Targeting news and ads to particular recipients on Facebook, Twitter and more, to sway voting 28 Oct 2017, by Issie Lapowsky

NEWS THAT CAMBRIDGE Analytica CEO Alexander Nix approached Wikileaks founder Julian Assange last year to exploit Hillary Clinton’s private emails has amplified questions about Cambridge’s role in President Trump’s 2016 campaign.

Shortly after The Daily Beast reported Nix’s contact with Assange Wednesday, the Trump campaign’s executive director sought to downplay Cambridge’s role. Michael Glassner said in a statement that the Republican National Committee was the campaign’s primary source of voter data. “Any claims that voter data from any other source played a key role in the victory are false,” Michael Glassner wrote. The statement did not respond to reporting in WIRED and elsewhere revealing a close relationship between the Trump campaign and Cambridge staffers. Cambridge did not respond to WIRED’s request for comment.

So, what gives? Such he-said-she-said battles are usually better left to Beltway happy hours. But as Congress and special investigator Robert Mueller turn their spotlights on Cambridge Analytica in their probes into Russian meddling in the 2016 election, it’s essential to get the facts straight about what the firm did—and didn’t—do for the Trump campaign.

Here’s what we know.

Cambridge worked both for the Trump campaign and a Trump-aligned Super PAC. In June 2016, Cambridge sent three staffers, led by chief product officer Matt Oczkowski, to the campaign’s San Antonio office. Oczkowski’s team eventually grew to 13 people, working under Trump digital director Brad Parscale and alongside his staff and outside consultants. According to Parscale, the Cambridge staff provided useful analysis of data about the American electorate. They did not, however, provide the raw data—things like demographic information, contact information, and data about how voters feel about different issues—on which that analysis was done.

That may sound like a small distinction, but it’s a crucial one. Ever since it burst onto the scene of American politics in 2015, Cambridge has trumpeted its massive data trove, boasting 5,000 data points on every American. Cambridge claims to have built extensive personality profiles on every American, which it uses for so-called “psychographic targeting,” based on people’s personality types. It is feared by some, including Hillary Clinton, for conducting a kind of psychological warfare against the American people and dismissed by others as snake oil. Both Parscale and Oczkowski have said repeatedly that the Trump campaign did not use psychographic targeting. Questions also have swirled about how Cambridge accumulated the data. Liberal voters in particular worried that their data had been harvested without their knowledge and used to elect Trump. But according to both Parscale and Oczkowski, the campaign didn’t use Cambridge’s trove of data, opting instead for the RNC’s data file.

The RNC was the voter file of record for the campaign, but we were the intelligence on top of the voter file,” Oczkowski says. “Sometimes the sales pitch can be a bit inflated, and I think people can misconstrue that.”

Parscale describes the firm’s work this way: “As I’ve said multiple times over prior statements, Matt Oczkowski (Cambridge Analytica) and his team created a daily tracker of polling, so that I could see how Trump was doing in key swing states. They provided that to me daily.” Parscale says Cambridge also helped the campaign with what he calls “persuasion online media buying. They also helped us identify potential donors. And they created a visualization tool that showed in each state which areas were most persuadable and what those voters care about.” 

Cambridge Analytica was paid $5.9 million by the Trump campaign, according to Federal Election Commission filings, $5 million of which went toward buying television ads, with the remainder going to pay Oczkowski and his team. But that wasn’t the only work Cambridge did for the campaign. Parscale says Cambridge’s head of digital, Molly Schweikert, managed an advertising budget of roughly $12 million on behalf of Parscale’s firm, Giles-Parscale. It’s a sizable, but still small slice of the $94 million Giles-Parscale was paid in total to purchase the campaign’s ads.

The Cambridge staff helped the campaign identify which voters in the RNC’s data file were most likely to be persuadable, meaning they were undecided but looked likely to swing toward Trump. They also created lists of voters who were most likely to become donors. In August 2016, a Trump aide told me Cambridge was critical to helping the campaign raise $80 million in the prior month, after a primary race that had been largely self-funded by Trump. This was the only period during which Oczkowski’s staff relied on Cambridge’s data, because (in August) the RNC was just beginning to share its data with the Trump team.

