Commentary

Rethinking Traditional Audience Measurement Through A Digital Framework

This week I am in London at the biennial Print & Digital Research Forum, perhaps the most rigorous and scholarly symposium in all of audience measurement, with a heritage as the Worldwide Readership Research Symposium.

Not surprisingly, the bifurcation of print media to print and digital delivery occupies much of the thought space here. As I listened to some of the presenters speak about the latest methods for measuring readership across platforms -- while others reminded us of the core verities in audience research regardless of platform -- I found myself pondering TV measurement. (To be fair, I spend a lot of time pondering TV measurement.)

In particular, I’ve been thinking about what a TV measurement -- or perhaps better, video measurement -- system would look like if one were to zero-base such a solution today, with the tools we have at hand, given the measurement challenges we face, while unencumbered by legacy systems.

For years we’ve been hearing that digital needs to be more like TV. This has always been the prescription for making the medium more appealing to advertisers, who still allocate the largest share of budget to TV.  At this point, though, we all recognize that the opposite is occurring: that the medium of TV, as a consumer experience and as an advertising platform, has gone digital. So I believe that were we to zero-base a TV measurement system today, it would look an awful lot like digital measurement.

What would such a system look like? It would need to build on three key lessons from the digital space:

User-centric, device-gnostic: We must come to media measurement today placing the consumer -- the viewer -- at the center.

In the digital space, we used to think that if you measured computers, you were measuring digital, and indeed once upon the time this was the case. In a simpler time, we could conceptualize digital measurement through the lens of the computer panel. Now we know that digital content is consumed on computers, smartphones, tablets, wearables, gaming consoles and so on.

So we cannot think about measurement through the lens of the device; we must think in terms of the consumer, and build measurement that cuts across all devices the consumer uses to access content.

In TV, then, we cannot conceptualize measurement simply based on a box affixed to the set. Consumers are watching TV on all sorts of devices, in and outside the home, via broadcast, cable, satellite and streaming. We must design systems to reflect this complex reality.

Future-compatible: I’ve been in audience measurement long enough to remember when cable TV was a new medium. I was at Arbitron back when we had to figure out how to assure that TV ratings could accommodate that sprawling and unruly new media platform. I’ve been at comScore since 2007, and in that time we as an industry have had to grapple with two new media platforms — smartphones, tablets — along with others nipping at their heels: wearables, gaming consoles and even the Google thermostat. We’ve learned that if you engineer your audience measurement system for the present day only, you are courting disaster.

Big-data-friendly: TV is still the most mass of media, but in this day and age, as media consumption becomes more and more fractured, fragmented and granular, there is no excuse for failing to measure the long tail. And, as we endeavor to robustly measure that long tail, it becomes clear that Big Data assets are essential. In media measurement sometimes we call these assets naturalistic data sets — set-top-box/return-path data, for example, or the dataset generated from site tagging (or by placing an SDK within an app). In the digital measurement space we spent a good 10 years debating the relative merits of site-centric, census measurement versus person-based panel measurement, until finally we experienced our collective, “Hold on! You’re both right!” epiphany.

I’m an old school audience measurement professional, but I’ve learned that the traditional panel, regardless of its quality, is no longer a sufficient asset by itself in constructing robust and comprehensive audience measurement solutions. Quality panels remain important, but over time they’ve become a proportionately smaller piece of the measurement puzzle. Site tags, SDKs, and machine data have all become essential components of quality audience measurement.

I believe that the philosophical approach to building a TV measurement system today must be to assemble quality data assets -- including both panel measurement and naturalistic, Big Data assets -- and then bring them together in a holistic fashion. We need to be diligent in collecting all the TV and video to which the viewer is exposed, and we must be prepared to measure and report at sufficient scale to accommodate the thousand-channel, OTT-enabled, broadcast/cable/satellite/streaming, viewing environment in which we all compete.

And we must be prepared to embrace new channels and distribution platforms as they emerge, even if it means breaking apart what we’ve built and putting it back together again. This is how we’ve been forced to live in the digital space, and it’s how the TV researcher will need to operate as that medium continues to go digital.

What do YOU think?

2 comments about "Rethinking Traditional Audience Measurement Through A Digital Framework".
Check to receive email when comments are posted.
  1. John Grono from GAP Research, October 22, 2015 at 5:56 p.m.

    I know this won't surprise you Josh but I totally agree.

    Our online data is based on a hybrid of tags, SDKs and panel data with the focus now totally on audience and not traffic.  The unique browser is dead and vendor and buyer talk unique audience.

    By the end of the year we plan to have streamed TV programmes sitting side-by-side with TV ratings with the stream as an average minute audience count.   De-duplication is the next bridge to be crossed.

    We launched MOVE to measure the OOH industry back in 2010 using 'big data' sets - vehicle and mass transport counts, mall foot-traffic counts, airline travel and airport counts, census data - and coupled them with household travel surveys and viewability studies to produce 'average day' audience data.

    Our CineTAM service couples official cinema ticket sales (by outlet by session by movie title) with the various cinema customer loyalty programs to produce demographic profiles.

    Our cable TV service (Foxtel) uses RPD data from 110,000 homes daily to get the HH tuning, then overlays that with panel-based programme-level heuristics.

    One of our readership services (EMMA) uses offical AMAA circulation data coupled with sample-based audience information in order to be able to report readership estimates with demographic breaks for the smallest of titles for community newspapers, and regional and rural newspapers.

    So, yes, I agree.

  2. Joshua Chasin from VideoAmp, October 23, 2015 at 3:09 p.m.

    By the way, as profound as I may find the gnostic legacy, That bullet should read, "Person-centric and Device-AGnostic."

Next story loading loading..