The Washington PostDemocracy Dies in Darkness

Opinion The real value of Paul Manafort’s polling data

Paul Manafort, President Trump's former campaign chairman, leaves the Federal District Court after a hearing in Washington. Manafort is suffering from depression and anxiety and is at times confined to a wheelchair because of gout. That’s according to a court filing from defense lawyers Jan. 8.
Paul Manafort, President Trump's former campaign chairman, leaves the Federal District Court after a hearing in Washington. Manafort is suffering from depression and anxiety and is at times confined to a wheelchair because of gout. That’s according to a court filing from defense lawyers Jan. 8. (Jose Luis Magana/AP)

David Measer is a senior vice president at the advertising agency RPA and an adjunct professor at the University of Southern California’s Annenberg School of Communications.

The goal of advertising, like politics, is to persuade someone somewhere to buy something, whether it’s a product or a candidate. It’s not a coincidence that we both call our persuasive pitches the same thing: a campaign. And at the heart of every campaign is data about markets, the competition and, most of all, the potential customers.

When Paul Manafort allegedly shared “polling data” with Konstantin V. Kilimnik, a Russian businessman and former military interpreter who the FBI says has “ties to a Russian intelligence service ,” many Americans may have shrugged and thought “so what?” To them, polling data is the result of simple surveys that tell us “52 percent of people favor Candidate A,” or “72 percent of Democrats oppose Policy B.” It’s not. It’s far more detailed, and the way advertisers and political campaigns use it has serious consequences.

President Trump denied knowing "anything about" his former campaign manager Paul Manafort sharing polling data with a Russian associate in the 2016 campaign. (Video: The Washington Post)

Polling data is the same contextual business intelligence that big brands use in their marketing activities, including advertising. It’s the raw material we use in the battle to win hearts and minds, and to get people to choose one product or service over another — to vote with their wallets.

If advertisers — and political campaigns — want to persuade you, first they need to really get to know you. For this purpose, marketing departments at big companies spend hundreds of millions of dollars and employ thousands of people in pursuit of data: generating it, managing it, and developing insights and implications from it. It’s very expensive, labor-intensive and takes a lot of time. Anyone who works for a major company knows that big data is the business battle of our time, and data and analytics operations are mission-critical to a company’s profitability and survival. Data is one of the most valuable resources a company has. It’s also one of its most guarded secrets.

The sort of data Manafort shared may have comprised the same types of information that advertisers covet when trying to persuade consumers to buy products. Polling data generally starts with simple demographic information about people, including age, gender, race, household income, education level and geographic location. But those data sets can be cross-referenced with more nuanced — and more valuable — information from other sources, including media-consumption patterns, affinity group affiliations and which consumer segments Americans belong to.

Much of that kind of information is fished from the same stream that’s under heavy scrutiny for dubious privacy practices, including data collected from unknowing consumers with every Facebook like and YouTube view, every mobile location-based GPS cellphone ping, and every cookied website visit. We’ve seen just how devious companies can be when casting their nets into these waters.

Cambridge Analytica manipulated the Facebook platform with a fake app that it used to harvest personal data from up to 87 million users . Each of these individuals became a data stream of actions they took on Facebook: what they liked, watched, shared and bought. In the political context, this created highly detailed profiles of voters with whom a campaign could speak in their language and according to their interests, their fears, their behaviors and their values.

As we learn in fits and starts in our digital age, the data that’s collected on us and from us — the type of data that constitutes polling data — is more detailed, more nuanced and often more invasive than we really know.

We don’t yet know why Manafort allegedly shared this polling data. As the New York Times reported, “He might have hoped that any proof he was managing a winning candidate would help him collect money he claimed to be owed for his work on behalf of the Ukrainian parties.”

But it’s a mistake to treat polling data as mere briefing material; it’s actionable information. Those of us in advertising use it to decide who to target; to position the brands we represent as distinctive from other brands; to develop messaging and ads; and to knock competitors out of their positions in consumers’ minds. We’ve known since 2017 that the Russian disinformation campaign during the 2016 presidential election did the same thing — aiming different posts at people who indicated that they “liked” patriotism or lived in Ferguson, Mo.

Passing on this kind of information gives a partner the ability to reach audiences in a very personalized way. And if that partner is a foreign country intent on influencing voters, exploiting divisions and disrupting elections, the data is priceless. It gives them the tools to get pretty close to the holy grail of marketing: to be in the right place, at the right time, with the right message.

Read more:

The Post’s View: Location data shows where we go — and who we are. And it’s being sold.

Pawan Deshpande: Your Facebook data is still vulnerable. I know because I made it that way.

Alexandra Petri: An appraisal of my data, for the companies who now own it

Steven Hill: How to rein in Big Tech

Letters to the Editor: Perhaps the problem with big data settlements isn’t their size