When Mad Men gave up being mad: how big data is shaking up the old order of TV ad media buying
In the fast-moving world of digital, ideas cannot be allowed to stand still otherwise they quickly fade and die. Contrast that with the old world, particularly in legacy businesses, where ways of doing things have remained the same, unchanged for years, decades even. Since the advent of the concept advertisers have bought their ‘spots’ on TV in largely the same way: according to the makeup and size of the audience watching. No longer. That model has been – pardon the expression – ‘disrupted’.
TVSquared, an Edinburgh-based startup, has in just three-and-a-half short years developed and honed a new way of buying television advertising, which looks in realtime across large swaths of response data and using algorithms to instantly analyse what is working and what’s not. That ‘optimisation’ process then feeds back into the media schedule, giving the advertiser all sorts of very useful information, usually within hours. If they have two adverts, it can tell them which is the most successful, it can tell them when is the best time to advertise in terms of the biggest response uplift and whether that’s by people flooding onto the advertiser’s website, downloading an app or phoning a call centre.
It can also work out which area of the country and which networks are the most effective, which is a huge ad- vantage in its biggest market, the US, where there are 210 different marketing areas. But most importantly it tells advertisers whether they are spending their money in the most efficient way, and allows them to ‘reweight’ the media they buy to reduce how much it costs to generate each response, lead or sale, the all important ‘cost per acquisition’. That is the ultimate metric which has seen TVSquared start off from humble beginnings at Edinburgh tech incubator CodeBase to become a company of 35 people, with offices in London, New York and LA, and is likely to double its head count this year.
When I talk to Blair Robertson, the company’s Chief Analytics Officer, a job title which will surely become more popular, he explains that the company is now on its 75th iteration of its Advantage platform. The constant re-invention, adding yet more dimensions to what the statistical modelling-based calculations can analyse, is breathtaking. “We really can use it to address almost any question we have from TV advertising,” says Robertson. “The platform is extremely scalable and the technology works regardless of whether you’re airing one TV spot a week or 100,000 TV spots a week; it’s really been built to deal with all the complexities. And that’s what happens when we move from version one to where we are now.”
Robertson insists that the platform is not a ‘regressive model’, it doesn’t look back. It is looking at the pattern of response data as it emerges, looking for the characteristics that drive people to go online, especially as so many are now ‘dual-screening’: watching TV and fiddling with their smartphones at the same time. And it seems to be working: TVSquared has managed to save its clients typically between 25 and 45% on its cost per acquisition, with some up to 80 or 90%. These are staggering sums. “If you’re spending $10million on TV that represents an extra value of $4 to $5million, so it’s very, very valuable and it’s becoming more so,” adds Robertson.
The success of TVSquared – whose clients include the likes of travel website Expedia and furniture retailer Made.com – is built on the growth of e-commerce, which has allowed much greater insight into consumer behaviour. Its platform can track people’s responses: they may come to a website after seeing an advert, but not make the purchasing decision until much later, during which time they can be re-targeted through the use of anonymised cookie data, with additional Facebook adverts or an email campaign.
“Because we’ve cookied that user we can essentially say, ‘well they came back two weeks later and they made a purchase’. But we can credit TV for driving them to the site for the first time and then we can credit the other digital advertising for bringing them back later on,” says Robertson.
Not bad for a technology that was conceived in one of the founders’ living rooms.
The group – including CEO Calum Smeaton, a 20-year veteran of founding innovative tech start-ups – got together to help friend Kevin Dorren of Diet Chef, the Edinburgh-based online diet food delivery service.
His advertising was working but he wasn’t able to determine whether it was a particular kind of advertising or creative that worked.
“He didn’t know and he was being told by his media agencies that he should use his phone responses to try and work it out,” Robertson explains.
“He said ‘well, hang on a minute, I’m an e-commerce company. I’ve got 35 times more sales coming through the website than I do from my call centre so why on earth would I use phone data?’ He kind of challenged Calum, to say ‘surely there was some way you could use data science to address this?’”
The rest, as they say, is history and now TVSquared is operating in 45 countries with around 330 different brands tracking about 200million visits to various websites.
The decision to stick to TV advertising seems deliberate: it is a global market worth $200billion annually. Whether that means TVSquared can propel itself to that next level, and gain ‘unicorn’ status, is another matter.
I put this to Blair after reading, post interview, that his company has been hotly tipped to become a billion-dollar company by CodeBase founder Jamie Coleman, who described it as a “rocket ship”. I haven’t heard back yet, but watch this space.