Why Facebook says its ads can reach more people in the US than live there
For one, Facebook counts residents and travelers.
Facebook may have messed up on its math again. Or maybe not.
Facebook claims that its ads can reach more people in the US than the number of people who live in the US. Pivotal Research analyst Brian Wieser picked up on the discrepancy after checking potential audience estimates provided by Facebook’s self-serve ad-buying tool and comparing Facebook’s estimates with estimates calculated last year by the US Census Bureau.
According to Wieser’s digging — and confirmed by my own checking of Facebook’s and the US Census Bureau’s numbers — Facebook estimates that its ads (running across Facebook, Instagram and its Audience Network ad network of third-party publishers) can reach 10 million more 18- to 24-year-olds and 15 million more 25- to 34-year-olds in the US than live in the US, per Census data.
Obvious question: Why would Facebook tell advertisers it can show their ads to 25 million more people in the US than actually live here? Maybe not-so-obvious answer: It’s not.
Comparing apples and apple sauce
Facebook’s audience estimates cannot be compared apples-to-apples with the US Census Bureau’s resident estimates. Each estimate is calculated differently and vulnerable to separate reporting errors.
Facebook’s figures include people whose Facebook profiles claim they live in the US, as well as people who happen to be in the US because they are traveling and let Facebook track their location, as detailed on Facebook’s site. The inclusion of residents and travelers would also explain why Snapchat’s self-serve ad-buying tool claims its ads can reach up to 56.4 million 18- to 34-year-olds in the US, or 75 percent of the Census Bureau’s estimate for that population, when ComScore estimates the app reaches only 60 percent of that demographic; Snapchat Ad Manager also pegs the high end of the app’s reach among 18- to 24-year-olds to slightly exceed the total population of 18- to 24-year-old residents in the US.
Meanwhile, the US Census Bureau’s figures only include the people who live in the US and rely on people raising their hands in a governmental roll call from which its estimates are extrapolated (Here’s the full methodology).
“Reach estimations are based on a number of factors, including Facebook user behaviors, user demographics, and location data from devices. They are designed to estimate how many people in a given area are eligible to see an ad a business might run. They are not designed to match population or census estimates,” said a Facebook spokesperson in an emailed statement. The spokesperson also said that the estimates provided by Facebook Ad Manager are only projections of how many people an ad could reach, not how many it will actually reach, and that the estimates are not used to charge advertisers for their campaigns.
Problems with self-reported stats
Given that both Facebook’s numbers and the Census Bureau’s numbers are estimates, either could be — and more likely, is — wrong. Both rely on people providing accurate information; therefore, both are susceptible to people fudging the facts.
Eleven-year-olds may want to be on Facebook like their older siblings, so instead of waiting until they turn 13, they register accounts claiming to be 15 years old to skirt Facebook’s minimum age requirement. Adults may also misreport their ages on Facebook, disguising who they actually are to evade Big Brother or to appeal to people in a different age group. And people may either elect not to fill out a Census form or inadvertently fill in the wrong information, leading to incorrect counts like the number of same-sex married couples.
Facebook’s past mistakes impart distrust
Over the past year, Facebook has been plagued by measurement problems that have eroded marketers’ trust in its math and mounded evidence of its fallibility. Since September 2016, Facebook has disclosed 10 separate measurement errors, including one in May 2017 that led to Facebook mischarging advertisers. As a result, any anomaly in Facebook’s numbers is more likely to be immediately interpreted as inaccurate instead of inconsistent for a reason.