Jeffry Nimeroff is the CIO of Zeta Global, a data-driven marketing-technology company founded by David A. Steinberg and John Sculley, the former CEO of Apple Computer and Pepsi-Cola, in 2007. The scope of Nimeroff’s role includes overseeing global/enterprise technology; harmonizing technologies across Zeta’s business units; driving product management, execution, and technical delivery, and architecting next-generation products.
CMO.com recently caught up with Nimeroff to hear how CIOs are thinking about marketing data, science vs. art, machine learning, and all things technology-related. Read on for a very interesting and unique view from across the aisle in the C-suite.
CMO.com: So, Jeff, how did you come to be CIO at Zeta Global?
Nimeroff: I started in academia in the early ’90s and actually solved problems that would today be considered machine learning and big data. My goal was finding patterns in data that were related to multimedia problems, such as movie-making. How do you find efficiencies in a particular set of processes? That set the stage for how I moved out of academia into the corporate world and ultimately ended up at Zeta.
I quickly moved through the ’90s and the early 2000s in internet music, including CDNow. I went through their $100-million days and their public offerings, until they wound their way to the Bertelsmann Music Group, one of the largest private companies in the world. I solved what were the R&D problems. So it was data. It was security. It was music deployment.
When I left that world, I worked with Steve Gerber, who’s the president and COO here at Zeta, in a marketing-analytics context. We built a platform that was the precursor to today’s web personalization. It was the ability to market or test marketing messages to learn and then automatically deploy an optimized model for the rest of the population.
I eventually moved through the world of eBay Enterprise, ultimately heading their solutioning division. They had an e-commerce platform for large brands and retailers—Ralph Lauren, the NFL, etc.—and there was a group that took the product and deployed the client. We did six to 12 very large e-commerce builds each year, and that included fulfillment center work and call-center work.
From my perspective, it was great, but it got a little repetitive, so I moved back into the world of marketing, marketing analytics, and data, and made my way to Zeta, with Gerber, David Steinberg, and John Sculley, as a consultant at the end of 2013. Then I found a permanent role in early Q2 of 2014, and I’ve been here ever since.
CMO.com: That’s quite a story. It sounds as if you were a very early “data scientist” or data IT guy who got a sense that marketing was going in that direction. Is that true?
Nimeroff: In my academic days, I was very much a data scientist before we called it that. We had gobs and gobs of data, and the learning, the artificial intelligence, was there to help us find patterns. Where we found patterns, our goal was to effectively compress data where it was very similar and point out the differences in data, then focus computational work only where there were differences in data and, basically, reuse data wherever we found mass similarities.
So you can view it as compressing the work you would do. Instead of, ad nauseam, doing this very large amount of work, we found points where you could reuse earlier work and cut the time greatly and then focus your efforts where there really was a value to be gained in doing all of the heavy lifting.
CMO.com: For those of us who are not that familiar with data compression and the like, could you be a little more specific about this concept?
Nimeroff: Sure. We viewed data as having patterns and distinct differences, and then we harmonized the work effort to the data, which is exactly what you do from a marketing perspective, right? You find patterns in data, and you look to harness what those similarities represent. Then, where you have differences, you put in extra work to capitalize on that.
If I can reuse a marketing message across a population because the data shows me that, that’s great. If I need to use a different message because I’m now seeing differences in the data that would tell me that an earlier message isn’t working, then we look to have the smarts to do that as well. So the difference is really the application of the techniques and seeing a very enjoyable and potentially profitable arena where those techniques can be applied—because they can be applied anywhere.
CMO.com: I wonder how many marketers really understand what you can do with information and data, given your expertise. How have you seen that changing in the marketing world in your work with senior marketers and CMOs?
Nimeroff: That is a great question, and it may very well be the fundamental question. If we go back close to 20 years ago, you had marketing as an art, right? The people who did it were very artful in what they did, and it was a burgeoning science. And where the science and the art collided—meaning data didn’t align with intuition—we had a challenge because it could be viewed as some form of automation potentially impinging on the work that marketers do.
