100 and 500, and for companies that are based in North America, to really get a sense of like, am I in the ballpark of what we can expect a company like mine to be doing?

And so I do think that kind of method of comparison, I think that’s probably what first comes to mind when people talk about benchmarks. And then as an individual company, if you were to look at, okay, here’s all the metrics I was asked, here’s the benchmark, what everyone is doing there, and now here’s my performance, that starts to tell you a really powerful story of maybe you’re right kind of in that, you know, middle 50%, those middle two quartiles and you’re happy with that performance.

But maybe you do see opportunities to jump up, or maybe you see areas where you’re actually really leading in a way that surprised you. So I think that’s something. There’s so many potential opportunities for how people can apply this on a company level and also at a portfolio level that it’s really just kind of a wealth of information to start to draw comparisons from.

Josh: Okay, let’s talk numbers. 49,000 benchmarks, 1.24 million underlying data points. That’s a lot. How do we maintain quality, reliability with that scale?

Caitlin: Yeah. A good question and one that I ask myself every day. So there’s a couple of different points where data quality and data reliability come into play. And I can tell you a little bit about some of the tools that we’re doing to ensure this because I think that’s what makes, you know, doing this for 1.24 million data points possible is our tooling.

So back behind the scenes this year, we made a really significant investment in the way that our benchmarks are produced. Before the advent of AI, this was a really, really robust data science process that was heavily, heavily driven by individuals with lots of expertise. We really relied on these individuals, as we should.

We’re a young data science company, and really making sure that those foundations are in place was core and critical to the mission. As we got really good at that process, so in those first kind of three years of benchmarking, we really started to nail it down that this year, for the first time, we were actually able to move that entire process, that entire service into Novata’s architecture and structure.

So no longer are we ever exporting and cleaning and running kind of manual reviews. All of this is happening not with the help of AI. It’s happening with code that was largely generated with the help of AI, but there’s no AI cleaning, but all of our processes are now cemented in these very repeatable, very strong from a coding perspective, steps that make it possible to generate a very large set of benchmarks, 49,000 benchmarks off of these data points.

So I think that’s something I know, you know, our users aren’t necessarily going to see us. You know, no one wants to see, you know, how the sausage gets made. But it’s a really, really robust process that we’ve now been able to automate significantly. And when you’re able to automate a process significantly like that, we can embed all of those quality controls that you would typically do alongside that and really make sure that it’s been strengthened overall and that it’s repeatable and that it’s defensible.

And so if we’re putting benchmarks out there that we can’t point to the exact sources, the exact data points, why different points were excluded or included as part of our cleaning process, then we’re gonna start to lose the plot on what makes our benchmarks so reliable, so defensible, and the private market’s leading collection of benchmarks is because of all of this kind of robustness that goes into our quality controls and that goes into our data cleaning and collection process.

Josh: So appreciate the point that nobody wants to see how the sausage is made, but I’m curious if you go back in time a little bit, were there mixed feelings about this type of investment and making the shift to this more code-driven process? Take us behind the scenes a little bit.

Caitlin: Yeah. I mean, talk about having a problem and then needing to solve it. 

It was just this kind of great all-hands-on-deck activity that I’m sad our clients don’t see on a day-to-day, like how committed our engineers were to bringing this to life because it really like, it gets me so excited because it means that our potential capabilities are really kind of unleashed right now.

We can create custom benchmarks. We can regenerate a set of benchmarks with about forty-five minutes notice, whereas we used to do this over three months. And I think we can’t understate right now how fast technology is moving, how much people are looking to rely on AI and just rely on something else.

But at the core, what we really wound up doing was reinforcing what the core value proposition of Novata is, which is our strength in our data. So we took that entire process, and we made sure it was ironclad, and we made sure it was ironclad to a point where it could all be containerized and moved into our own architecture.

So for me, that’s something so exciting that not everyone will get the benefits of. However, we all get the benefits of our time back because we can now do this faster and more effectively than we were ever able to run it before. And so for me, that’s massive because it means we can do so much looking forward.

Even kind of a small example of this is, you know, by virtue of our dataset growing and us being able to query the data in more ways than we have historically, we are now able to generate for the very first time our first set of Middle East benchmarks. And so now we can kind of set up these capabilities where we’re constantly looking for Latin America.

