Adtech’s surveillance ambitions are decades in the making

Debates over internet marketing and surveillance often focus on the unprecedented access companies have to people’s online behavior. But data collection for marketing isn’t new. Companies have been trying to collect data on which ads actually work, and find ways to target optimal customer segments, for generations. The push for numbers grew especially since the mid-20th century, when World War II-era advances in technology and management science permeated the business world.

That history is the subject of the new book Selling the American People: Advertising, Optimization, and the Origins of Adtech by Lee McGuigan, assistant professor in the Hussman School of Journalism and Media at the University of North Carolina, Chapel Hill. In an interview with Fast Company, McGuigan spoke about the parallels between today’s adtech and 20th-century ambitions, and what the future may hold as consumers and regulators grow more privacy-conscious. The interview has been edited for length and clarity.

Fast Company: In your book, you point to this heightened interest in mathematical approaches to advertising and marketing especially after World War II. Why do you think that came about then in particular?

Lee McGuigan: There was this group of technically inclined mathematical experts who had cultivated during World War II efforts, and then spread in business and corporate worlds and wanted to flex their muscles in some sense.

And they made these really impressive-seeming claims about what they can do in terms of delivering new efficiencies and optimizing the sort of decisions that advertisers or agencies needed to make on a regular basis. In many ways that, I think, parallels the arguments that what we think of today as adtech companies have made with regard to the digital world—what they can do with new media that delivered all this data about what people do all the time, and how that can be marshaled for making predictions and speeding up decision-making and investment. 

I think it’s a very similar thing that happened in the middle of the 20th century with these technical experts from operations research and the new field of management science, and particularly a lot of the excitement that was not exclusive to the advertising business, but felt throughout industry with regard to digital computerization.

FC: Do you think the adtech industry today is aware of that history and those parallels?

LM: That’s a great question that I really don’t know the answer to. My sense is no. 

I’m speaking in somewhat general terms. If you looked in detail at different companies, you might see variation on what I’m saying here. But on the whole, I think you see the adtech sector and people looking at it are making claims about new wine that’s really in an old bottle, but they haven’t recognized the oldness of the bottle, if that makes sense.

The discourse used to justify adtech processes today, and to claim how they’re so fantastic and almost magical, looks extremely similar to the claims that people had been making across the second half of the 20th century, in regard to lots of different information technologies and mathematical techniques. And I don’t sense in the proclamations made by contemporary adtech companies much recognition about lineage. 

A lot of these claims, although they came to fruition in some sense over a long course of time, were not immediately successful in any sense. Much of that stuff I look at in this book, even as it tracks the ascendancy of this thing we now call adtech, is really about failures and persistence in the face of repeated failures to make these things work. And many people would argue that adtech today doesn’t work particularly well, that lots of its claims are overblown, a lot of it relies on magical thinking, and, really, a lot of people just don’t know what’s happening under the hood of the systems that they use.

FC: What’s a good example of a claim that’s a bit overblown?

LM: I think one of the most controversial areas probably has to do with what people will call attribution. Fundamentally, what it’s trying to do is make causal claims about the effects of advertising—to attribute marketplace outcomes to some advertising or marketing event. And it’s a very fraught area, because in some sense, it gets to the very substance of what this whole industrial sector is about, which is trying to affect people’s purchasing decisions or consumer behaviors in discernible and productive ways for these companies. 

And yet, there has always been and still remains an enormous anxiety about the possibility that advertising just doesn’t work at all. And so for decades, intermediaries between advertisers and media have all tried to spin up evidence to show how well advertising works. Lots of people doubt that these measurements are particularly good, particularly today as they’re generated not only by direct forms of evidence but through machine learning processes that basically mix direct observation with speculative prospecting from various types of modeling.

FC: In the book, you talk about how it’s hard to distinguish between causing a sale and targeting someone who’s going to make that purchase anyway. Why has that been so difficult to achieve over all these decades, despite collecting more and more data?

LM: There’s an old saw that the best way to make sure people redeem the coupons you send them is to hand your coupons to people as they walk into your store.

It’s really difficult at, like, a philosophy of science level to make a causal claim about this process in which you’re intervening at multiple points, where you don’t really know what would have happened in the absence of your advertising.

