Identifying visits coming from TV advertising: elementary?
Identifying visits coming from TV advertising and being a crime investigator have a lot in common. Thankfully, surfing onto an advertiser’s website after having been exposed to media pressure is no real crime.
It is now obvious to all that the TV generates a real impact in advertisement. However measuring this impact is no easy thing, for a simple reason: we are talking about offline media. Any online advertiser owns the native ability to identify visits coming from ads. As long as technological solutions are in place, a few clicks and tracking devices are all what it takes to perform proper identification. It is as if fingerprints were intentionally left on our crime scene…
The issue is quite different – and harder – for a TV advertiser. Drawing a line between the profiles which have been exposed to a campaign and those who have not is a difficult task. This is a TV-specific paradox: as this media is less intrusive than the online one, televiewers actually leave less evidence behind, hardening the job of the detective.
The number one step in identifying visits coming from TV ads is a good understanding of how people consume this media and how it interacts with the online world.
When facing an apparently insoluble mystery the best detectives often use their ability to recreate the crime scene. The equivalent of that gift in the world of TV advertisement is a global vision of telewatchers – web users’ behaviors. At Realytics we strive to implement something a real bloodhound would indeed rely on: robust scientific methods whose aim is to best model these behaviors.
Our top-down, granularity-seeking approach allows us to differentiate the following categories – always relying on all data at our disposal:
– the baseline (the innocents): those are the visits with no relation to the ongoing campaign;
– the indirect visits (the suspects/accomplices): these visits have benefited from the halo effect of the campaign, through repeated exposure, buzz etc.
– the direct visits (the culprits): those sessions are most certainly coming after having watched a spot.
Using technology and machine learning: best having 10000 Sherlocks on your side than one.
An accurate modeling of behaviors stemming from exposure to advertisement is achievable through a good command of both technology and data management. In a second step our machine learning and time series analysis algorithms allow us to deal with the sheer size of data in a disciplined fashion – very much in the manner of a well-equipped team of detectives. At the end of the day data granularity vouches for accuracy when identifying the visits coming from TV advertisement.