Maybe we’re looking for South America. Like we can kind of set these custom parameters and go after our data in this way and really interrogate it in a way that there’s just so much capability there that I’m really excited about. So for anyone that was ever thinking about what does a GICS benchmark looks like relative to a SICS benchmark, or can I create, you know, a custom geographic grouping?

Those are all now capabilities we are knocking on the door of. And so the value of that process to us as a business is something that can’t be overstated.

Josh: I mean, I was literally in a customer meeting where they asked if we could do GICS benchmarks instead of SICS benchmarks.

Caitlin: Every day. Yeah.

Josh: Yeah amazing. Okay. So obviously we love our benchmarks, and we get excited about the work that we’re doing around them. Let’s get to the customer-facing piece of this and the insights.

So, have you played around enough with the data to be able to give us some insights that pop out from the data itself?

Caitlin: Yeah. I think when I look at our data set as a whole, there’s a lot of different things happening. So people are always like, “Wow, so many benchmarks. How, what, how is this going on? How are you doing it?” We do have so many more metrics that are getting benchmarked for the first time, and there’s stories under there.

And so a couple of those stories. Big headline would be quantitative environmental data is where we’ve seen the biggest gain. So when we look at ones that we really, you know, know and love here Scope 2, total GHG emissions, Scope 1, Scope 3, the growth over there is really serious, and it’s also serious in terms of the different combinations people are looking at.

So we’ve always kind of carried total GHG as a benchmark, but now we have people doing various combinations of Scope 1 and 2, Scope 1 and 2, looking at Scope 3 with a materiality lens. So we can definitely tell that there’s a lot more being asked of our users as it relates to those quantitative environmental metrics, which we love to see.

In terms of governance, I think a really cool story happening here is kind of moving away from policies and then into enforcement outcomes. So if I were to put, you know, my hat on of previous lives where I was really focused on evaluation, especially in the social and governance sectors, we would say, okay, there’s lots of process indicators, there’s outcomes, there’s performance, there’s all different ways that we can classify and look at these metrics.

When you’re talking about input indicators, that’s the very, very earliest entry in terms of your theory of change. Do you have this? Do you have a board? Do you have a board that meets four times a year? Like we’re talking very, very basic inputs. We’re seeing a movement away from those metrics, so we’re seeing a decrease in terms of how many people are asking that, what does that look like?

It’s still being asked. We still see it there, but it’s looking more at enforcement. So now we’re talking about what actually were the fines that came out of those proceedings. So the things that a board is designed to prevent, like corruption and bad governance and, you know, bad practices, misconduct, how are we seeing that play out?

We know you have this policy. I don’t need to ask you for the third time in a row, but I am seeing that there’s maybe a reduction in fines, that there aren’t these violations coming up, that you don’t have a higher incidence of material ESG incidents or money laundering or a number of data breaches that affect PII.

So there’s all these things that, you know, we see kind of a movement away from do you have this? And we can all understand, you know, yes, do you have a policy is very helpful in terms of like, is this something you care about? But now we’re transitioning into what happened? So now that I know you have that, do you see any kind of outcome on it?

And that to me shows growth in this area because anyone can ask, do you have all these things? Great. That’s not really what matters. It matters is like, is it having an effect? And that effect is a reduce in fines, a reduce in emissions, whatever it might be. We are starting to see more of a transition into those particular areas and really driving kind of that conversation around what happened.

Josh: Interesting.

So all this sounds like the sustainability ecosystem is maturing. Is that a reasonable read of what you’re describing?

Caitlin: Josh, you and I are optimists, so yes, it is maturing. I think the biggest takeaway if you were to say one thing benchmarks tell me this year, it would be that this continues to matter to people. No matter what is happening with CARB’s legislation, no matter how many more delays or weird things can happen in Europe with Omnibus, this still matters, and it is still driving business decisions.

We don’t see full scale that people don’t care about these metrics anymore or that they’re watering them down. We’re starting to see the opposite. They’re asking more granular metrics. They’re asking more frameworks that are more involved. We’ve seen a massive proliferation in terms of the number of VSME-related metrics that are coming into frameworks.

