How Greater Sandusky Partnership Measures Visitation
Who provides this data?
GSP works with Placer.ai, a location analytics company (placer.ai). It measures how many people visit physical places across the United States, and it is one of the most widely used platforms of its kind, relied on by thousands of local governments and civic organizations along with retailers, universities, and real estate firms. Placer.ai does not collect anything from people directly. It receives anonymized location data from a large network of mobile apps and turns that data into estimates of how many people visit a place. The company is independently certified for data security and publishes its privacy practices openly.
What is this data?
An estimate of how many visits a place gets, based on the movement of anonymous mobile phones. A place can be anything with a geofenced boundary: a store, a festival, a park, a downtown, or a whole county. The same method works at every scale. It describes places, not people.
Where does the data come from?
From ordinary smartphone apps that people have given permission to use their location, things like weather or navigation apps. Those apps share anonymous location signals with Placer.ai, which cleans and aggregates them into estimates for places. GSP accesses the finished, aggregated data through Placer’s platform, the same way it would read a published report.
Does it use cell towers, or track or "ping" my phone?
No. It has no connection to your phone carrier, your phone number, or any cell tower. Nothing pings your device, and nothing follows you in real time. The data comes only from apps a person chose to share location with, and it arrives already stripped of anything that could identify a phone or its owner.
Whose phones are included?
Only adults in the United States who opted in through an app. Phones belonging to anyone known to be under 18 are excluded. The panel is built entirely from U.S. devices, so it does not include international phones or track foreign travelers.
How does it know someone visited a place?
Every place is defined by a geofence, a virtual boundary drawn around it on a map. A geofence can be as small as one building’s footprint or as large as a county line. When an anonymous device is observed inside that boundary long enough to signal a real visit, rather than just passing by, it registers as one visit. GSP can set how long a device must stay inside the boundary before it counts, which screens out people who are only passing through and tailors the count to the place being measured. Because the geofence and that threshold define every number, drawing them correctly matters, which is why GSP verifies both before trusting the data.
What is the difference between a visit and a visitor?
A visit is a single trip into the geofence. A visitor is a unique person. One visitor can make several visits, so a place that records 1,000 visits in a month might come from 600 visitors, some of whom returned more than once. Visits measure activity. Visitors measure reach. GSP reports whichever one answers the question being asked.
What else can it measure?
More than a headcount. It estimates how long people stayed, how often they return, and the general area they traveled from, reported at the ZIP code or neighborhood level and never as an address. It can also separate a place’s traffic into visitors, residents, and employees. That split is based on patterns of time, not identity: a device at a location about eight hours a day on a regular schedule reads as an employee, one that regularly spends nights there reads as a resident, and everyone else is counted as a visitor.
Can it show how visitation changes over time?
Yes. The data goes back to January 2017, so GSP can run any time frame and compare it against another: this summer against last summer, this year against pre-pandemic, one event against the same event a year earlier. Most of the value is in the trend, not a single number. Watching how a place performs across seasons and years is how GSP measures whether an investment, an event, or a corridor is gaining or losing ground.
How do a sample of phones become a full count?
The data covers a large sample of phones, not all of them, the same way a poll surveys a sample of voters. Statistical algorithmic models then scale that sample up to an estimate for the full population, correcting for who is and is not represented. Because it counts devices rather than people, and a group may share one phone or carry several, the result is a calculated estimate, not an exact headcount.
How accurate is it?
Estimates typically fall within about 5 to 8 percent of the actual count, in either direction. Placer.ai reports accuracy of 92 to 96 percent, and the Urban Libraries Council, testing it independently against authoritative visitor counts, measured correlation above 90 percent. Federal researchers at the U.S. Environmental Protection Agency have tested this class of data against National Park Service counts and confirmed it a reliable way to estimate visitation.