We often see articles about various types of data with categorization of first party (my) data, second party (your) data and third party (vendor aggregated audience) data. As we were developing the Inscape census smart TV data, I struggled with these definitions.
On the one hand, you can think of census TV data (e.g., device level data from set-top boxes or Smart TVs) as an aggregation of the first party data sets for all the networks, local stations and advertisers that are in the dataset. That’s kinda exciting (and valuable). Such data sets are aggregated and provided by a vendor. And they can be used as third-party data. For example, a census data vendor can supply a list of TVs that have displayed college sports or ESPN or Big Ten Network, or what have you. In this case, it really is third-party data.
But that isn’t the primary use of such data. There are a great many uses, but the one that is most interesting at least presently is to use the data for what I decided might be called, “Match Fabric.”
Why? Because the fascinating value of census TV viewing data is unlocked when we enable advertisers, agencies, networks, and stations to match specific audiences to the viewing data. This tends to occur in something of a cycle.
On the buy side, advertisers, long frustrated with crude demographic targeting for TV advertising, can now match first, second or third party data to viewing data. An example of a first party data match might be generating a list of people that bought a Ford truck five years ago. Such people might provide a useful target set for a new pickup truck ad. Ford, or it’s agency (or data vendor!) can match their truck owners list to the TV viewing dataset to produce a list of networks, dayparts or shows that have high concentrations of the target list.
With initiatives like OpenAP, Ford can then ask TV ad sellers to sell them a media buy focused on shows with high concentrations of such viewers. Then the campaign runs. With the list matched to TV display devices, the ad seller or an intermediary (or Ford’s data vendor!) can then measure the actual reach and frequency of the campaign against the target list to measure effective target GRP.
But we’re not entirely done. With another match – this time Ford truck buyers – we can produce a deterministic measurement of TV ad effectiveness. This not only reveals which ad spots not only reached the most people in the target group but also provides the highest number of truck purchases. We can see the impact of frequency on conversion as well as recency and combinations of recency and frequency (e.g., the optimal result for truck purchases might come from a frequency of 3 within 11 days).
So, census TV viewing data turns out to be extremely valuable centrally because it can be used to match audience targets to audience delivery, to target media-buying more efficiently and make it more efficient. I use the term ‘Match Fabric’ because we are using the census smart TV data connect audiences to behavior and the phrase illuminates the virtue and value of the data.
Over time, we expect that efficiency will quickly evolve from effective CPM, or the cost to reach my target audience, to conversions per dollar of media spend. Once the seller of mobile games know how many downloads they're getting per dollar, that quickly becomes much more important than audience measurement – at least for those advertisers that can rapidly capture and apply consumer behavioral data.
However, such systems are only as good as the quality of the data that is employed, and the quality of the systems that glue it all together. After serving my term at Inscape, I joined Dativa because I think there’s great value in high-quality sourcing, matching, transforming, modeling and implementing census data as well as the audience and behavioral data that we match to it. We don’t make data, but we do help generators, owners and licensees build solutions based on Match Fabric.