Could this lead to the place where a marketer goes away and is replaced by a complete notion of technical automation? What we found then—and I think it still resonates today—is that there still will be art. The art to marketing is really performed in a different place and at a different level now because the science is being harnessed. So it really is a function of having a marketer who, I believe, understands and is comfortable in leveraging the data or the science as one of their tools and then performing their magic as an adjunct to the data.
CMO.com: So, basically, CMOs and other marketers had better get used to dealing with data?
Nimeroff: Indeed. The simplest example would be intuition, from a marketing perspective, saying something is going to work, with the data saying it’s not quite the way we expect. You could either view that as a clash or you can view that as a partnership. And the marketers that are really taking off are those that say, “I’m OK, potentially, with the data representing a different pattern than my intuition would expect,” as opposed to fighting with it. So the data will say, “Certain messaging at certain times of the day seems counterintuitive, but the data shows that it’s successful.” And a marketer can choose to harness that or they can choose to—I wouldn’t say collide with it, but they can choose to view that as a conflict of their intuition.
Back in the day, it was much more of a challenge. We have marketers now who very much appreciate the science and know how to perform their art using the science as one of their tools. It’s enjoyable, from my perspective, participating that way, and it’s very satisfying to see people get great results when they let the patterns that exist in the data or the learnings within the data help them out.
CMO.com: Creativity has taken two forks from the marketing side. There still is the creativity that would be the conventional “Mad Men” kind of campaign and cleverness, but then there’s also the creativity that surrounds the use of all of the data. That takes a new and different kind of creativity, wouldn’t you say?
Nimeroff: Oh, I completely agree. As we break down the barriers for utilizing channel skipping and channel integration and channel crossing, it opens up a whole new set of avenues for thought leadership on the marketing side. Back in the day, with a direct mail campaign, results would come in 45 to 180 days later, and those results were used to time or sequence the next direct mail campaign.
We now can gather data much more quickly, and we can certainly extend the idea of what could be thought of as a singular campaign to a larger conversation that an entity, a marketing unit, a brand, a retailer, would have with their individual clients. That can be driven by what we’re seeing in the data. And that’s a dimensionality to the creativity that wasn’t really available before.
CMO.com: So the data isn’t the end but the means to the end, which is marketing success?
Nimeroff: Exactly. The data shows the time in which brand X should talk to Jeff, but it also starts showing how you might want to use the different channels that Jeff has expressed interest in to foster that communication, to have that longer term conversation.
The data unveils that nuance as well: Email Jeff first. If he follows up within a prescribed amount of time, continue in whatever that channel is. If Jeff doesn’t follow up, then maybe you want to ping him with an SMS message, or maybe you want to let the internet do some work. And do you want to gently nudge him in a social context or display context?
So if you bought in, on the marketing side, to using the data, the data will start to show you things that now pull you forward, from an omnichannel marketing perspective. It’s really cool.
CMO.com: One of the things I think marketers are interested in, or maybe even worried about, are new technologies, which would include machine learning and, to some extent, AR, VR, and IoT. From your CIO vantage point, where do you see all that landing in the marketing space?
Nimeroff: With internet of things—and we’ll assume we’ve solved the security challenges—it becomes very much an adjunct data source for you to potentially leverage. It may be useful to know, from a marketing perspective, how Jeff interacts with a particular item that would be representative of the IoT. Jeff constantly keeps his house cooler than one might expect. That could ultimately be used by a machine-learning algorithm to represent a pattern, and then that pattern can help us learn things about Jeff or Jeff’s family. So IoT is potentially a great data-collection mechanism, once we’ve gotten over, we’ll say, some of the fledgling tech challenges.
CMO.com: Someone I recently spoke with said that marketers have to be careful not to think of IoT, for instance, as a deliverer of campaigns, as opposed to a collector of data.
Nimeroff: I would agree. My Nest thermostat has quite an interesting small display. It’s probably not best viewed as a delivery channel or a delivery mechanism, although it is very targetable, and you very much know you’re getting the specific individual or the specific household. I don’t know if we push IoT as a distribution mechanism early on, because we still have untapped channels that we can work through if we want to get more ubiquity or more coverage.
