After some big conversations with outdoor advertising industry leaders, we got to work on the next iteration of our billboard analytics system. Our original intent was a bit simpler than our delivered work, as per usual, but the performance gains we’re seeing are making it all worth it.
The biggest excitement we’re seeing is directed toward how our API is validating OOH inventory, building targeted audiences of IP data on the fly, and then re-targeting via our DSPs. So, we’ve made the API even better.
Billboard Vector Filtration
Woah. That sounds neat, eh? Usually billboards don’t find themselves paired with high-tech sorting and directional vector algorithms, and we think that’s kinda sad. Because we’re sensitive to these sorts of things, our team decided to put our analytics to work for us.
When your platform can process 35 billion data points a day, you can get a little spoiled. Thankfully, our moms raised taught us to share. Our API now includes this feature as a baseline.
But what is this magical billboard analytics algorithm? Is this related to the 3D billboard mockup tool that you’ve said nothing about publicly?
Easy now, we’ll get to that another day.
Billboard vector filtration (TM) centers around an existential question at Adder: if you’re not looking at the ad, does it exist?
Short answer: yes, it does exist — Elon Musk might be the center of the simulation but we are not.
The longer answer: yes, the ad still exists, but does it matter if people aren’t looking? A third unsolicited rhetorical answer: No. It does not.
Being mindful of those philosophical concerns, our engineers set out to filter out bad traffic we could be counting as outdoor impressions. After all, verification is everything, and getting as close as possible to the truth of outdoor attribution is exactly what we’re doing in this space.
In other out of home attribution systems, the direction of an individual might not be known, and thus, your system might accidentally count people that are facing away. Worse, it might count people moving away from your OOH billboard or bulletin or sign spinner, and so on. It’s quite a processing challenge to understand all of these things, and some heavy duty calculations are needed to extrapolate directionality from our massive data sets that are reported daily.
This attention to the real-world details, of course, means fewer reported impressions overall. This can be frustrating to marketers who like to see big numbers, but there’s a bigger picture to keep in mind — the numbers don’t matter if they aren’t real. In fact, they’ll just end up screwing up your marketing efforts in the long run.
So, to put a pin in this feature: If you’re facing a billboard, we’re counting that as an impression. And if you aren’t, we aren’t. It’s pretty simple, really, but it’s more complicated than it might sound.
The Other Thing…
Our changes on the validation end of the impressions and analytics API inspired us to make additional tweaks on the backend. These focused primarily on networking, APIs, and making sure everything is locked down to prevent intrusion on listening on our connections.
If you’re not a developer, and have persisted to this point, you have my appreciation. To put this in laymans terms, the Adder Analytics portal is now even more secure, and loads faster than ever. We owe extra special thanks to our CTO, Brandon Bush, for the multi-threading optimization, which distributes workloads about our onsite cloud servers.
Oh, hey, now there’s a topic to write on!
Coming soon: The Adder Analytics Cloud.