Cambridge went on to conduct hundreds of thousands of voter surveys for the Trump campaign to better understand the likely Trump voter and sent a full-time staffer to the New York headquarters, who could relay these findings to senior staff, including Parscale. Based on these surveys, RNC data, data the Trump team collected itself, and commercially available information from data brokers, Oczkowski’s team developed a heat map of the country to pinpoint where Trump should visit to maximize his impact on potentially persuadable voters.

Oczkowski views this as a collaborative effort between his team, the RNC, the campaign, and other vendors, including Deep Root Analytics, which helped the campaign target television ads. “At the end of the day, when candidates win elections, it’s a big team effort,” he says.

The RNC played a very important role in that team. Gary Coby, director of advertising at the RNC, managed the bulk of the campaign’s advertising purchases on Facebook. The campaign famously ran 175,000 variations of the same ad on Facebook the day of the third presidential debate in October 2016, a tactic Coby refers to as “A/B testing on steroids.” The RNC also ran the campaign’s field operations and worked with Parscale to plan get-out-the-vote advertising campaigns on television and online.

What’s also clear, however, is that the Trump campaign seems to have ample motivation to distance itself from Cambridge, a firm whose tactics have sometimes raised questions. Adding to the intrigue is the fact that shadowy billionaire and Trump supporter Robert Mercer is Cambridge’s main financial backer. Former Trump campaign manager and chief strategist to President Trump, Steve Bannon, also held a position on Cambridge’s board. The company itself is an offshoot of the British firm, SCL, which has roots in government and military operations.

Now, Assange’s confirmation that Cambridge’s CEO wanted to join forces against Clinton has renewed suspicions about the company’s business tactics, suspicions that the Trump team would very much like to avoid in the face of ongoing investigations into Russian meddling in the election.

“I had absolutely no understanding any of this was going on, and I was surprised as everybody else when I saw the story” about Nix’s approach to Assange, Oczkowski says. During the campaign, he says his team was walled off from the rest of Cambridge, because the company was also working with a Trump Super PAC. Federal regulations prevent campaigns from coordinating with Super PACs. Of the 13 Cambridge staffers who worked in Trump’s San Antonio office, only four remain at the company.

Still, for some in Congress, the web of connections among Nix, the campaign, and now, Assange, seems too close for comfort. The House Intelligence Committee has acquired Cambridge staffers’ email records, which it is currently analyzing for clues of inappropriate contact with foreign actors trying to meddle in the election. Next week, representatives from Facebook, Twitter, and Google will testify before both the House and Senate intelligence committees and will likely face questions about their interactions with Trump’s digital team and members of Cambridge’s staff.

And investigators will, no doubt, continue to question members of team Trump about Nix’s communication with Assange. The panels will be seeking answers. But, as is often the case when it comes to Cambridge, each answer will likely only lead to more questions.


FACEBOOK IS ENMESHED in several investigations into Russia’s interference in the 2016 election. Details continue to emerge about the 3,000 political ads linked to Russian actors that it sold and ran during the 2016 election cycle. It’s now handed those ads over to congressional investigators, as well as special investigator Robert Mueller, and will join representatives from Twitter and Google parent company Alphabet at a House Intelligence Committee hearing on Nov.1. The constantly evolving story of how the social-networking giant found itself at the center of all of this, and, crucially, what that could mean for President Trump, can easily get confused amid competing headlines around healthcare, hurricanes, and an escalating nuclear standoff with North Korea.

To help, we’re here to walk you through everything we know—and don’t know—about Facebook’s role in the 2016 election and the subsequent investigations. We’ll update this list of questions and answers as we learn more.

What did Facebook give investigators?

