- Is DAI Already DOA? (AdExchanger, Dec. 15)
I am likely to spend quite a bit of my time in 2017 attempting to educate the more youthful and considerably less experienced in our industry on the folly of believing everything they read about the power of data and technology to solve what appears to be any problem.
Truthfully, my more youthfully exuberant colleagues have no understanding why some things work the way they do, how some things have worked for decades and that those things don’t necessarily need to have today’s tech applied to them to fix what isn’t broken. To wit, dynamic ad insertion (DAI) is not dead on arrival (DOA) – though perhaps it should be.
The first and most alarming aspects of DAI are the “fixes” it supposedly brings to TV. Actually, TV isn’t broken. Well-targeted mass-reach delivers scale that literally makes markets. Choose a business category, any business category where TV has been used, and you can see how the sheer scale resulted, for the best creative, in making something out of nothing. Where no business previously existed, a new one sprang to life.
This is not because of hypertargeting – precisely the opposite. Human beings analyzed the available data and determined that the best approach was to use mass-reach TV in order to impress a mass market. The reason? It’s not because we don’t want to have better data; it’s because we know that businesses are built not just on immediate response but over weeks, months, quarters, years and decades.
There are so many unpredictable human complexities inherent in every market that there is no effective way to target just the buyers and have it work to the scale needed to pay for the production and the advertising. The only way the numbers actually work is with the human capacity to come up with a well-targeted group of mass-scale outlets so that the response rate can cover the costs associated with the ad and advertising. The whole idea of paying a premium to super-selectively target an ultra-narrowly focused group of consumers who fit a tidy little set of massive data selects is absurd.
Want an example?
Let’s try this hypothetical: Our advertiser is a home-warranty company. Let’s say it spends $20,000 producing a TV commercial. (That’s an almost laughably small production budget, but it’ll do for our purposes.) And then it produces a second version of the spot for another $15,000. That’s also absurdly low, but for this example it lets us see that we’ve got $35,000 invested in TV production. Now we go to market to test which spot will work better. We decide on a weekly budget of $50,000 with which we might be able to buy three national cable networks delivering about 18 percent of the client’s 35-to-54 target market with weekly frequency of about 2.0.
Leveraging Mass Reach
We’re now $105,000.00 in and don’t have the first sales dollar yet, but our media plan will deliver well-targeted reach in excess of 20 million households (homeowners ages 35 to 54 with income of at least $75,000) per week. The first week’s spot produces a meager tenth of a percent response rate (1,000), and the client converts 23 percent of callers (230) at an average sale of $799.00, or about $183,770 in revenue. Week Two’s ad doubles the response rate (2,000), which the client converts at a lower, 17 percent (340) rate at an average of just $699.00, for revenue of $237,660. Both weeks the client is in the money. The return on investment (ROI) is through the roof, and they keep on buying, adding staff and spending more and more to replicate these efforts. Obviously, with increased scale comes softer conversions, but they’re still at a 2-to-1 ROI.
Now we take the same scenario but use hypertargeting to buy only hypercostly “premium” inventory based upon what the ad techs tell us are 12 different, critical pre-selects for this consumer group. The investment is identical to that in our first example; we’re in for $105,000. Our hypertargeted campaign reach is a little more than 5 million households, but it projects a four-tenths of 1 percent response rate and anticipates a client-conversion rate of 30 percent. These are real estimates by the way, not made-up figures. So let’s do the math. It should yield us an impressive 2,000 respondents, with 600 converting at $799.00, producing a whopping $479,400.
But it fails to meet projections by 50 percent at each stage of the funnel, yielding just 1,000 respondents and 150 conversions at $799.00, for $119,850 – and an ROI of 1.13 the first week and 2.28 overall, compared with 4.0 in the first example. That is because you cannot parse out all the undesirable elements of a target market without taking the segment of the market that is responsible for producing the most profits. The latter are the engine that leverages all the power in business and make rapid and successful expansion possible. Why? Because you cannot accurately predict human behavior or motivation from past behavior.
The securities industry, and nearly every other business sector, gets that and disclaims the notion that past behavior must not be used to measure future results. Why is it the advertising industry persists in demonstrating its complete lack of business acumen, focusing instead on fairy tales and myths to earn its living?