“On TV And Video” is a column exploring opportunities and challenges in programmatic TV and video.
Today’s column is written by Ian Ferreira, EVP of programmatic at WideOrbit.
Marketers have turned to digital video in droves because it’s supposed to be a great way to laser-target audiences, eliminate waste and place brand messages next to premium content.
The reality is that premium content is hard to come by. While there’s no shortage of opportunities to buy video inventory, the brand-safe and target-matched content marketers crave has turned out to be pretty scarce.
A colleague used to tell me about his experiences buying premium video. Once his carefully targeted ad unit was served to the right audience on the top 10 sites, the 11th placement was likely to be found adjacent to a video of a cat in a Santa hat riding a Roomba. Cute, but hardly premium.
Now it may be time to ask if digital video’s ability to focus targeting down to the individual level isn’t really a consistently attainable benefit either.
This is a tough thing to admit. After a decade in digital ad tech and machine learning, I can’t help but be a big proponent for using data and algorithms to hypertarget consumers. There’s no problem I love more than developing algorithms to find needles in haystacks.
But what do we lose when we shift dollars from traditional media with greater reach and ostensibly more waste to digital? The premise that precision targeting should be the top priority might actually be a loser when marketers can opt for time-tested reach methods.
Some brands have realized that being seen by as many people as possible may still be more important than being seen by only the exact right people. P&G’s recent announcement of a spending pullback on hypertargeted media is the canary in this proverbial coal mine.
There’s always the chance that hypertargeted ads miss huge swaths of potential customers. Facebook, the most targeted and widely used digital medium, reaches only half of Americans. Nobody can tell you how many of its users are concurrent at a given moment in its walled garden.
Let’s take a moment to remember the time-tested power of TV’s reach and quality. In aggregate, there are Super Bowl-like levels of viewership every day. TV is in 95% of US homes. People watch a lot of it: Viewers average more than 35 hours weekly, 90% of which is live. Its ads are the most effective of any medium. And aside from an occasional wardrobe malfunction now and then, there are very few brand safety issues.
Despite this strong evidence of TV’s effectiveness and reach, many in ad tech are still addicted to the idea that their most important functions are hypertargeting and eliminating waste in the name of possibly increasing ROI. The temptation to advance the science of ad targeting is strong. Just as a surgeon naturally has a greater bias to operate than medicate, a data scientist prefers to solve problems with data.
As a result, we have become dependent on technology to solve problems that might be better solved by applying Occam’s razor: The simplest answer is likely to be the right one. I recall a conversation with a friend who was working on a gender classification project for a search engine. His project involved processing millions of rows of data and running a variety of classification algorithms on hundreds of machines. I asked if he could simply ask users to volunteer their gender. His facial expression spoke volumes.
This tendency continues today in the name of evolving digital advertising’s capabilities when the improvements may be too incremental to provide real benefit. Take the recent rise of lookalike targeting technologies that are meant to expand small sets of known audiences to a single large set with shared attributes. This method requires cobbling together a cookie pool large enough to gain effective reach for an advertiser.
Why do this when we already know we can precisely measure users with single cookies? It’s like building a car part by part instead of driving one out of the dealership.
Marketers whose careers paralleled the growth of digital may have reservations about the simpler parameters of TV targeting, including age and gender. They will (rightly) argue, “What is a male 18-34 like? What do ‘average’ 18- and 34-year-olds really have in common? How relevant is a measurement that projects the behavior of one person to a thousand others?”
Fair enough.
But this is no longer an excuse to pass on TV. The tools for finely targeting within TV’s heritage demographic parameters are here now in the form of programmatic TV. Consumer marketers who crave reach along with a measure of hypertargeting should be thrilled that programmatic TV gives them a mechanism for using their hard-earned digital assets to tap into the world’s most premium pool of video inventory in new and innovative ways.
Programmatic gets marketers the two key elements they need for targeting TV audiences with the same finesse as digital video. They gain the ability to draw on the battle-tested data they’ve collected in data management platforms during the last 10 years. They can also buy canonical inventory units asymmetrically with a built-in method for comparing their value.
If there’s anything we’ve learned in ad tech’s last 10 years, it’s that incredible things are possible when data, technology, audiences and advertising converge. It will be intriguing to see the innovation in TV creative and audience targeting when marketers combine TV’s strengths with digital media’s data-driven buying.
Follow WideOrbit (@WideOrbit) and AdExchanger (@adexchanger) on Twitter.
