While mobile advertising is expected to grow into a $100 billion industry over the next four years, much of the targeting and attribution currently used is reliant on inaccurate location data. This leads to misplaced ads, false attribution and dubious analytics. To solve this problem, Skyhook cleanses all inaccuracies in location by leveraging its 13+ years of location leadership and expertise, its global location platform, and its ground-truthed, quality-controlled venue set. Advertisers and publishers can create high-value mobile targeting and consumer insights with our premium, location-derived mobile segments.
To build the best behavioral profile of a user we have some required data parameters. User sample locations must contain a unique device ID, a timestamp and a latitude and longitude with at least 4 decimal places. We want (and can handle!) a continuous stream of data. By ingesting more data per user, we can identify more behavioral patterns and provide deeper insight into user preferences.
The Proper Process for Refining Personas
At Skyhook, user location samples undergo a rigorous refinement process to become Personas:
Step 1: Quality Check
Skyhook’s location processes trillions of location signals each year which we use to constantly refine our network and source of ground-truth. By processing so many on-device location signals, we have unique insight into what good location data looks like. We use this knowledge to certify location sample quality. Lat/Longs that do not meet our criteria for precise and accurate location are excluded from the Persona creation process.
Skyhook can detect when submitted latitude and longitude samples have been generated from poor positioning, such as IP-derived location, and scrub data points that have been coarsely positioned at the midpoint of a centroid. We’re also able to throw out inaccurately positioned latitudes and longitudes based on historically verified device location samples.
Our quality filter is unique in that we do not derive location accuracy from programmatic bid stream data or probabilistic guesses of user movement.
Step 2: Venue Match
In the Venue Match process, we bind the cleansed location samples to curated points of interest in our database of precisely located venues.
Skyhook passes venues through a meticulous scoring and quality assurance process, curating a highly accurate and actionable database of places of interest. Cross referencing the location of ad requests with our database of over fifteen million venues yields user mobile behavior patterns that are the foundation of user insights and effective targeting.
Step 3: Homing
Skyhook passes cleansed location samples through our proprietary probability density functions to identify with confidence which location samples represent a user’s home. By determining where a user lives we can determine Census demographics such as income level, education level, ethnicity and age of that user. Once the user has been “homed,” we add the demographics that align with that user’s census block to each user’s Persona set.
Step 4: Behaviors Process
Personas are made by combining venue visits and frequency along with user demographics. In the Behaviors process, Skyhook determines if users meet any of the Persona definitions.
Partners can build mobile audiences with both ease and flexibility. We offer forty-eight Core Personas - our Core Persona algorithms go through a rigorous research and development process to ensure that they accurately reflect human behavior and movement.
If your Personification needs extend beyond our Core Skyhook Personas, we open up any of our 3,000+ On-Demand Personas to choose from. For example, you can choose any user that has visited a Walmart in the past 30 days, or choose any user that has visited both a Walmart and a movie theatre in the last month.
Once your data has made its way through our process, we package your data and send it back to you on a regular basis. The output is Persona segments in the form of a list of your user IDs that are decorated with Skyhook Persona IDs.
These enhanced Personas offer the most accurate and actionable portfolio of location-based behaviors in the industry, resulting in compelling insights into publisher audiences and unmatched targeting and attribution for advertisers.