In early September, Facebook said it had identified $150,000 of political ads purchased by fake accounts linked to Russia. It attributed about $100,000 of the total, or 3,000 ads, to 470 accounts related to a Russian propaganda group called Internet Research Agency. It found another 2,000 ads worth $50,000 by searching for ads purchased through US internet addresses whose accounts were set to the Russian language. The ads touched on hot-button social issues such as immigration and LGBT rights, as well as content aimed at stoking racial resentment against blacks and Muslims. About 25 percent of the ads geographically targeted certain regions of the United States. The majority of these ads ran in 2015. On Sept. 21, Facebook confirmed it had shared the ads with Mueller’s team and would do the same with congressional investigators.

How many people did the ads reach?

In an Oct. 2 blog post, Facebook said roughly 10 million people saw the ads and that 44 percent of those impressions took place before the election. Considering that fewer than 80,000 votes cost Hillary Clinton the election, that number is intriguing. And researchers have since pointed out that the fake accounts that purchased those ads likely reached even more users through old-fashioned viral posts.

According to Jonathan Albright, research director of the Tow Center for Digital Journalism at Columbia University, organic content posted by just six of the 470 fake accounts Facebook pinpointed may have been shared 340 million times. Albright used the social-media-monitoring app CrowdTangle to analyze traffic to several of the now-deleted accounts that have been made public, including one page called Blacktivists and another called Muslim America. Those same posts received roughly 19.1 million interactions in the form of people clicking the “like” or “share” buttons.

These numbers dwarf the 10 million figure Facebook has provided. That’s because Facebook is only accounting for the reach of ads paid for by a Russia-linked troll farm. The reach of the viral content those same trolls posted is far wider—and tougher to catch. Unlike ads, viral content leaves no money trail.

How did Facebook find these ads?

The only detail Facebook has shared publicly is that it looked for US internet addresses set to the Russian language, then “fanned out” from there, as a Facebook spokesperson told WIRED. That makes it impossible to know whether Facebook has identified all suspect ads or just those the Russians were laziest about hiding.

It’s likely, however, that Facebook’s search has not covered everything. On Sept. 21, during a Facebook Live address, CEO Mark Zuckerberg admitted as much, saying, “We may find more, and if we do, we will continue to work with the government.We know, for instance, that Internet Research Agency, the propaganda group, has officially shut down. But similar firms, including one called Glavset, operate with the same people at the same addresses. The Facebook spokesperson would not discuss whether its investigation would have caught these other shell companies. 

What changes has Facebook announced, and will they make a difference?

During a Facebook Live announcement in September, Zuckerberg outlined how the company plans to overhaul its election-integrity processes. For starters, it will require political advertisers to disclose—on the ads—who paid for the ad. It will also require political advertisers to publicly catalog all the variations of ads that they target to different Facebook audiences. The goal here is to make it easier for the public to see when politicians send different messages to different groups of people. President Trump has been criticized for using so-called “dark posts” to send messages about his proposed border wall to core supporters that conflict with his more public statements. That kind of targeted advertising is par for the course in the internet age, but now, Facebook says, it will ensure that when it’s used in politics, the public will be able to learn more about those messages. Facebook also said it would add 250 people to its election-integrity team to more thoroughly vet who’s buying political ads.

Questions remain: What constitutes a political ad? Are campaigns and super PACs the only ones subject to this disclosure on Facebook? Or will anyone who wants to advertise about a political issue be subject to the same scrutiny? And what about fake news publishers that pay to boost their own articles? Facebook isn’t providing much detail about how it will implement its plan, but answers to those questions are critical to understanding how effective this self-regulation will be.

Could Russians have placed other ads that Facebook hasn’t yet identified?

Absolutely. In the case of the $150,000 in ads, one digital breadcrumb led to the next until Facebook uncovered a cohesive effort by the Internet Research Agency to spread misleading information to US voters. It’s easier to spot such a coordinated campaign than it is to find every ally of Vladimir Putin who might have spent a few thousand dollars to give a fake news story some extra exposure. Facebook sold $27 billion in ads in 2016. Combing through that pile of cash for signs of Russian dirty work is a tremendously complex, if not impossible, task.

Is there anything the government can do to close these loopholes?

