×

What Twitter thinks it knows about you is often wrong

  • Newly available details show users the aggregated third-party data.
  • But some marketers are skeptical of the accuracy of targeted advertising.

CNBC's Wilfred Frost is retired, and Landon Dowdy makes over half a million running her own business. Those are some of the conclusions drawn by Twitter's newly available demographic inferences, which are based on behaviors outside of the social media site.

In May, Twitter made some updates to their user data controls and privacy settings. The changes were designed to give users more access to their information, and more control over how it's used, the company said in a blog post. In essence, the changes allowed users to see how Twitter and its advertisers had been categorizing them based on their social media usage and other, often offline, behaviors.

You can find out and edit what Twitter knows about you in the "Your Twitter Data" section of your account's settings section. It's broken down into two parts: what Twitter gleans about you from your use of the social media site and what it's "partners" know about you. That's based on what you do off the site, like making online purchases attached to your email address, and offline behavior like in-store purchases.

It's that third-party data that users often find speculative or at worst, flat-out wrong. CNBC's Eric Chemi checked in with some of the network's on-air talent to see where Twitter's partners fell short on their inferences.

Driven by ad dollars

Twitter having your age or income wrong may not seem like a big deal, but it means that advertisers who want to appeal to a specific demographic are paying big bucks to serve ads to exactly the wrong people. With this change, Twitter hopes users will adjust the settings so they're more accurate — "self-declared data," in ad-speak — thus improving their ad-targeting.

"By surfacing this information, we're going above and beyond what other companies provide and give people the ability to review and edit their data, or opt out in a single click," said Dan Jackson, a Twitter representative. "This gives people on Twitter industry-leading control over their information and how it's used."

One of the reasons behind Twitter's new policy is to be more open with users, the company told CNBC, but experts say the change also benefits the company financially.

"It's no secret that they're struggling," said Jordan Cohen, CMO of Fluent, a data-driven marketing firm. Twitter's ad revenue was $473.8 million in the first quarter of 2017, the company said in filings. That's down 11 percent from the previous year.

Worldwide digital ad spending is expected to hit $224 billion in 2017, according to research firm eMarketer. Facebook and Google will take in about half of that, and Twitter is likely to get just $2.3 billion, about 1 percent. Twitter, and its investors, are anxious to increase that share, according to the Associated Press.

At the time time, Twitter's daily active users in the first quarter were up 14 percent over the previous year. In a May 26 research note, analysts from BTIG said it would take a few quarters for the surge in users and engagement to translate into advertising dollars.

Bigger isn't smarter

If you look through your account's privacy settings, you'll likely notice that the "interests from Twitter" section is pretty accurate. That's based largely on your Twitter usage, and is often correct, according to CNBC's analysis.

Then there's the "interests from partners" section, where Twitter shows you what it's partners — companies like Datalogix, Acxiom and Epsilon — have gathered about you based on those other behaviors. Using that data, users are sorted into groups, or "audiences," for advertisers.

It's important to note too that Twitter isn't the only company that categorizes users like that: Both Facebook and Google also allow users to see what categories they use to target them with ads.

The trouble for advertisers is that's often inaccurate, as CNBC's informal study showed. Although Twitter's representative declined to comment on the errors in the "interests from partners" section, marketers have questioned the quality of third-party data for years.

An experiment from 2014 by media-buying agency Mediasmith showed that many data-targeting tools were "useless," according to the Wall Street Journal. And even with expanded data availability and increasingly sophisticated ways to slice that data, it hasn't gotten much better, some experts say.

"Bigger data isn't necessarily smarter data," Cohen said. "There's been a deluge of big data and most marketers are drowning in it."