Where Does the Data Come From?

Placer’s location data originates from tens of millions of mobile devices across the U.S. The aggregated location data is sourced from hundreds of partner mobile applications that collect the data from devices “in the background”, even when the apps are not actively used. Then, the data is aggregated to form a large and representative panel of the U.S. population.

A few facts about Placer’s panel data:

• The data is sourced from mobile devices distributed across the 50 U.S. states and represents approximately 7% of the entire U.S. population. The data currently excludes U.S territories (i.e., Puerto Rico, U.S. Virgin Islands, Samoa, Guam, and Northern Mariana Islands).
• The data is sourced from both iOS (55%) and Android (45%) mobile operating systems.
• The data is available starting January 1, 2017.
• The data originates from hundreds of mobile apps across diverse categories. Partnering with many different and diverse apps helps us reduce biases in the data.

Placer applies a wide set of processes and methods to ensure the quality and accuracy of the data.

How Does the Panel Data Account for Visits?

Placer’s observed panel data is aggregated and processed to generate insights into any location, using an approach that is similar to a statistical survey. By applying advanced statistical techniques, we are able to project and generalize the findings from the panel in order to make inferences about the broader population –– thereby providing valuable insights for decision-making and understanding trends at a larger scale. 

Placer runs a proprietary extrapolation algorithm to produce an estimation of visits to a given location, as well as a range of metrics that provide visibility into visitor & consumer behavior.

For instance: Placer observed 74.8K panel device visits to a grocery store in Jackson, MI during a certain time period, which resulted in an extrapolated estimation of 2.4M visits overall.

Visitation and panel visitation metrics of a Meijer location in Jackson, MI

Placer displays the number of panel visits as well as the estimated number of visits for each POI examined, in order to explain the origins of foot traffic estimations and the strength of the underlying location data.

Does the Data Accurately Represent the Population?

Yes. To make sure that the data from Placer's panel accurately reflects the entire U.S. population, the data goes through a process called debiasing, which helps remove any biases and ensures the metrics and insights are reliable and representative.

Similar to other consumer panels, Placer’s panel data may have certain biases or limitations that need to be reduced or eliminated. This is needed since Placer receives data from multiple mobile apps, and each app has its own systematic bias. For example, some apps may be used by mobile devices that are concentrated in specific geographies; others might be installed only on Android devices; and some apps are used by devices with certain visitation patterns.

More specifically, in order to account for a range of panel biases and variations, Placer applies a multifactor model using proprietary unsupervised machine learning methods.
The main variations that are addressed during this process are: 

1. Geo-Debiasing. Accounting for demographic, geographic, and other variances in coverage at a local level.
2. Population & Data Source. Resolving differences in source apps and types of devices as representatives of the larger U.S. population.
3. Panel & Sample. Accommodation for our continuously growing device panel.
4. User Behavior. Accounting for differences in user behavior, such as the frequency of visitation to the same location.

Learn more about Placer's visitation data accuracy and POI data accuracy

Is Mobile Location Data More Accurate Than Manual People Counters or POS Transactional Data?

First, mobile location data is typically not used to merely measure traffic counts, as it provides broader and deeper insights than other solutions such as manual counters, cameras, in-store sensors, and POS data. The wide range of location insights includes visitor behavior, cross-shopping, brand affinity, True Trade Areas, and more.

Second, the level of accuracy differs across traffic-counting solutions:

• POS/ Transactional data can be helpful in quantifying the number of purchasers at a retail location. However, it would usually miss out on the additional visitors that visited the store but did not make a purchase.
• Manual counters, cameras, and in-store sensors are more effective at counting every visitor that enters a physical location. However, those solutions require extensive investment and maintenance to sustain accurate measurement.
• Mobile location data can be used to generate an estimation of the total number of visitors to a location. While this method enables visitation data at scale, the mobile-derived estimation is likely to be slightly different than manual counting.

Lastly, unlike other traffic-counting solutions, mobile location data provides visibility not only to customers-owned locations but also to any location of interest, such as competing stores, tenants' locations, public venues, regions, and more.

Should Visitation Data Align Perfectly with Sales Data?

At Placer, we use visitation data as an indicator of performance. However, when compared to sales data, you may not always see a 1:1 correlation and find that a business’ visit trends differ from their sales trends. Multiple factors can cause these differences:

• Online orders: These purchases aren’t accounted for with visitation data since there weren’t visits to the business’s physical location.
• Basket size: It may be that many people visited the location but each spent a small amount of money, or fewer people visited the location and each spent a large amount. These situations will show opposite trends in visitation data compared to sales data.
• Group purchasing: It may be that a group of people visited the location but only one or few in the group made a purchase, showing an increase in visitation data but not sales data.
• Visit durations: In the Placer platform, you can choose to look at visits over 10 minutes, under 10 minutes, or both. You may be looking at the visit duration segment that doesn’t properly capture the visitors who make purchases at the location. For example, if you’re looking at visits over 10 minutes but many visitors quickly pick up purchases and don’t stay at the location for long.
• Chain coverage: When looking at chain-level data, it may be that the visitation data doesn’t cover the same locations as the sales data. You can learn more about chain coverage here.