With PlaceIQ, location data is all about the apps

Maybe we've OD'ed on too many episodes of 24, but when it came to location, we've always thought that our smartphones gave us away. In reality, when it comes to location data, it's a jungle out there, literally.

Pokemon Go proved the value of location, location, location as it provided a bounty of geolocation data to the game provider, Niantic that it could monetize to marketers. But much of that data was probably garbage.

For starters, you may not always know that the data was real; bots could take over your phone and transmit bogus data. And even if the data is legit, there remain a myriad other factors that cloud location visibility. For instance, those apps on which you rely to send location data might not be well-behaved, sending data that is stale, corrupted, or truncated.

Your mobile device may have a GPS, but what the carrier registers is triangulating your position relative to the nearest towers. Providers like Placecast harvest and anonymize this data. But they must supplement it with analytics to cleanse the data and provide "context" by time of day and what type of place the anonymized person was at, such as in the office, at a Starbucks, going shopping, or in school.

If you want absolute precision, there's nothing more accurate than a beacon placed inside a store that knows exactly which aisle you're shopping. That's fine for identifying traffic patterns inside the store and hot spots around specific shelves or displays, but their range is limited and can't give the retailer any advance warning when some of their customers are parking the car or walking nearby.

PlaceIQ is one of a growing list of providers offering to deliver location intelligence. The six year-old company went around carrier data and instead relies on mobile apps to supply location data. The company initially operated an ad network, but as that field has grown extremely crowded, it has moved up the value chain to delivering geolocation analytics. Alibaba recently placed an undisclosed equity investment to gain access to PlaceIQ's analytics engine. Marketing data providers IRI and Comscore are now also consuming PlaceIQ's location analytics.

Now it is getting into the data as a service business. PlaceIQ's data merges GPS and mapping/location data with partner data covering area such as purchases and TV viewership to profile consumer behavior. It recently announced the new Landmark service providing cleansed, anonymized location and behavior data collected from 165 million opt-in North American consumers. Landmark can compare activity at specific stores by day and time, considering the impact of Wal-Mart's Thanksgiving night openings on Costco, which gives its associates the holiday off.

They munge raw data through a mix of heuristics, clustering, and regression analysis to isolate patterns, combine it with trajectory data that provides context, and then anomaly detection to cleanse the data. Understanding context is key to ferreting out anomalies; for instance, a downward spike in activity at a Chic-Fil-A on Sunday in the middle of spring break in Florida can be explained by the fact that the chain has a corporate policy of being closed that day. Making it consumable, they have partnered with the cloud data warehousing provider Snowflake to present the dashboards that can integrate with internal enterprise systems such as CRM

Location intelligence is a classic big data application. PlaceIQ collects billions of data points daily. Yet in spite of all the sophisticated heuristics and machine learning techniques that are thrown at the data, it's surprising that collating our digital footprints through the physical world still requires a lot of art with the science.