And so there’s a lot going on that all of these stories, optimist or not, tell me that this matters, and it is driving business conversations.

Josh: Let’s go back into the weeds for a second in terms of the Novata experience that our customers have. So my understanding is we now, when we’re surfacing the benchmarks, provide a “best-fit” benchmark automatically. What does that mean, and how do we do that?

Caitlin: Yeah. So great question. And it’s something that really serves two of our new products from the first half of 2026 really, really well, and that’s actually what led us to make this change in the UI for how people experience benchmarks in their report summary pages today. So to tell you a little bit about these two projects I’m alluding to, one is the Novata Emissions Estimator, and one is our proxy service.

So as we were going through these two things, and we kind of were, like, at the drawing board of how do we help people estimate what their emissions might be when that first source primary collected data isn’t there, isn’t enough, we don’t have it, how can we help people get there? And that led to kind of long conversations around, okay, if I’m looking at this series of benchmarks, and scope one, two, and three, by the way, are our largest collection of benchmarks.

We’ve got it in the thousands in terms of kind of the granularity you can get to. We quickly realized it was gonna be really overwhelming if we just kind of pointed people in a general direction, and instead, we’d have to be really prescriptive around the best benchmark that they should use. So when you go through this, what’s happening in the scenes, it happens in a millisecond, but what’s happening behind the scenes is that a waterfall algorithm is going through that’s helping to identify what the very best version of that benchmark is for you.

So if you told me, “I wanna find out my scope one emissions,” what is happening in our back end is that it’s looking at your industry. Can I match your industry? If yes, I’ll use it. If I don’t have that, I’ll look to your subsector. If I don’t have that, I’ll move to your sector. Then I’ll look at your geography.

If you’re in North America and I have North America, I will surface that. So now I’m at, you know, the sector level in North America. Then I will go on to look at your FTE to say, “Are you a small company you know, a hundred to a hundred FTE? Are you larger than five hundred FTE?” And we’ll try again to match you.

What this means is, if you were asking for, say, a metric that’s a little bit obscure or we didn’t have a ton of data coming in on it, maybe there’s only a universal benchmark available. It does happen. Not in scope one, two, and three, we don’t produce universals for scope one, two, and three. But we would also score that for you.

So we would tell you, “Hey, best I can do is a universal, and it’s gonna have a score of sixteen,” because it’s gone through all those four areas and it fell to the bottom on each one of the waterfall algorithm, and now you’re at a sixteen. It means it’s the most generalizable, the most universally applicable.

It applies to everybody no matter what. Whereas if you’re getting a score of a one, that means I matched on your sector, your subsector, your industry, your size, and your location. And I’ve done all of that to take a normalized value that’s going to serve as kind of a median emissions intensity, and then we’re gonna multiply that by your revenue.

So for us, instead of just being like, “Oh, you are wondering what tech companies that are small in Europe are kind of saying for their emissions? Here’s the number.” We’re not saying that. We’re saying, “Of all companies like you, when I take the median intensity by millions of dollars USD, it’s telling me this.”

So when I take that number and multiply it by your revenue, I now have something that’s specific to you in two ways. One, it’s matching your context and your profile on all of those points, and two, it’s being controlled by your revenue. So this we do a little bit differently, you know, if you’re a pre-revenue company, if your revenue is lower than a million dollars, a couple other, you know, levers that we’ll pull on that front.

But by and large, what we’re talking about is now giving you a wealth of access, a peer-based benchmarking system. We’re telling you, “Use this one, that’s the best one for you.” And then we’re running that multiplication to tell you what your estimate is. And that’s something that we’ve found really, really resonates with clients when they are, say, trying to create proxies for an SFDR disclosure for an LP.

Maybe they’re in credit, and they don’t have kind of direct connection to people, but they do have a good estimate on what industry they’re in, where they’re located, and what their revenue probably is. And using only those factors, we can kind of really lean into the proxies on that front. So a very long way of saying these methodologies that we really refined for Novata Emissions Estimator and for our proxy offering have really come in handy because now instead of just kind of throwing out there, you know, 49,000 benchmarks, we can say, “Actually, use this one. It’s the best one in this way, and here’s why.” 

