A primer for researchers on use of smartphone location data offers a glimpse of how we’re watched
As we go about our private lives with smartphones in hand, most of us regularly send out rather public notices of our whereabouts. A roughly $25 billion industry collects these anonymous location signals, packages them into coherent data and sells it, mainly to companies involved in investing, retailing, marketing and real estate. At least some of your own mundane movements are quite likely to be included in such transactions.
Increasingly, economists and social scientists, too, are using smartphone location data in hardcore research, the kind that goes into peer-reviewed publications, and then into corporate strategic plans and public policy debates. Data on where a single device shows up throughout a day is just a much more believable indicator of a person’s movements than all but the rarest of source material researchers had before. There’s a whole world of possible research redos by replacing once-required theoretical constructs with actual data on where, and alongside whom, people actually visit.
A team of economists recently wrote a primer for peers who want to use smartphone location data in their own projects, particularly as they relate to business strategy. As described here, the authors explain how the data is generated and how to get it. They offer examples of studies that used it and preview other research topics that probably will soon. They detail technical aspects of preparing the data and limitations of its answers, and they warn of its quirks that could lead to embarrassments.
“The potential of these data to expand our knowledge in strategy is largely untapped,” reads the working paper by University of Virginia’s Young Hou, UCLA Anderson’s Christopher Poliquin and Mariko Sakakibara and University of Miami’s Marco Testoni. “This paper serves as a primer on using mobility data for research in strategy and highlights several promising directions for future exploration.”
If you’re a researcher looking for ideas, browse through Dewey Data’s list of 800-plus published papers that tapped the company for data. They don’t all use smartphone location data, but short descriptions and color coding are helpful for finding those that do. (To start, look for those using Advan, Veraset or Safegraph.) Although the primer authors focused on business strategy topics, guidance throughout the paper, and repeated here, applies to other fields as well.
Not an academic researcher? That’s OK. For the rest of us, the paper offers a revealing look at what our phones are telling the world about ourselves.
When the Product Is Free…
If you ever allow location data on your smartphone, even only “while using the app,” some portion of your travels already exists in these massive databases. Cellphone carriers have always had a grainier, less accurate version of what’s available today. Location intelligence companies now buy that anonymized carrier data and supplement it with more precise location data from apps.
Fitness trackers, navigation and weather apps — apps that require continuous location data for functionality — can send back intricate pictures of how one moves across a day. Away in a car at Oak Street and 22nd Avenue; into the office building 20 minutes later; quick trips to a coffee shop, a pharmacy, an elementary school; eventually back to a place it spends the night. Other apps that need location permission at least sometimes pinpoint locations, less often but just as accurately.
Location intelligence services own some location-gathering apps, such as GasBuddy, CityGuide and coupon apps like Flipp. They buy more data off other apps and rarely disclose which ones.
Not all apps sell their location data. Not Google Maps, for example. Yes, for The Weather Channel and probably your dating app. With a lot of free apps, feeding the developer your movements is required payment for the service. The fine print of their privacy policies may, or may not, clarify if they do. The location aggregation industry is expected to grow to $63.8 billion by 2032.
Foursquare, once best known for its social app allowing users to find nearby friends and entertainment, has at times collected location data from some 50 million smartphones. Advan (includes SafeGraph now) is very popular for databases that show foot traffic to specific points of interest. (Like, your particular grocery store.) Veraset’s raw mobility data offers far more complete pictures of where each phone goes throughout a day.
Recently, Apple and Google made it easier for smartphone users to opt out of location tracking on websites and in apps. Those changes likely made aggregators’ jobs harder, but each still gets data from billions of smartphone location pings.
Requesting macro data gives you traffic to some point of interest, at some points in time, that you pick. Companies like Advan track foot traffic to millions of places, especially individual stores. It’s good, for example, to study events that affect traffic to specific retail stores, competition between restaurants and corporate allocation of resources to various specific locations. It’s relatively easy to request and to acquire, with Dewey Data acting as a gateway for academics.
It usually takes micro data (disaggregated) to study human interactions. That gives you an anonymous ID per phone, as well as latitude, longitude and a timestamp for each ping. Micro data requires a lot more work to get desired insights, including sometimes tricky geocoding and safeguarding against errors. Large datasets can include tens of billions of pings for each month of data.
Quantitative Data for Gut Feelings
A lot of economists study human interactions within and between organizations, such as how departments share information, innovators get inspiration from others or customers react to corporate’s bad news. But there’s a big hurdle in this work: Scientists rarely get to see the actual interactions. Instead, studies are often based on things that might facilitate interactions, such as shared public transportation, formal alliances, a popular watering hole or close physical proximity of offices.