Possibly. Democratic senators Mark Warner and Amy Klobuchar are working on a bill that would require political advertisers who purchase at least $10,000 in ads to publish disclaimers right on the ad. It would also require tech platforms with more than 1 million users to publicly track all “electioneering communication” purchased by anyone spending at least $10,000 on the platform. The FEC defines electioneering communication as ads “that refer to a federal candidate, are targeted to voters, and appear within 30 days of a primary or 60 days of a general election.” For now, the term applies only to broadcast and radio ads.

The Federal Election Commission is also reopening comments on rules related to online political ads. In 2011, Facebook asked the FEC to exempt it from rules requiring political advertisers to disclose in an ad who paid for that ad. Facebook argued its ads should be regulated as “small items,” like campaign buttons. The FEC failed to reach a decision on the issue, so Facebook and other platforms have run political ads with no disclosures. Now, the FEC is revisiting the issue.

Facebook, for its part, has warmed to the idea of working with Washington, sending chief operating officer Sheryl Sandberg to the nation’s capital earlier this month to meet with lawmakers and members of the Beltway press. In an interview with Axios’s Mike Allen, Sandberg said, “We want as much help as we can get,” from government officials that can inform Facebook and fellow tech companies about potential threats. The company has also met with people working on the Warner-Klobuchar bill in hopes of helping shape whatever legislation they eventually introduce.

Will these ads ever see the light of day?

It seems likely. The House Intelligence Committee has said it plans to release the ads to the public as soon as it can remove personally identifying information from them. In her interview with Axios, Sandberg confirmed Facebook is helping the House scrub that information. The company plans to release information about who those ads were targeting at the same time.

Were the ads targeting the same audiences as Trump’s campaign?

So far, that is still an open question. In her interview with Axios, Sandberg wouldn’t say, and Facebook has only said that none of the ads targeted specific email addresses. That at least dampens speculation that the Russians were using the Trump campaign’s email list.

Did the Russians use other platforms as well?

Yes. After an initial meeting with congressional investigators, Twitter acknowledged it had found more than 200 accounts related to the Russia-linked Facebook accounts. It has since handed over the list of account names to Congress. Google, meanwhile, has also reportedly found tens of millions of dollars worth of ads on its platforms. Even Pinterest has acknowledged that users saved some of that phony Facebook content to their Pinterest boards, though there’s no evidence the Russians used Pinterest directly. This demonstrates just how difficult it is to calculate the reach of just one fraudulent Facebook post or ad.

How did the Russians decide which Americans to target with the Facebook ads?

The short answer is, we don’t know. There are suspicions that the Russians might have had help from the Trump campaign or its allies. But the Russians may not have needed more than the targeting tools Facebook offers to every advertiser.

Facebook allows any advertiser to upload lists of names or email addresses that it would like to target. In most states, voter files are publicly available for free or for purchase. Advertisers can then design so-called lookalike audiences that have much in common with the original list. They can target ads based on geography, profession, and interest. Facebook knows where you live, the news you read, the posts you like, and what you shop for, along with a million other things about you. The company stitches this information together to make educated guesses about what kind of person you are.

According to Facebook, only 1 percent of the Russia-linked ads targeted users who had visited the advertiser’s website or liked its page.

The Russians also likely used the news as a guide. Reports have said that some of the ads, which discussed Black Lives Matter, were targeted to people in Ferguson, Missouri, and Baltimore, two cities that have become synonymous with racial tensions in America.

Armed with so much readily available information, a Russian operative would hardly need to call in help. That doesn’t mean they didn’t. It just means we have no evidence so far that they did.

What kind of evidence would there be?

One way to find out if the Trump campaign helped Internet Research Agency would be to compare the targeting criteria the campaign used on Facebook to the targeting criteria the Russian propagandists used. If both groups targeted the same audience, that’s worth looking into. Investigators could do the same with any further suspicious accounts Facebook unearths.

What about Cambridge Analytica? What’s their story?