That really kind of takes out, I think, a lot of the noise, a lot of the guesswork. It eliminates potential for greenwashing should you try to, you know, choose a different benchmark. And so I think for us it’s really about making it super, super contextually specific and really, really relevant to those companies that are now using those benchmarks.

Josh: Okay, let’s lift out of benchmarks and talk more generally about sustainability. And if I can, I’d love to ask you a career question as well. So sustainability, you know, it’s been through some ups and downs over the last few years. Where do you see it headed over the next few years?

Caitlin: Hmm. Well, anyone with any access to any kind of news is seeing kind of the collision of a bunch of things on a couple of fronts. One is the mass proliferation of AI data center usage and the demands that that places on resources. If you said, “Caitlin, I have, you know, $10 million. I want you to start a new business on something that people are gonna love and they’re gonna need.” I’d be like, “Hmm, it’s gonna be on collecting metrics related to AI data centers, how they’re being used in communities, and how they’re either being combated, remediated, or attended to.”

I think we can’t underestimate kind of the proliferation of these data centers in communities and the impact it’s likely to have. So with that entire sentence in context I just said, I never once said sustainability or ESG, but the entire conversation is around sustainability and ESG. Sustainability has kind of had, you know, since its existence in the mid-’80s, a crisis of what it really meant and for whom, and I think that’s really, really gonna come into focus in the next two to three years.

We’re going to see heat waves. I’m sitting here with my window curtain down on a lovely Thursday afternoon because it’s too hot for me to be near the window. We’ve got people, you know, sweating, you know, in our London offices, in Berlin, in places that were not designed to withstand these temperatures.

Josh: Right.

Caitlin: So when we look at sustainability in different ways, be it through communities, be it through resources, be it through just longevity and people’s ability to live in their communities, these are all things that are gonna rise to the forefront of the sustainability debate, whether or not they wanna call it sustainability.

Anyone could call it anything. It’s still going to be the interaction of people and their environment and how they live within that environment. So until we permanently, you know, blast off into Mars and sustainability becomes an afterthought here on Earth that’s the direction that I think it’s going.

And not to catastrophize, but I just think all of these things are really gonna come into focus. And for early adopters that have been keeping tabs on their data and doing their best and trying to make those decisions every day that we as consumers can often feel like this doesn’t really matter in the long run, I think we’re gonna see that in very, very sharp relief in the next two to five years.

Josh: Want to ask advice for folks who are just at the beginning of their careers. I think you’ve done an extraordinary set of things just in the time I’ve known you at Novata but clearly you’ve had a remarkable run doing a wide variety of things, and I’m curious if you have any advice for people who are just getting started in sustainability either in terms of, you know, advice about how to weather the ups and downs or advice on how to think about specialization versus generalization. Just what comes to mind?

Caitlin: Yeah. Hmm. That’s so good, and I love the trajectory that I’ve had at Novata, and I think that trajectory was really driven by curiosity. So that’s what I would tell any young person right now is that if you’re curious about something, learn everything you can about it. Where I started my career, not where I am now but it, you know, just came from pulling on threads of this is something really interesting.

I think there’s something here, or I think I want to learn about that, or I actually am interested, let me figure that out. The more that you can do that, you have to worry less about your job being eliminated or something’s not gonna exist in a few years because that kind of core expertise and passion can’t be replaced by anything else.

So committing yourself to something and learning about it and like really spending time with it is something that I could talk to you all day on a number of topics that have nothing to do with my job here. But really staying curious I think is what led me to so many opportunities in my career, and I think it’s gonna be, you know, what eventually separates humans from the computers is gonna be our passions and our ability to connect.

So I would say don’t lose that. Keep, keep being curious.

Josh: Keep being curious. I love it. Caitlin, you are such a key driver of the Novata business. Thank you for all the work that you’ve done, the many roles that you’ve played at the company, and for walking us through benchmarks and the learnings from the benchmarks today. It’s been a fabulous conversation. Appreciate it. 

Until next time, let’s keep building sustainable intelligence together. As always, you can find out more about Novata at Novata.com.