Why Data is King


People who work in advertising are getting nerdy. The “data dorks” are now sought out.

We believe the advertising industry will benefit from more data wonks — people who ask insightful questions and make honest, no-nonsense observations so the field can mature and so practitioners can get a deeper understanding of the laws of online advertising. One law of online advertising we follow is getting a deep understanding of what the goal of a campaign is instead of settling for the target audience description.
For example, an advertiser might say, “I need to target women ages 35-55” when the overall goal of the campaign is to increase brand affinity towards “I can’t believe it’s not butter!” A data-driven partner is going to ask why you’re targeting that demo, and then they’re going to come back with a way to use technology to achieve the objective more effectively. Real-time bidding is transformational in a context where nerdy buyers are focused on achieving goals versus indulging their beliefs about how to make campaigns work. Response prediction is key.
It has been said that display is on its way out, because people dislike banner ads and the like. Let’s define display as an ad served in some context other than immediately following a search. Display in this sense will be with us forever.
I’m not sure people dislike advertising overall. They dislike bad ads about belly flab or some housewife in Milpitas who has developed a miraculous tooth whitening process.They hate disruptive ads that expand to cover their screen and make them hunt for the “close” box. But honestly they love aspirational advertising — just look at the ratings of Sex and the City, which is essentially a huge product placement vehicle, or MTV’s Cribs for that matter.
There’s nothing inherent in the display channel that requires ads to be annoying, irrelevant or non-personalized, it’s just the way things turned out to be based on limitations in the pre-digital era that many advertisers have carried over to the way they buy online display. The kind of buying that says, “I want to pay a huge premium to show my ad to whoever visits some major portal’s homepage tomorrow!” is likely to become extinct, replaced by the kind of thinking that says, “I want to drive maximum brand awareness lift with my budget.”
The question has been asked that if data can create amazing audience segments, every campaign, including display needs all the data that it can get. Campaigns are like snowflakes: they’re all different, and I wouldn’t say there is any data that they all need. What they do all need is scientific optimization based on a test-and-learn framework that is happening at a scale and frequency that you can only get with machines. The bottom line is that if you can listen to people– and by “listen” I mean “program computers to analyze the response patterns in the billions of ad-serving impressions seen daily,” you can find patterns that let you deliver campaigns well.
Author and humorist David Sedaris wrote in SantaLand Diaries about his experience working as an elf in “SantaLand” at a Macy’s department store. Surprised and disappointed at the strong patterns in people’s responses to the various stimuli involved, Sedaris wrote “you would like to believe that everyone is unique, but then they disappoint you. ”

What was disappointing for Sedaris is gold for smart advertisers — just as people tend to make the same jokes when you concentrate us geographically and demographically in a store, we also tend to respond similarly to advertising when clever media buying concentrates audiences with similar brand affinities, interests, and immediate mindsets.
In this context where people are beneficially studied less as individuals and more as members of various clusters, rocket science is going to beat a human marketer’s guesswork every time. It’s not like Wall Street where a monkey throwing a dart can pick a portfolio of stocks as good as a professional; it’s more like a monkey trying to win Jeopardy playing against Watson and Ken Jennings. That is the power of predictive analytics. Predictive analytics are crucial to the future of advertising, including any evolution of display.
As a former academic who taught courses on artificial intelligence and advised the National Science Foundation on its investment in AI research, I would say that it’s very unlikely we’ll discover the limits of predictive technology in our lifetimes or our children’s’ lifetimes.
As a practical tool for optimizing an advertising campaign, predictions are immensely useful. It’s like having tomorrow’s newspaper today, only the headlines and articles are about the user segments, pages, times of day, cities, and so on, where the best response rates can be found for different campaigns. With real-time bidding, Rocket Fuel’s servers get poked billions of times a day by various exchanges who ask, “Hey, what would you bid for this impression?” We score each impression against each eligible ad. If an ad is part of a campaign with a brand awareness goal, we score the impression to predict the likelihood that the user would later respond positively to a brand survey. If an ad is part of a direct response campaign, we would score the impression to predict the chance that the user would later visit the advertiser’s site and buy or submit a lead. Then we figure out which is the best ad to show and the value to our advertiser in showing that ad right now. We might end up bidding $1.284278 for an impression, for example.
A lot of companies who claim to do real-time bidding are actually just writing little tools to submit human-set bids. For example, a strangely large number of impressions sell for exactly one dollar, which is the most common clearing price in aggregate across all the exchanges we work with currently. Roughly a third as many clear for exactly $1.01, so you can tell there are lots of buyers out there that have been learning some lessons from The Price is Right.
We can’t just place bids that represent human guesses– a bid must be customized in real-time to represent our estimate of the fair value of the impression to us and our clients.
I was disappointed to discover recently a quote from my favorite author from childhood, H. G. Wells: “Advertising is legalized lying.” This was more true in his day than in ours, with various tonics advertised in the late 1800’s that were said to cure an implausibly broad spectra of ailments; but not completely foreign when compared with some cure-alls still marketed today. It’s unfortunate that the bad apples are allowed to contaminate public perception of our entire industry.
Critical to an innovation economy is the ability for an innovator to get the attention of the consumers who would benefit from their new products and services. Jeff Bezos, speaking at an Amazon shareholders meeting and apparently feeling pretty good about Kindle sales at the time, said, “Advertising is the price you pay for having an unremarkable product or service.” Some months later he went on Charlie Rose to submit himself to journalistic interrogation on such topics as the battery life of the Kindle 2 and the differences between the $139 and $199 models. Ron Popeil probably could have learned a thing or two from that interview.
George John is CEO and Co-Founder of Rocket-Fuel. Previous to founding Rocket Fuel he led advertising optimization teams at Yahoo!, IBM and salesforce.com

More in Media

Inside The New York Times’ plans to correlate attention levels to other metrics

There’s a lot of buzz around attention advertising right now, but The New York Times is trying to stay grounded even as it develops its own plans.

Why publishers are preparing to federate their sites

The Verge and 404 Media are exploring the fediverse as a way to take more control over their referral traffic and onsite audience engagement.

Why publishers fear traffic, ad declines from Google’s AI-generated search results

Some publishers and partners hope for more transparency from Google and other AI companies related to AI-generated search.