What impacts does the SDK have in terms of battery consumption?
Pilgrim SDK is extremely battery efficient, as it uses less than 0.5% of daily battery when running in the background, as measured with normal usage across our owned and operated as well as partner apps.
When a user is stationary (at a visit) Pilgrim shuts down background location usage and is idle. Pilgrim is only active when the user is moving, which makes it very inactive for 95% of most days and times.
Beyond that, improvements at the OS level have further helped overall battery drain issues in recent years.
Furthermore, Pilgrim SDK uses the iOS region monitoring framework to set a single iOS geofence while the user is stationary to shut down background location usage and improve battery efficiency. The geofence will be removed when the user leaves their stationary location.
How accurate is the position detection?
To ensure that Pilgrim delivers reliable attribution accuracy in complicated environments like malls and cities we conducted an experiment across a variety of locations over several months.
In the experiment we found that Pilgrim Snap-to-Place performed over five times better at placing a device at the correct venue in dense cities than polygons, a manual mapping methodology. Commonly used in geofencing, polygons often struggle to accurately attribute visits in multi-story buildings. And in compact spaces like malls, where multiple stores are nested close together, Snap-to-Place was over five times more accurate than polygons. To see a relative comparison of how Snap-to-Place compares to polygons, venue radii and more, check out the grid below.
How does it handle altitude (elevation) in a building?
By leveraging barometric sensors and verticality index signals from mobile devices, Pilgrim SDK is adept at gauging altitude (elevation) in buildings where there are retail commercial activities such as malls and shopping centers, as opposed to a corporate office building. WiFI and BLE triangulation also help to round Pilgrim determination of a user’s position in relation to a venue where verticality may play a factor.
How does it detect the position in a building (integration with wifi, cellular network)?
Wifi and cellular network are only two of the different signals (GPS, bluetooth, accelerometer, time of day) which inform our Pilgrim model. They operate in conjunction with check-ins and human validation signals.
Our Pilgrim Snap-to-Place technology uses sophisticated machine learning models on top of this first party data set. Our multi-sensor technology allows us to go beyond GPS and lat/long, to also receive Bluetooth and beacon signals, wifi signals, and information from the compass, accelerometer, and barometric sensors that are built into smartphones. By using those signals, we understand exactly where a user is.
What is the minimum dwell time in a location to detect it (how many samplings to be confident about the location)?
There is no standard minimum dwell time in a location per se (on average, it is at least 2-3 minutes) - it is variable based on a few different factors, such as duration and speed of movement.
Specifically for geofencing, the SDK can tell you when a device's GPS has entered, exitted, or remained within a place for a certain amount of time (dwell). Geofences fire Entrance, Dwell, and Exit events. The Dwell event fires after the device has been inside the geofence for a certain amount of time. The dwell time is also configurable.
How often do you sample?
The sample rate is variable.
Are frequency and mode of sampling are parameterizable?
Frequency and mode of sample are not configurable by the SDK partner.
Can it detect the means of transport used for the user's movements?
Is it able to understand if the user is the driver or passenger of a vehicle (car)?
What is the frequency of data transfer from APP to central system (proprietary platform or post systems)? Is it in real time (pro engagement)?
With Pilgrim SDK integration, the data transfer is in real time. An alternate option available is Enrichment - audience segmentation based on users, matched against Foursquare’s broader targeting device graph (typically significantly less scale than via Pilgrim SDK integration) and there is a week-long latency.
How is it integrated with engagement and campaign systems?
The true power of Pilgrim begins to showcase itself once you start integrating its event detection with other services. Pilgrim supports a number of third-party integration options in addition to delivering data to Cloud Data providers.
Many teams use a Mobile Marketing Automation (MMA) or a Customer Data Platform (CDP) solution to manage events, create content based on those events and/ or associate those events with their users. Similar to webhooks, we will send a notification to any partners you have configured when an arrival or departure event occurs.
You can directly integrate Pilgrim SDK events into the platform your company is using. You can also receive a daily delivery of Pilgrim events to an S3 bucket you set up. If your preferred CDP, MMA or Cloud Data provider is not already integrated, let us know and we can look into supporting an integration with them.
Are there proprietary or open analytics to integrate location with socio-economic layers (POI, etc)?
As a Pilgrim SDK partner, you can have access to the largest and most accurate Places database. Foursquare’s database has Point-of-Interest data on 130 million places globally, including: Name, Address, Venue Category, Location, Hours of Operation, Photos, Tips, Reviews, Ratings, etc.
Are there proprietary or open analytics to identify customer behaviors?
As a Pilgrim SDK partner, you can use 50 targeting segments from Foursquare taxonomy (800 segments in total) and apply them to your user data to create look-alike audiences and modeling. Segments are a sophisticated way for you to understand users, powered by years of Foursquare data science and analysis of visit patterns. Intelligently target your users based on their visit behavior for deeper customer engagement with pre-built user segments. Build personalized experiences for your users based on “who they are” according to each user's visit patterns and history.
Updated 3 months ago