Diving Deeper: A Look at New Breakthroughs in Unacast Products and Features

Diving Deeper: A Look at New Breakthroughs in Unacast Products and Features

Even amid dreamy summer vacations and onboarding new Unacasters, I’m excited to say that our product team delivered big in Q3, and we’ve never been more ready to share our breakthroughs with the market.

Uncontextualized Visits translates raw pings into a meaningful stream of events representing where a device rested during the day, unlocking insight into 75 to 100 million visit events and 17 to 20 million daily active users. Meanwhile, new features, including Maximum Dwell Time, update our core Visits product and further increase the available scale by 55 million Visits.

What's New from Unacast

So, let’s start with Uncontextualized Visits. Like its name suggests, this data feed contains all Visits except no points of interest (POI), enabling our clients to use their own POI to contextualize or to use Visits without the contextualization. The result? 75-100 million newly available Visits and 17-20 million daily active users. Second, a new Visits clustering algorithm - proving we are consistently making improvements to our algorithms to add quality and value across all of our datasets. And finally, calculating Maximum Dwell Time for all of our Visits, which is what I really want to dive into in this piece.

If you are an avid reader of the Unacast blog, and we know you are, you’ve read our complaints about the lack of data quality/transparency in the industry. Candidly, because we put so many eggs in the quality basket, it’s tough to also achieve scale. We spend lots of time working with our clients finding the best way to achieve quality and scale. Impossible you say? With our latest release, we are proving you wrong.  

Before I get into how we are achieving this new scale, let me pull back the curtain on the dirty secret the location data community doesn’t want you to know. They do anything they can in order to achieve scale.  

That means assigning one visit to many places at once and also taking any signal they can in order to say that a device visited a location. What do I mean by that? Well, if you have a GPS signal from when you were driving past a Walmart that is close to the road, they will say you were at Walmart. Now, they’ll tell you that those few visits don’t really matter. Ask them what it means if that Walmart is near a busy intersection with a traffic light.  Hundreds of visits per day will be assigned to Walmart, that were really just for devices driving by. What does this mean for data quality? It means it’s terrible! Previously, Unacast counteracted this issue by never creating or validating a visit that didn’t represent a dwelling device, requiring 2 GPS pings that are at least 8 mins apart. This strict rule was great for quality but bad for scale. We have plenty of clients that need scale, but don’t want the bad data with it.  

So, how did we solve this age-old scale vs accuracy problem? We’ve started calculating Maximum Dwell Time for all our visits.

A Closer Look at Maximum Dwell Time

Maximum Dwell Time describes how long a device could have spent at a location by looking at the data before and after each Visit. When we determine that a device was driving past a store, we still won’t sell that as a visit. However, when we’ve determined that a device could have dwelled at that location for longer than just a drive by, we’ve determined this is a potential Visit. The same rules apply as a Visit for duration, the max duration will need to indicate that a commercial visit was feasible, by lasting longer than 8 minutes.

Let me walk you through it.

See picture A. It’s nice. We have a lot of pings from a device and by looking at the pings, we can see the device traveled, dwelled at Joe’s Crab Shack, and then traveled from Joe’s Crab Shack. In this case, we can see there was a Visit and that we observed pings for most of it. Not much uncertainty here. Observed duration and the max duration the device could have dwelled here will be nearly identical.

Now let’s look at Picture B. This is more realistic - location data from cell phones is collected at varying cadences, with varying rules as to what causes the device to check to see if it has moved. Therefore, it can be difficult to understand how long a device actually dwelled at a location. Here’s where maximum dwell becomes particularly useful.

We can see that the device still visited Joe’s Crab Shack, but we only observed pings for 44 minutes. Does that mean that the device was only there for 44 mins? Maybe, maybe not. We do know, though, by looking at the pings surrounding the Visit that the device could not have been at Joe’s Crab Shack for longer than 74 minutes (less travel time to and from the Visit).

The difference between the observed duration and max duration is 30 minutes. The person could have been there - or he could have stopped somewhere else along the way. Either way, we let you into the secrets surrounding the Visit for you to decide how to use it.

And back to scale - How does this help us solve our clients’ scale problem? The transparency into the underlying data surrounding a visit enables us to deliver Visits that we previously couldn't. We will now deliver One Ping Visits that the industry has been giving you this entire time. The difference, though, is that we will give you valuable information surrounding the ping, and we won’t give you any One Ping Visits that couldn’t possibly have dwelled at a location.

Maximum Dwell Time enables Unacast and our clients to have a powerful view into how a device, and the person behind the device, interacts with the real world - achieving both quality and scale. In a world where not all data is created equally, transparency enables us and you to make better decisions around how to use data. And with this - go on, use Maximum Dwell Time and all the scale that comes with it!

Want more information about our latest releases? Read about them here.

Related Articles