I think attribution is such a fascinating thing. And I talk about it a lot in this book. And it is controversial, because it’s just such a difficult problem for the advertising industry. And I don’t think it’s something that they’ll ever really solve, because it’s existential for the business to presume that what they do has a positive impact. And yet, trying to disentangle the influence of their efforts from all the other things to which we could attribute the eventual choices that people make in marketplaces or when they download apps or whatever is virtually impossible.

And so again, it’s this deep anxiety. And in some ways, I think it’s really productive for the industry to have this anxiety because it generates a huge amount of investment for companies that say they can solve this problem. And it’s helped allow technical experts, data brokers, and these companies that do all sorts of surveillance to find a market for their services that trades on that anxiety.

FC: And how did this push for quantifiable audience data and attribution shape recent media, like cable TV and the internet?

LM: I think this is one of the key promises that marketers have spun up around the internet: “Look, we can track all the traffic, we can see what people click on.”

For decades now, there’s been enthusiasm among marketers for what sorts of claims they could make based on click-stream data and all that sort of stuff. Internet-based communication allows for a huge generation of behavioral data just to send people to the appropriate websites and so on. 

And so there’s a potential that you can know exactly who’s doing what, and try and better understand these people and better measure the impact of your advertising. And some of those claims started to get made in the context of cable television, particularly from the ’80s onward, when large cable systems invested in what they call addressable systems, which is to say that everybody’s set top box is connected back to a computer database with subscriber information. 

These addressable systems weren’t designed for the purpose of advertising—it was much more to manage bread-and-butter parts of the cable business like turning on and off different subscriber services without having to actually send a truck to someone’s house. 

But early on in that process, there are people who recognized that if you can do that— if you can tell the difference between each individual potential recipient of messages based on their wired connection to your computer system, we don’t have to stop at governing the tiers of service that they’re getting. You can send them individual messages. And you can send one whole household a certain set of advertisements that’s different from the house right next door to them.

There were lots of impediments in the ’80s and into the ’90s. Even though you could do it technically, it was not at all feasible within the bandwidth of the cable system. But still, you could see people starting to talk about this stuff in a way that I think permeated the eventual thinking that developed advertising on the internet and the way that we know it today.

FC: Do you see any fatigue on the part of advertisers and even tech companies after decades of pursuing this dream? Or are they still pursuing it as vigorously as ever?

LM: Honestly, in this moment, it’s a little hard to tell, because there’s a lot going on, partly in terms of new sets of laws and expectations around data governance.

It’s changing the discussion about adtech. Also, changes to things like interest rates affect how investment flows into companies that are promising to spin attention into gold, without necessarily knowing how that’s going to work. So there are a lot of factors affecting the way that enthusiasm, hype, and money are flowing into the adtech sector. And the potential for a more private orientation is probably the biggest of them.

I think companies are trying to dodge, or at least contain, the force of a privacy imperative through a variety of technical solutions and self-regulatory measures.

So I do think the industry continues to chase a lot of these dreams. And this is kind of one of the main points I’m trying to make in my book: that it’s because of this underlying basis of advertiser-supported media, these twin goals of monetizing attention or behavior and influencing people through advertising and marketing. And so that creates a variety of pressures that don’t always play out in predictable ways but still continue to influence the ways companies try to make money and to control these sorts of systems.

FC: And you speak in the book a bit about the possibility of moving away from these sorts of models. What paths for doing that would make you the most hopeful?

LM: It’s a tricky question because it’s politically fraught, not only in the way that people have come to expect what media systems are supposed to look like but also about the actual nuts and bolts of making decisions about how to govern these systems, outside of some of the market forces that have been relied on for a long time.

But I do think it’s important to confront the fact that the problem runs deeper than just adtech. I’m not convinced that some of the reforms that try to clean up the things that people find obviously distasteful about adtech— lax data control, the invasive surveillance, the profiling, and all the rest of it—that we can fix those things with small reforms, without addressing the broader question of, where did these things come from? What sorts of priorities, ambitions, and imperatives do these developments respond to within the heart of the business?

But of course, there are different layers to how you have to approach this problem. There’s the big question of how do we organize media systems so that they serve public life within democratic society? And then there’s the question of what are we doing tomorrow to make things better than they are today?

And so from that standpoint, I do think that moving the needle on things like privacy will help clean up some of these things, and certainly antitrust regulations and many of the things that we see happening in the U.S. between the Federal Trade Commission and potential options through Congress would possibly make benefits. Of course, it’s a really complicated area. There can be lots of unforeseen consequences in choices made in these spaces.