CMO.com: It also could be very intrusive and creepy.
Nimeroff: That is the singular issue, right, because I’m not really using it in any broad sense. I might go up to my thermostat, and I’m using it for a very small window. If it had a message waiting for me, that would feel a bit much, whereas interacting with addressable TV boxes and other devices that are more broadly utilized could be viewed as both a data collection and a distribution mechanism, or at least more naturally viewed that way. So that’s where the AR and the VR and any notion of an augmented channel is better as a collection mechanism.
Jeff uses an AR device or VR device or multimedia device, and he consumes in a certain way, so you learn from a data perspective. But it also is naturally viewed as a place where content is presented. The content could be self-selected or it could be targeted.
I would probably be OK in a video game that has advertising aligned with the environment—fake names on fake billboards in a shoot-’em-up game. It may be interesting to view that as the landscape for a real ad and then balance the idea of how much it detracts from the game, versus what ROI you can get on that form of messaging.
CMO.com: Then, of course, there is machine learning and programmatic advertising.
Nimeroff: Correct. Machine learning has had various definitions for a while. If you view machine learning as a particular implementation with AI of moving toward a human-like understanding, the algorithms tend to look for patterns in data, and it very much is an overarching science for, in my opinion, filtering, categorizing, organizing, and making actionable these large sets of data that we have access to.
I don’t really want to be using the full data set very often, if at all, but a machine-learning algorithm that can turn it into the patterns that are very concisely represented, that I can carry around in my pocket, so to speak, or I can deploy is a really powerful use of machine learning or artificial intelligence. I now have all of the predictive capabilities that the data set represented, but I have it in an efficient model that I can use in a computer system and harness.
Machine learning is great in that targeted environment. But it is equally great and more interesting when you try to let it freely learn about certain aspects of data. You can provide a machine-learning algorithm to unstructured data, and it will start to learn things that may or may not be germane to the problem you’re trying to solve, and that’s where scientists would hope you break into the world of artificial intelligence. Human-like intelligence.
I’m not sure if we’re there, but looking through gobs and gobs of marketing information—this message led to this disposition and this success or failure—and gaining learnings that speak to that specific use of data is really powerful for a marketer.
CMO.com: We have an opinion piece on the site titled “Information Is ‘A Thing’ And Should Be Treated Like A Brand,” written by Tom Daly, who was a senior brand marketer for Coca-Cola. His concept is that information really is a thing, that digital interactions replace physical, and you need to understand the attributes of the information and manage them like a brand if you’re going to succeed in the future. Any thoughts on that?
Nimeroff: Yes. At the highest level, from my perspective, an older school computer science guy, we had tons and tons of raw data, and the processing would try to bring out what we would call the next-level information. And then the information, when connected well, provided knowledge. It still was data, but it’s how you processed it. In raw form, it’s just data. If you process it, you learn from it, you get information, and you then build higher-level constructs that relay sets of information to each other, and you start to build knowledge.
So I agree wholeheartedly that our process is to turn data into knowledge. And with that, I would agree that when you have information or knowledge, especially tied to individuals from a brand perspective, prospects, and clients, it’s super powerful in driving all of the metrics that they would be concerned about.
If I have information that have either been specifically self-reported or I gleaned from data, I can turn it into money or build a relationship with you very easily, so I would curate the information. I would make sure it achieves everything we need from a privacy perspective, so the relationship doesn’t feel weird. But that relationship is explicitly managed appropriately, and I would curate the information and treat it like the asset that it is.
CMO.com: What would your words of wisdom be for marketers about working with data and working with the CIOs and CTOs of their companies to further marketing’s success?
Nimeroff: The CMO-CIO relationship should really be a partnership that leverages and accentuates the skills in each world: the CMO doing the work to engage in the forms of marketing that they want to, with the CIO bringing critical-thinking skills and a tech arsenal that should help them by enabling the how.