The House investigation has now extended to Cambridge Analytica, President Trump’s data-mining firm during the 2016 election. The Trump team, led by digital director Brad Parscale, worked with Cambridge, as well as the Republican National Committee, to analyze data about the US electorate to guide decisions about where and how to advertise on television and online. That’s not unusual. Hillary Clinton’s campaign tapped similar analyses from a data-analytics firm called BlueLabs, as well as the Democratic National Committee.

What is unusual about Cambridge Analytica is its backstory. The company, which is a US spinoff of UK-based SCL Elections, is financially backed by billionaire financier Robert Mercer, who spends liberally to advance his fiercely conservative views.

Cambridge has also been accused of amassing data from Facebook users—such as what they like on the site and who their friends are—via silly personality quizzes. (Facebook has since closed this privacy gap.) Cambridge combined those results with data from elsewhere to sort people into categories based on their personality types, so advertisers could send them specially tailored messages. Cambridge calls this approach psychographic targeting, as opposed to demographic targeting.

During the election cycle, some Republican operatives outside the Trump campaign accused the company of overselling its technical wizardry. Now, Cambridge’s approach is viewed by some, including Hillary Clinton, as a form of ugly psychological warfare that was unleashed on the American electorate. Parscale and others, however, say the campaign didn’t use Cambridge’s psychographic methodologies.

Cambridge’s parent company, SCL, has been known to use questionable methods in other countries’ elections. In Trinidad, it reportedly staged graffiti to give voters the impression that SCL’s client had the support of Trinidadian youth. And Cambridge is currently being investigated in the UK for the role it may have played in swaying voters to support Brexit. It’s worth noting, though, that the UK has stricter laws around how citizens’ data can be used near elections. The US does not have the same protections.

Is Cambridge involved with these Russian ads on Facebook?

Not as far as we know. While Cambridge helped the Trump campaign target its own advertisements, there’s no evidence so far that Cambridge did the same for any Russians. Whether any connection exists, of course, is a key question both Mueller’s team and Congress will continue to investigate.

Facebook, Google, and Twitter had staff inside the Trump campaign headquarters during the campaign. Is that normal?

Tech companies regularly assign dedicated staffers to political campaigns that advertise on their platforms. Clinton’s campaign also worked closely with Facebook and other tech companies, if not physically side-by-side.

Still, perhaps the least secretive part of the whole affair is the outsized role digital advertising played in the Trump campaign’s strategy. Shortly after the election, Parscale told WIRED, “Facebook and Twitter were the reason we won this thing. Twitter for Mr. Trump, and Facebook for fund-raising.” The Trump campaign ran as many as 50,000 variants of its ads each day on Facebook, tweaking the look and messaging to see which got the most traction. Days after the election, Andrew Bleeker, who ran digital advertising for the Clinton campaign, acknowledged that the Trump team used digital platforms “extremely well.” He said the Trump campaign “spent a higher percentage of their spending on digital than we did.”

Could Facebook have prevented this?

That’s complicated. While the ads that Internet Research Agency purchased were about election issues, they weren’t explicitly about the 2016 election. It’s not clear those would have been considered election ads, even if Facebook, Congress, and the FEC took a stricter approach to election ads online. Plus, the US Supreme Court has given nonprofit groups wide latitude to raise money to influence elections both online and offline without revealing their donors. That’s why it’s called dark money.

Senator Warner recently said Facebook “took down 50,000 accounts in France. I find it hard to believe they’ve only been able to identify 470 accounts in America.” What did he mean, and does he have a point?

Yes and no. In April, Facebook disclosed that it suspended 30,000 accounts, that were spreading fake news in France ahead of elections there. According to Warner’s team, Facebook has since told the senator it suspended at least 50,000 accounts.1 It did not explicitly tie those accounts to Russian actors. Instead, it shut down those accounts after updating its tools for identifying fake accounts, adding flags on accounts that, for instance, repeatedly post the same content or suddenly produce a spike in activity.

That means the French example is not directly comparable to the election ads purchased by accounts that Facebook connected to Russia. Facebook is not asserting that those 470 accounts represent the totality of fake accounts on the platform. They’re merely the accounts Facebook has so far linked to Russia. That said, Warner’s point is well taken: Without more information on how Facebook found those accounts, it’s impossible to know what the company may have missed.