Smartphone data, some scientists already found, helps tamp down skepticism around location proxies. A coffee shop situated between two research institutions may foster collaboration between their inventors, but it doesn’t show that co-authors of any patent ever set foot there. Smartphone locations can’t prove coffee-assisted collaboration either, but it might show multiple occasions when the two inventors spent an hour at the coffee shop at the same time.
These databases are possibly the next best thing to actually witnessing, for example, hundreds or even millions of subjects wait in line to vote, meet with their acquisition targets, distance themselves from politically different relatives, avoid mingling with those outside their economic class or socialize with potential patent partners.
- An early study tracked some 10 million Americans to Thanksgiving gatherings in 2016 and found data to back something something long suspected: American politics were fraying families. Dinners including guests likely to have divergent political views were nearly half an hour shorter on average than those where everyone supported the same party, according to findings by UCLA Anderson’s M. Keith Chen and Washington State University’s Ryne Rohla.
- Do America’s various social classes rub shoulders much? Naval Postgraduate School’s Maxim Massenkoff and MIT’s Nathan Wilmers addressed the question by tracking devices from specific census tracts to specific points of interest, like restaurants, parks and stores. Their working paper finds that the rich mostly stick with the rich and the poor with the poor, but classes in between mingled more. They also uncover some potential explanations, such as locations of dollar stores versus luxury vendors. They find libraries and parks better at drawing diversity than museums and historical sites. Fun finding: People cross class lines to eat at Olive Garden.
- Chen and MIT’s David Atkin and Anton Popov used smartphone location data to uncover the R&D effects of serendipitous encounters — situations researchers have found virtually impossible to study in the lab or replicate after the fact. In a 2022 working paper, they tracked 218 million encounters between seemingly unrelated workers in Silicon Valley as they visited coffee shops, elementary schools, bars and dozens of other out-of-office locations. They calculate that these encounters boosted the number of Silicon Valley patents by about 8%.
- One study, published in Nature Human Behavior, watched device movements to some 3.9 million stores, amenities and other points of interest across U.S. cities to find out how many trips people were making within a 15-minute walk from home. (The “15-minute city” is a goal for places that want to be called “walkable.”) They found about 12% of those trips done within the time frame.
- Taking location data into business strategy research, a 2022 study measured the effects of face-to-face meetings between executives at an acquiring company and their intended takeover targets. While it’s widely believed that such meetings pay off for buyers, the field of research was constrained by problems of proving they actually occurred. With location data, Testoni, Sakakibara and Chen could see when (likely) pertinent executives were in the same place together and for how long. They found that returns to acquirers generally rose with the number of meetings before a merger was announced.
- In recent years, tweeting out support for, or rejection of, a newsy social movement has been a marketing strategy for some firms. Historically, researchers looking at customer reactions to these formerly verboten topics might rely on imprecise or quite delayed measures, such as customer surveys or experiments meant to identify changes in purchase intentions. Smartphone location data allowed a closer look. In the days after some CEOs went before Congress to push for stricter gun laws, Hou and Poliquin watched, via device pings, foot traffic fall off at the companies’ stores in politically conservative areas while staying steady at others. Within two weeks, it had all returned to normal, according to findings published in Strategic Management Journal.
On a technical note, a seemingly mundane finding in a new study based on smartphone location data could change the way researchers measure travel in multiple fields. People tend to visit multiple places when they leave home, and when one stop disappears — perhaps the work commute — other stops also go away, Boston University’s Yuhei Miyauchi, Hitotsubashi University’s Kentara Nakajima and Princeton University’s Stephen J. Redding demonstrate in a working paper.
Moreover, they find that the commonly used method of measuring travel as single destination trips leads to inaccurate results. They offer a fix for that: a tractable travel itinerary model.
Rethinking Business Strategy Research
The primer authors stress that business strategy is particularly ripe for new studies using smartphone locations. To assess performance of, say, a new store, outside researchers historically relied on end-of-quarter data or similar reports from public companies. But quarterly data for specific business units or stores is often messy — difficult to disentangle one brand or single store from a larger corporate whole, and subject to human manipulations and mistakes — as well as limited and old. It’s also hard to come by for privately held firms.
With smartphone location data, researchers can count individuals walking into that mom-and-pop shop days after it opens and simultaneously watch traffic at competitors. They can also build once-elusive demographic profiles on those customers with aggregator-provided stats or by matching it with other databases, like Census data.
