“Averages are made to be broken” is one of my favorite sayings.
By this, I mean that average performance should always be broken down into more detail — giving us greater insight to improve even further. Know how your average TV ad is impacting consumer behavior? Great, what happens when you split that average between linear and streaming TV? How about when you break it out by publisher? And then again by creative? And by audience segment or DMA? With the right granular datasets, more possibilities — and opportunities — to learn and grow will always present themselves.
And across all marketing investments, few areas are more ripe for a broken average than TV ad frequency. For years, we’ve attempted to show the “right” number of ads using broad-stroke media mix modeling tools that suggest the profits a marketer might earn by showing X number of ads, to Y number of consumers, Z number of times *on average*. Wash, rinse, and repeat.
Unfortunately, this uber-macro approach to TV frequency fails to account for the increasing diversity of audiences, media, streaming apps, ad types, and creatives in today’s convergent TV (CTV) landscape. Some of our favorite ads (like GEICO’s delightful “Scoop, There It Is!”) engage consumers time and again (and again), while others might start off hot and wear out quickly. Don’t know which of your creatives is which? Well, then you’re probably leaving money on the table, or worse, unwittingly burning your budget on excess creatives and airings.
TV outcomes are the missing piece of the frequency puzzle
Fortunately, marketers can now go beyond these overly broad legacy tools — and finally optimize frequency in-campaign — at the media, creative and household levels. In a world of tighter budgets and increased competition, this is a great lever for any marketer to immediately improve the ROI of their CTV investment.
The leap forward in frequency optimization opportunity lies in the fact that we now have the metrics necessary to know how EACH ad airing impacts consumer behaviors that are highly predictive of future sales.
Back in the day, you’d have to wait months for an attribution report to get a broad assessment of the *overall average* QUARTERLY reach and frequency targets that might be best for your brand. But now, powerful, predictive TV outcomes data is helping brands see how every ad impression impacts consumer brand engagement. And since online engagement — like brand searches and site visits — are highly predictive of future sales, brands can quickly understand how their campaign strategy and tactics contributed to driving business outcomes.
With this data in hand, we can construct frequency curves demonstrating how consumer engagement fluctuates depending on the number of brand impressions consumers see. And we can see this at the campaign, media, creative, household, and DMA levels.
Here’s a real-life, blinded example of one of these frequency curves — showing how increased exposures impact marginal consumer engagement for a single campaign across specific publishers. Note how these curves differ a great deal between publishers — this variance can also be found across creatives, demographic groups, and DMAs.
For frequency and beyond, more precise data analysis leads to even more effective optimization
Sure, big, broad media mix models allow for guesstimates about how many airings you should show your target audience. But the only way to know what’s actually moving the needle is to carefully track how different strategies drive increases in engagement on an ad-by-ad, impression-by-impression level — and then compare that response against your internal ROI goals. Ultimately, you’ll find yourself optimizing for the RIGHT number of touches for each and every campaign — limiting wasted or underutilized TV spend. Getting this sort of optimization right, over and over again, is far more effective than the outdated set-it-and-forget-it style of one-size-fits-all reach and frequency goals.
The beauty is that frequency optimization is just one of the ways that precise, decision-worthy data can make a huge impact for modern marketers. And the more granular you can get, the more questions you ask, the more likely you are to uncover insights that can make a real impact on your ROI. So if you can slice your averages with a gamma knife, why would anyone even consider still using a sledgehammer?