ProPublica recently found that it’s possible to target ads on Facebook to categories of people who identify as “Jew haters” and other anti-Semitic terms. How does that relate to this?

These are distinct issues, but there is some overlap. ProPublica recently reported that it had purchased $30 of ads targeted at users Facebook thought might be interested in terms like “Jew hater,” “how to burn Jews,” and “why Jews ruin the world.” Facebook’s advertising tool had scraped these terms from users’ profiles and turned them into categories advertisers could target. Those categories were a tiny subset of the 2 billion Facebook users, but ProPublica showed that it could assemble such a cohort and send its members targeted ads in 15 minutes. Facebook temporarily changed its ad tool to prevent these user-generated terms from being turned into advertising categories.

The company views this as a separate issue from Russian ads. And yet both incidents point to a lack of oversight of Facebook’s advertising platform. The reason Russians could easily buy political ads to sway US voters is the same reason anyone can target ads to neo-Nazis: Facebook’s advertising systems are largely automated, and anyone can set up an ad campaign with little human oversight from Facebook.

In September, Facebook COO Sheryl Sandberg issued a statement saying Facebook had restored the ability of advertisers to target user-generated terms but had taken measures to weed out the bad ones. It’s also adding additional human oversight to the process of selling ads and is setting up a system through which anyone can report abuses of the ad tool. Something tells us they’re in for an onslaught.

1UPDATED 9/27/2017, 9:57 AM: This story has been updated to include additional information from Senator Warner.

UPDATED 10/17/2017, 6:02 PM: This story has been updated to reflect developments since it was initially published.


Brad Parscale, the Digital Director for the Trump administration, arrives at the Trump Tower in New York on November 17, 2016.

AS SPECULATION MOUNTS about whether President Trump’s digital team helped Russia target voters with fake news during the 2016 election, the campaign’s former digital director Brad Parscale has agreed to an interview with the House Intelligence Committee. But if investigators really want to get to the bottom of how foreign propagandists tried to sway voters, the answer may lie within Facebook’s servers.

Parscale tweeted a statement about his decision to meet with committee members Friday morning, defending the work his San Antonio firm Giles-Parscale did for the campaign. Parscale rejected the notion that his team shared data with Russian operatives to help them target receptive voters. In his statement, Parscale says the campaign “used the exact same digital marketing strategies that are used everyday by corporate America.” And he specifically points out how closely the campaign worked with Facebook, Google, and Twitter.

View image on Twitter

View image on Twitter
The campaign had designated liaisons from both Facebook and Google working inside Parscale’s San Antonio-based office, who were intimately involved in the inner workings of the digital and data team, according to Parscale’s statement. They helped carry out an effort of great scale and sophistication. During the campaign, the Trump campaign ran up to 50,000 variants of its Facebook ads a day, learning which ones resonated best with voters. It also deployed so-called “dark posts,” non-public paid posts that only appear in the News Feeds of the people the advertiser chooses.

Parscale has credited that collaboration with delivering Trump’s victory. “Facebook and Twitter were the reason we won this thing,” Parscale told WIRED shortly after the election. “Twitter for Mr. Trump. And Facebook for fundraising.”

To be sure, there is nothing unusual about this arrangement. Large advertisers working with social media companies are often assigned designated representatives. “Someone from Facebook or Twitter’s ad sales will call you and be your account manager,” says Adam Sharp, who ran Twitter’s government and elections team until December 2016.

Which is why investigators exploring the Russian social-media operation may learn as much from platforms like Facebook as from the Trump campaign. Congress could subpoena the company for data on which entities made large scale ad buys—the kind that can actually help swing an election. Facebook keeps lists of who it extends lines of credit to, though of course, those lists would only reflect the agency doing the advertising and not the many entities that might be funding the ads. Facebook also allows political advertisers to upload their own voter lists for targeting purposes. Investigators could ask the company whether any advertisers used duplicate lists to disseminate pro-Trump or anti-Clinton ads. That could indicate a coordinated effort by some outsider to influence the election on Trump’s behalf, though it’s possible that data is inaccessible because of the way Facebook hashes the information in its system.

