I have news for you. Many of you were misled in 2016. You bought questionable digital inventory to get a slightly cheaper rate, used probabilistic data, but thought you were getting deterministic, reached a tiny fraction of your target universe, and got some really bad advice.
Unfortunately, this was all too common a problem. This cycle saw an unusually high number of new entrants into the consulting market, particularly on digital. Knowing there was a flood of newcomers, with irrationally large sales goals, it’s easier to understand why our credibility is threatened. My view is that the best thing we can do as an industry is call BS.
The cheapest inventory providers aren't giving you a better deal. They’re just selling skirt steak and calling it Kobe strip. You can have affordable inventory, high viewability inventory, and high reach inventory, but not all at the same time. The best vendors will help you to prioritize two of those. The worst will promise all three.
Since the vast majority of our political and public affairs buying this year was aimed at persuasion, our priority was typically maximizing reach. As much as we’d like, it’s virtually impossible to rely solely on premium, private marketplace deals with a narrow target set and a narrow timeframe. You should use caution when publishers suggest the opposite. In fact, I’ve seen them supplement first-party data with lookalikes, just to meet a delivery goal.
Lower quality data and fuzzy targeting
We utilize tens of thousands of data points, mainly based on commercial behavior and demographic data. Modeled data is, by definition, less precise. So when a vendor offers pre-constructed data on an insanely-narrow issue, they may be fibbing.
National surveys scaled to congressional districts or zip codes are much cruder than their vendors may lead you to believe. But it's an attractive pitch and I understand why lesser informed clients are excited. Last cycle, I lost business to a nonpartisan targeting shop offering a pre-constructed file of “people who are interested in infrastructure.” Folks, that’s absurd. Issue data so specific is likely modeled on demand to keep costs down. As an alternative, we suggested targeting known car owners who commute more than thirty minutes. Alas, that didn’t seem as sexy.
Now, there are worst practices going on than just putting lipstick on a pre-constructed pig of a data file. In 2015, we found a general consulting shop pretending to offer Audience Partners’ targeting solution. Except they weren’t, and their customers were likely buying cheap, non-targeted inventory. It was the first case I’ve seen of possibly counterfeit voter targeted digital ads, but it likely won’t be the last.
Buyers may have also misunderstood the precision (or lack thereof) of IP targeting. Only IP data provided by the carriers is accurate at the household level. We create Targeting Pools for most carriers but that is just now beginning to enter the market. A lot of you bought IP targeted ads without using carrier data and hit the right zip code, if you were lucky.
Smaller reach with any single tactic
Pick one tactic plus any narrow targeting segment and you've reduced your reach significantly. Mobile-only solutions are no better than display-only solutions. Twitter’s reach among active users is less than 17 percent, mainly reaching the youngest and least persuadable voters in America.
The median likely voter in America is 59 years old. But according to comScore, 93 percent of Snapchat users are under age 55. And video-only solutions over a short period of time can have reach issues when small segments are overlaid. Even Facebook-only advertising limits your reach. The point is “all of the above” solutions nearly always give you the biggest reach.
And lastly, any single DSP will also have reach problems. Multiple DSP’s allow you to aggregate more inventory, higher reach, and drive higher frequency among targeted voters in a tighter time frame. Just because you’re targeting all likely voting independents, it doesn’t mean they are all on any single platform.
Ugly programmatic TV
Programmatic television today is terrible. Again, you have reach, quality, and affordability, but you can only pick two. Today, eMarketer says that total programmatic television spending will be about $700M. That’s nothing.
It’s moving slow because of technical hurdles, hesitant television networks, and increasing fragmentation in how we consume content on screens. Spending on programmatic TV will more than double next year, according to some estimate, and then double again in 2018. Between 2018 and when programmatic TV actually takes off, it will take several rounds of boom and bust before supply and demand converge. The solutions you were pitched in 2016 will be replaced soon.
Pile on low-reach formats, junk data, and inexperienced media planning and you've got crap on crap on crap. Roughly 95 percent of our political and public affairs business runs through consultants, and the best ones ask questions. Don’t be afraid to learn how targeted ads are delivered and challenge conventional thought.
The takeaway from 2016: If it’s too good to be true, it probably is. Folks who play the long game, build trust. Still, it’s good to say no sometimes, even to those you trust. It helps avoid mistakes.
Jordan Lieberman is politics and public Affairs lead for Audience Partners (@AudiencePartner) and the self serve platform CampaignGrid Direct.