Economists, hedge funds, and health officials are using this startup's data to understand the pandemic

In the century since the Spanish flu gripped the globe, the extent of the disease’s economic destruction remains an enigma.

Good data about the influenza of 1918 simply does not exist. The pandemic took hold as World War I barreled toward its conclusion, making the two world-shaking events nearly impossible to disentangle.

The limitations have left historical analyses wanting. “The scope of research on the economic effects of the 1918 influenza pandemic is scant at best,” said a report from the St. Louis Federal Reserve in 2007, which relied on old newspaper clippings for most of its study.

Today, as the novel coronavirus and a related business slowdown exact a huge economic toll, the situation has changed. Modern technology, like smartphones and their location-tracking, are providing unprecedented details about a pandemic’s impact, down to the ebb and flow of foot traffic at a given store.

What’s your step count?

The precision and granularity of the data collected today, from the individual consumer to countrywide, is unmatched by any other point in human history. It’s as though the entire economy now wears a Fitbit at all times.

Chart showing foot traffic patterns across various industries in the U.S. “Normal” behavior is represented by “1” on the y-axis. (Courtesy of SafeGraph)

In mid-March, brick-and-mortar businesses lost the majority of their shoppers. Through April, foot traffic nationally fell 80% at bars, 70% at restaurants and hotels, and nearly 60% at cafés, according to SafeGraph, a startup that collates geolocation data collected from apps on 45 million devices in the U.S.

Some industries weathered the downturn better than others; a couple even received a boost, at least initially. In mid-March, activity at general merchandise stores leaped 30% and 40% at supermarkets, as people stockpiled supplies, according to SafeGraph’s data.

The comeback has already begun. By May, foot traffic at supermarkets and general merchandise stores had ticked back up to normal levels.

Other industries are recovering too. By June, counter-service restaurants were almost back to normal, operating at 95% of usual capacity. By that time, bars were seeing 60% of their usual customer inflows, sit-down restaurants were receiving 70%, and cafés 80%.

The recovery for other industries looks grimmer. Airports had only a quarter of their usual visitors, while attendance at movie theaters remained sparse at a fifth of normal activity.

Voting with your feet

“We predict the past,” says Auren Hoffman, SafeGraph’s cofounder and CEO, during a recent online presentation.

The company provides customers and research partners with data about what has happened, in other words, and leaves the analysis up to the others. “We let customers predict the future,” he says.

Chart showing foot traffic patterns across a selection of retail brands in the U.S. (Courtesy of SafeGraph)

On its publicly available dashboard of movement patterns, SafeGraph shows that, during the pandemic, some businesses bucked generally dismal trends.

Foot traffic at home improvement giant Home Depot surged as much as 60% between mid-March and the end of May, perhaps as homebound consumers began redecoration projects. After initially falling 40% between mid-March and mid-April, foot traffic at Target recovered more than 60% through mid-May.

Other businesses face more persistent drop-offs. Foot traffic to Starbucks stores fell by half between mid-March and mid-April, and fewer people—only three-quarters of the usual clientele—appear to be lingering there today for macchiatos and lattes.

McDonald’s, on the other hand, appeared to be all but back to normal as of the end of May.

Location, location, location

SafeGraph says it buys its geolocation information from well-known consumer apps involving health, safety, navigation, and transportation. The resulting data sets are anonymized and aggregated to protect people’s privacy, says Nick Singh, the company’s marketing lead.

While Singh declines to disclose the exact sources of SafeGraph’s data, he claims the company does not use sketchy sources that hoover up information without people’s knowledge. The origins are, generally, well-known apps that users have granted permission to monitor their whereabouts.

Privacy advocates will find such pervasive tracking troubling, of course. Many people do not realize how far and wide their geolocation travels as it changes hands through little-regulated marketplaces. Experts may call into question the ability of data scientists to fully anonymize their data sets too.

SafeGraph’s customers include hedge funds, which use the data to inform their investments, and hotels and other chains, which use them to determine where to open new retail locations. Of the 11,000 U.S. hedge funds, there are “probably only 50 hedge funds that buy alternative data” of the sort SafeGraph sells, CEO Hoffman says, but he believes more businesses will become customers as they build up internal data science expertise and as better data-crunching tools become available.

SafeGraph provides data to academic researchers, policymakers, and journalists too. The Centers for Disease Control and Prevention used the data to determine the efficacy of stay-at-home orders. And the Brookings Institution, a Washington, D.C., think tank, used the data to show social distancing differences within black and white neighborhoods.

While even Hoffman admits his company’s data is imperfect—cleaning up raw data is a science all in itself—the information is better than anything researchers had access to in the past. “Every day the data is improving from the day before,” Hoffman says.

Feeding the Fed

In the depths of the pandemic, the Federal Reserve of Dallas acknowledged that the delays associated with gathering usual economic data, which tend to be published quarterly, “are crippling.” It cited “an acute need for real-time measurement” to scrutinize the wreckage of—and hoped for rebound from—COVID-19.

The Dallas Fed has been using SafeGraph’s data in its own economic modeling and forecasting. The resulting charts constitute “a key metric in forming our assessment of economic conditions and the outlook for future activity,” it says.

Since May the central bank branch has been publishing a continuously updated “mobility and engagement index,” which measures how much people are moving around. The gradual easing of government-mandated restrictions and the onset of summer shows how behavior is slowly but surely beginning to return to normal.

It’s a far cry from using century-old periodical archives.