The use of alternative data sets for informing stock picks or long-term investments in industries such as retail, airlines, real estate and auto is growing. In the case of mobile location data, investors and others use it as a valid indicator of consumer interest or even sales. But without regular system updates, these indicators may be less than valid. In fact, the information used by money managers to make extremely important business decisions could be dated or even faulty because some location data providers may not regularly revise data baselines to account for factors such as artificial foot traffic growth and historical shifts in mobile usage patterns.
Like marketers and advertisers, more and more investors, fund managers and consultants today include location data showing consumer foot traffic to business locations such as restaurants or car dealerships in their data and analytics tool chest. In fact, an April report from capital markets consultancy Opimas predicts the market for mobile device location data in the money management industry will grow 40 percent annually to $250 million by 2020.
Despite all that excitement around mobile location data, people are unaware of the risks that lurk when providers of that information fail to update their data models.
Why is it so important to recalibrate location data baselines? At the core of the issue is the huge growth in the amount of information generated by our mobile interactions, prompted by greater adoption of mobile devices, more robust device capabilities, decreasing costs of data plans, and an explosion of data-heavy websites and apps. In November 2017, Ericsson reported that mobile traffic in North America will reach 48 gigabytes per month per smartphone by the end of 2023, up from an estimated 7.1 GB in 2017.
Protecting that $250 million prediction
All that mobile growth results in tons of additional individual data points, which over time reflect changes in usage patterns. The increase in data points can be misattributed as growth in foot traffic because some location data providers do not regularly recalibrate their data models. But if the industry is to hit that $250 million mark, that should change. Revising data models should be common practice to ensure the location information used to gauge foot traffic in store and restaurant locations actually reflects what’s happening in real life.
Before listing several reasons why below, I’ll put this in the context of an investor scenario. Let’s say you’re a mutual fund manager evaluating the performance of Chipotle Mexican Grill and Qdoba-owner Jack in the Box to determine whether they are solid long-term stock investments. Of course, you’d consider recent earnings reports from both companies, and delve into articles about new menu item launches, retail location openings and promotions.
However, if you also include alternative data sets in your evaluation of these fast-casual Mexican restaurant rivals, it likely includes mobile location data indicating how many people visited certain locations and possibly information such as the amount of time patrons spent in those outlets, as well as related demographic data. You might set up a search in a software platform for foot traffic data during a given period for all Chipotle and Qdoba locations. But if your location data provider has not updated its stale models, you might be putting key business decisions at risk due to misleading information.
More mobile data means more reasons to update location models
Because so much of the location data gathered to gauge foot traffic at restaurants is gleaned via mobile apps, there are several variables that can affect the validity of measurement models:
- More mobile devices: While the same number of people may be visiting restaurant locations, a larger percentage of those customers could own smartphones today than did in the past. We need to factor that into data models to prevent mistaking more phones for more people.
- More app partners: Most location data firms obtain their data through relationships with app publishers and app networks. If data firms have added apps to their list of app publisher partners, this change must be accounted for when recalibrating the models. Are location data firms taking in more data because there actually is more usage among the same number of devices represented by the same number of apps as six months ago, or could that data influx be the result of new partnerships with additional app providers? Either way, it affects how we estimate the number of people visiting a location.
- Shifting mobile app usage patterns: The way in which people interact with certain apps evolves and fluctuates all the time, altering the app usage pattern landscape. Sometimes app publishers change features, more users adopt app categories such as ride-sharing apps, or people get excited about a new gaming app one month and have stopped playing it a few months later. If you recently added the Lyft app to your phone, or haven’t played Pokémon Go lately, you know what I mean.
- Geographic and seasonal usage distinctions: People engage with apps differently depending on where they live, too. If you’re on the east coast in the springtime when weather can fluctuate hour-to-hour, you might open a weather app several times a day. People in New Mexico expecting sun may not open their weather apps for days. When fewer people open apps used to pick up on location pings, it doesn’t necessarily mean that fewer people were physically present in a location.
Each of these factors has an impact on the validity of the location data used by investors, marketers, municipal planners and others to make crucial business decisions that affect how budgets are allocated, what stocks are added to a portfolio, or whether a neighborhood gets a new restaurant that could generate jobs.
When location data providers make a point of revising data model baselines regularly, the value of mobile location data has a better chance of reaching its full potential.