Say hello View the video

Ads and Insights From Social Interest Data

Geographic Drift: Understanding Location in Social


Marketers value location for targeting their social ads. But identifying where people are isn’t as simple as it seems. Users’ locations are more nuanced than simply “where are they now?” Social ad firms discussing location often leave this out, for simplicity. We don’t blame them: with billions of interest connections being generated every day, social offers a lot of data for analysts to crawl.

Public social data can show us the difference between where people say they are and where they’re actually spending time. For example, you might have added “Austin” to your social profile, but all your latest check-ins and geo-tagged posts are in Louisiana.

This article is part one of a two-part series on understanding location in social.

What is Geographic Drift?

So you may state you’re somewhere but be observed somewhere else. The difference between your stated location and your observed location is your geographic drift.

It’s tempting to think of geographic drift as a measure of honesty: why would people say they’re somewhere they actually aren’t? But people’s conception of location doesn’t stop at cartographic borders.

In fact, stated location can illustrate how people imagine a geographic area. We’ll explore geographic drift and the reasons behind it with a few visualizations.

When a City Is Bigger Than a City

US map

The map above marks out six different metro areas. The circles around each city is a hint at how how widely (or tightly) people conceive of the city.

Take Chicago, for example. Of all social users who say they’re in Chicago, more people were observed outside of Chicago’s blue circle than inside it. Put another way, people who say they’re from Chicago are often not even near Chicago.

By stark contrast, the geographic drift for most who state their location as Jackson, Mississippi is much more tightly confined. The same principle applies to this part of the chart — more people who say they’re in Jackson are observed outside the circle — but the drift is much less.

Measuring Drift: Defining Location in Social

The idea that people aren’t where they say they are is huge. Let’s back up and talk about how we measure geographic drift. How do we know to attach locations to people?

Contrary to what many people might think, there’s no single way that people record and report their locations. In social platforms, it happens a number of ways:

  • People on all social platforms add real or vanity locations to social profile bios
  • Foursquare and Facebook users check in at locations on platforms
  • Twitter users elect to attach GPS location data to their posts
  • Apps use IP addresses to distinguish their users

Stated vs Observed Locations: What’s the difference between where I am and where I say I am?

The 140 Proof Labs team differentiated between stated locations (“I’m from Chicago!”) and observed locations (You’re actually in Indiana!) by starting with a few assumptions. One assumption, for example, is that automated location tagging is less likely to be affected by user bias and thus can be an observed characteristic. So observed location data includes GPS tagging and IP addresses. If it happened without much help from you, you were observed. And stated location data includes profile-reported locations and check-ins. If you clicked or tapped it, you stated it. 

Why Do People Drift?

Major factors affecting drift can include: commuting, clarity, credibility, and social graph distribution.

The map below (right) hows how commuting can increase geographic drift for a population in Los Angeles, a city known for its freeway traffic.

-borIt’s the same story for New York (left): wide availability of transit makes it easy for self-proclaimed New Yorkers to show up well outside of the five boroughs.

Geographic drift for New York and Los Angeles

The chart on the left is mapping people who say they’re in New York (defined with a yellow border). All the colored shapes outside the five boroughs show where those “New Yorkers” actually are. Same for Los Angeles: people who say they’re in Los Angeles are actually observed in a wide distribution across Southern California.

Another explanation for drift around very large cities could also be that people understand that names like “Los Angeles” and “New York” are more recognizable than “Bridgeport” and “Santa Ana”, so people in social are more likely to just say they’re in LA or NY. It’s simpler that way.

Why DON’T People Drift?

Here is an example of two cities that, by contrast, show little geographic drift.

Geographic drift for Napa and Irvine, Ca

People who say they’re in Napa tend to be observed in Napa. Why? You might already know that Napa, California is famous for its wines and attracts tourists year-round. But what about the residents? They’re largely retirees, families, and local business owners. So they don’t have as much reason to leave Napa as a high-octane Manhattan resident or an LA road warrior.

And Irvine, California? It’s another suburban town of about 215,000 people — most of whom are families and over 10% of which are college students. While Irvine-dwellers get around a bit more than the Napalese, they generally stick closely to Irvine.

Stay tuned for part two, in which we examine how diverse interests (defined as personas) are represented regionally.

This study was authored by @williamcotton, @jm3, @NotoriousKRD, and @vnaylon of the 140 Proof Labs team.

Related articles:

March 14, 2014 - 6 months ago

Posted In trendstechnologydataviz