(When reached for comment, Facebook said that it has found no evidence of Russian entities buying ads during the election. If true, that would imply that Russians spread their propaganda the old-fashioned way—by creating viral content that Facebook users were compelled to share, without engaging in any demographic targeting.)

The investigation may give Congress the opportunity to shed some light on another opaque digital company, the data analytics firm Cambridge Analytica. Few political vendors have garnered notoriety as quickly as the data-driven campaign firm, backed by the Trump campaign’s top financier, Robert Mercer, which claims to be able to target individuals based on a uniquely detailed database of psychographic information. Some have charged the company with engaging in “psychological warfare,” while others allege that the firm has inflated its abilities.

Now, it appears the House Intelligence Committee is going directly to the source to cut through the hearsay. Parscale was among the first members of the Trump team, joining the campaign before it was even officially announced to the public. The 6-foot 7-inch digital marketer had worked with the Trump family since 2011, building websites for Trump Winery and the Eric Trump Foundation, before he received a call asking him to create a website for the real estate mogul and reality star’s presidential campaign. “When I got the phone call I was elated because of how much I respect this family,” Parscale told WIRED last year.

Parscale’s role at the campaign grew, from building the website and creating social media content for the would-be Tweeter-in-chief to managing $250 million in online fundraising. Over the course of the election, Giles-Parscale took in some $90 million, the vast majority of which went toward buying Facebook ads for the campaign. As November neared, Parscale had evolved into something of a deputy campaign manager, working alongside Trump’s son-in-law and current senior adviser Jared Kushner to strategize what was always a digital-first campaign.

Parscale’s testimony may help illuminate an often misunderstood aspect of political campaigning. Democrats, including Hillary Clinton and Sen. Mark Warner, have asserted that if Russians wanted to target fabricated news stories at American voters who were on the fence about who to vote for, they would have needed inside information about those voters and where to reach them. But digital operatives on both sides of the aisle have countered such conjectures, asserting that targeting news stories and ads on platforms like Facebook and Twitter is far simpler and much more difficult to detect than some outside the industry might expect. Between lax campaign finance laws that don’t require digital platforms to disclose who’s paying for ads and social media platforms’ ability to target people based on age, gender, location, and interests, finding the right voters to bombard with propaganda isn’t all that difficult.

Parscale, for one, maintains he’s “unaware of any Russian involvement in the digital and data operations.” That may be. But that wouldn’t preclude other members of Trump’s inner circle who had access to the digital and data team’s insights—including Kushner—from transferring that knowledge to a Russian operative during one of the undisclosed meetings he had with them during the campaign.

As a former staffer put it: “When it comes to the kids and close family, I’m not really sure what the hell is going on.”

Those family members will likely be subpoenaed. But investigators don’t have to rely only on them for the answers. All they have to do is ask Facebook.

The British number-crunchers and PhDs started working with down-ballot candidates in the US two years ago, and came to prominence this year touting its “psychographic targeting.” The approach supposedly builds on traditional ad targeting metrics like demographics (age, race, income) and behavior (voting, spending, online habits) by adding a person’s psychological profile.

To generate models, Cambridge, which employs data scientists and psychologists, draws on personality surveys it has conducted by telephone, email, and social media since 2013. It uses those samples to predict the personality traits of voters—traits like, say, neuroticism. Candidates can use those findings to tailor their message to a specific audience.

Cambridge CEO Alexander Nix claims his company’s data models “absolutely, indefatigably” raised Cruz’s poll numbers. “Every decision of the campaign—where to spend the money, who to target, how to target them, what to speak to them about, what channels of communications to use, what messages to send—was all driven by our data,” says Nix, who was among those included in WIRED’s Next List of business leaders.

Those who have worked with Cambridge say the company’s data minds are strong. “I think their data scientists are some of the best, most talented I’ve ever worked with,” says Chris Wilson, who was director of research, analytics, and digital strategy for the Cruz campaign.