To illustrate the nuanced insights these databases can provide, the primer authors whipped up an event study around a highly publicized fight between fast food chains Popeyes and Chick-Fil-A for chicken sandwich customers. Using Advan location data, their findings suggest that the Popeyes’ sandwich launch in 2019 immediately led to more foot traffic at both chains’ locations. Popeyes gained younger, richer and more college-educated customers after the launch, according to their findings.
With the event study, the authors include a way for researchers to practice with smartphone location data before diving in with a new project. The experiment details how to replicate their analysis, including event study data and R code. You can read here how other researchers also used smartphone location data to analyze a different aspect of the same Popeyes’ launch.
Potential Pitfalls, for Everyone
Privacy breaches top the list of potentially devastating consequences of using smartphone data in research, for both academics and individual device owners.
Datasets for research are anonymized long before the researchers see them, and re-identifying supposedly anonymous location data is illegal in the U.S. and many other nations. It’s also explicitly banned by prestigious universities, many academic and professional organizations like the American Economic Association and major funders of research.
The more likely privacy breach would involve studies published in ways that allow someone else to identify device owners. For example, it’s not uncommon for a study to peg a phone’s base location; aka, where it spends long hours outside of work. This is most likely where the device owner sleeps. If the data is published without hiding longitude/latitude coordinates, it wouldn’t take a coding expert to figure out who lived there.
Without care, pairing smartphone location data with traditional databases can lead to result-skewing mistakes. For example, Advan-assigned ticker symbols, for various reasons, do not always represent the company that trades with that ticker today. It’s an easy overlook when researchers match location databases with those from Compustat, a long-trusted source of financial information on public companies.
There’s also room for error when researchers need to geocode addresses and define the perimeters of buildings of interest. For example, people moving around outside of a building sometimes appear to be inside. Buildings that host multiple companies — or say, a deli, a parking garage and offices — pose particular issues.
Of course, the primer itself takes a deeper dive into these and other potential hazards around working with smartphone location data, just as it goes into far more detail than this article on culling its sources and manipulating it for results. By demystifying the data and processes, the authors lay out best practices for a very new tool they expect to revolutionize a lot of research. Remember, the general public hates being tracked. They’re offering some practical information for navigating the minefield.
Featured Faculty
-
Christopher Poliquin
Assistant Professor of Strategy
-
Mariko Sakakibara
Sanford and Betty Sigoloff Chair in Corporate Renewal; Professor of Strategy
About the Research
Hou, Y. and Poliquin, C. Sakakibara, M., & Testoni, M. (2025). Using Smartphone Location Data for Strategy Research. Strategy science, 10(4), 281-299.
Chen, K.M., & Rohla, R. (2018). The Effect of Partisan and Political Advertising on Close Family Ties. Science. DOI: 10.1126/science.aaq1433
Massenkoff, M. & Wilmers, N. (2025). Rubbing Shoulders: Class Segregation in Daily Activities. Journal of Public Economics, 244, 105335.
Akin, D., Chen, K.M., & Popov, A. (2022). The Returns to Face-To-Face Interactions: Knowledge Spillovers in Silicon Valley (No. w30147). National Bureau of Economic Research.
Abbiasov, T., Heine, C., Ratti, C., Sabouri, S., Salazar M., Santiat, P., & Glaeser, E.L. (2024). The 15-Minute City Quantified Using Human Mobility Data. Nature Human Behaviour 8(3), 445–455. https://doi.org/10.1038/s41562-023-01770-y
Testoni, M., Sakakibara, M., & Chen, M.K. (2022). Face-to-Face Interactions and the Returns to Acquisitions: Evidence from Smartphone Geolocation Data. Strategic Management Journal, 43(13): 2669-2702. DOI: 10.1002/smj.3435
Hou, Y., & Poliquin, C. (2025). CEO Activism and Political Mobilization. Journal of Business Ethics 200(2), 269–285. https://doi.org/10.1007/s10551-024-05901-x
Miyauchi, Y., Nakajima, K., & Redding, S. J. (2025). The economics of spatial mobility: Theory and evidence using smartphone data. The Quarterly Journal of Economics, 140(4), 2507-2570.
Ashutosh, B., Harsha, K., & Norris, B., (2024) The Clashing on Social Media: Exploring the Imoact of Twitter Banter Among Competing Fast Food Brands. The Journal of Social Media in Society Spring 2024, Vol. 13, No. 1.
Pyrgelis, A., Kourtellis, N., Leontiadis, I., Serrà, J., & Soriente, C. (2018, December). There goes Wally: Anonymously sharing your location gives you away. In 2018 IEEE International Conference on Big Data (Big Data) (pp. 1218-1227). IEEE., doi: 10.1109/BigData.2018.8622184.
