Dativa was at TV Week in New York last week, with our Chief Data Scientist Tom Weiss speaking on a panel about how data analytics is becoming TV's new table stakes at the event's TV data day. Here are our key takeaways from the show
TV is in rude health…or is it?
TV events usually contain a healthy dose of chest thumping about how healthy the industry is. One could, if one desired, make a long checklist of the types of phrases reliably trotted out by industry execs ($72bn transacted…check! Best medium for building brand awareness…check! Gold standard..check!). So it was illuminating to hear Turner's Larry Allen kick off his afternoon keynote by getting to straight to an entirely different point, namely, that the current TV advertising model is broken. His argument, loosely summarized, is "too many spots, too little targeting". That's not hugely surprising to hear at a data conference, and there were plenty of robust defenses of the status quo on display. However, it's a sign of where we are that a speaker from a top network group is so openly challenging said status quo on a stage at an industry event
How much TV data complexity will advertisers take?
New data is needed to solve some of TV's challenges – on this, everyone now broadly agrees. The more interesting question is what data sets we use, and how best to apply TV analytics to them. CBS' newly promoted analytics chief Radha Subramanyam hinted at a possible answer – all of them - by highlighting almost a dozen different sources of viewing data that her organization uses, including set-top-box data, Smart TV data and OTT data. The sheer volume of these datasets presents huge challenges for the buy and sell side – including many of our customers – around understanding how to get the most out of these data sets, and how to transact on them in a more streamlined fashion. Publicis' Jonathan Steuer highlighted that TV needs to "behave more like a platform in terms of planning, execution and measurement". That's true, but perhaps if TV really is trying to be more like digital, as most speakers hinted at over the days, then acceptance of some of that complexity – at least from a data standpoint – is going to come with the territory TV wants to stake out.
Goodbye attribution, hello outcome-based marketing
Outcome-based marketing (essentially TV attribution, but with a much sexier name, as far as we can tell) was comfortably the biggest buzzword of the show and another sign of TV's desire to embrace the good of digital. A plethora (www.cubeiq.com) of (www.ispot.tv) vendors (www.dataxu.com) are targeting the space, interest in which has also been peaked by networks like A&E offering delivery-based outcomes at this year's upfronts. There are a few obvious attractions to outcome-based marketing - competing with you knowwho, more granular targeting - but a less heralded upside is attracting new advertisers to TV. Outcome-based marketing can help grow the pie by attracting both brands - particularly direct-to-consumer – and marketers who are very comfortable with approaches around sales conversions, search lift, and so on.
Attribution isn't perfect. Challenges remain on how to properly attribute actions in specific categories (how do you attribute the purchase of a car, or any category with a six-month decision cycle, on a single spot?). However, perhaps the bigger challenge for TV is how to balance its newly acquired ability to track outcomes with its current turf as a brand- and awareness-building medium - and a very good one at that. As Hulu's Julie DeTraglia nicely put it, nobody is going to subscribe to her service if they don't know it exists.
Bad datasets are going to get found out
"Data science is a bullsh*t term, and we're all bullsh*t artists" – words not spoken by us, but by a leading network's head of data science. He was joking (at least that's what we told our data science consulting team) – but he did inadvertently strike upon an issue we see more and more with our clients: data quality and transparency.
If TV data is still a bit of a Wild West, inevitably, not all providers are going to be Rooster Cogburn. Not all TV datasets are created equal, and at the show our CTO Tom Weiss shared our experience of working with poor quality audience segments that we've seen being traded – something we've blogged about before.
All datasets have strengths and weaknesses, and it was refreshing to hear Inscape's Zeev Neumeier telling delegates "they should be careful what they wish for" when using his dataset. If buyers and sellers are to accept compromises in the data they use to trade - in the same way digital buyers and sellers do - dataset providers must be transparent about what their data can and cannot do.
Don't forget about the consumer
It's easy for us data wonks to get caught up in match rates, upscaling, panel balancing and all the rest, and forget that part of the benefit of data-driven TV advertising is to improve the experience for consumers. To that end, Turner showed some fascinating research the network had carried out with consumers around the kinds of new TV advertising experiences they'd like to see. One result was a single 5-minute long central spot, which tested very well. Turner also shared psychographic research around ad receptivity – a consumer that watches CNN in the morning is going to be more receptive to certain types of ads than TBS in the evening. We've written before about how the future of TV advertising is also about bringing in more of these kinds of contextual insights into the mix. This kind of small data, if you want to call it that, is going to be just as important as its shoutier, limelight-hogging big brother in the future of TV.