Can Your Man Beat My Machine: Lawd, -- Lawd?

There are those in the online advertising industry (well, maybe more on the periphery) who are convinced that all of the effort and money thrown at trying to technologically optimize a client's ability to reach the right person, at the right time, with the right ad, is pointless. To substantiate their opinions (which tend to support businesses that don't rely so heavily on said technology) they point to low click rates. They also point to the Great Pumpkin and Catherine O'Leary's cow, but that's for another column.

The "art vs. science" debate, which seems to warrant at least one session at every ad conference from here to Beijing, centers around the growing encroachment on both the media and creative sides of the ad biz -- with some arguing that machine-produced imagery and copy can be just as effective as Don Draper copy, since it can be constantly refined and adjusted to provoke the intended response. Moreover, persistent advancements in targeting -- be they contextual, behavioral, semantic, intender or knowledge gleaned from social interactions -- only help the ads become more relevant to the users. The art side argues that in all this we are losing our humanity, compromising user privacy and probably spreading swine flu.



What then is to be made of this week's study from Dynamic Logic that says it's bad creative that makes online advertising ineffective? After analyzing the highest and lowest performers from its database of more than 170,000 online ads, DL determined that creative factors such as persistent branding, strong calls to action and even human faces -- and not super-targeted or high-profile ad placements -- make for better ad recall, brand awareness and purchase intent.

"But media people spend the most time trying to optimize and measure campaigns, because the creative quality is outside of their control. If they got good ads to begin with, that would help," Dynamic Logic's Ken Mallon told reporters.'s Scott Briskman said to Ad Age: "The media buying and planning are the delivery system. It doesn't matter how good your delivery system is if the creative sucks."

For decades, the ad industry has tried to use science to inform its creative, going so far as to scan audience brains to see if they react more positively to one creative option than to another. They have tracked eyes, run a million focus groups and even checked vital organs to see if they speed up or slow down in response to this message or that one. So it seems to me that we have never been too far from science on the creative side. The problem has been that nobody has built the algorithmic ability to come up with "Things Go Better With Coke" or "I can't believe I ate the whole thing!" Not yet, anyway.

On the media side, I see no reason not to automate the entire process as soon as traditional media can digitize its delivery systems and we can keep humans from screwing up what should be a pretty straight-ahead equation. "This amount is a fair price to pay to reach an audience segment consisting of men 18-24." This broad demo can be further refined by the user data collected by the various targeting providers. Yes, more technology. And still, technology can assure the ads actually run, run in a safe place, and are run in such a way that they stand the best chance of success.

On the creative side, there are tech providers who claim they can change the copy and imagery on the fly in order to eliminate unsuccessful ads and replace them with units that pull higher response rates. Is this just too much hair-splitting to matter? Perhaps. But if, as a marketer, you can start with a good basic creative idea and let machines tweak it a hundred or a thousand times and in the process teach you something about your audience, why wouldn't you welcome it? Especially since the cost of doing this is ridiculously low (for now).

It is said that if you put enough monkeys on enough typewriters, eventually one of them will reproduce Shakespeare's prose. One might think it possible to come up with "Fly the Friendly Skies" with considerably fewer monkeys.

3 comments about "Can Your Man Beat My Machine: Lawd, -- Lawd?".
Check to receive email when comments are posted.
  1. Scott Brinker from ion interactive, inc., October 23, 2009 at 7:21 a.m.

    Always an insightful column with an amusing tone.

    However, one serious question: when you "let the machine figure it out," how many bad -- terrible, awful, offensive -- test experiences are you willing to let it subject a small slice of your audience in the quest for the perfect combination? Sure, the fact that the people exposed to that message bail right away and never come back leads the algorithm to eventually remove that combination from the trial space. But was the cost worth it?

    BTW, I'm a technologist more than a marketer, so this isn't a hang-up of "the good old days of Mad Men." It's a real conundrum with using real audiences as lab monkeys -- without their knowing that they're participating in an experiment.

  2. Paula Lynn from Who Else Unlimited, October 23, 2009 at 10:05 a.m.

    Then there are those who follow the others who preach ditherings from psychotic episodes (or phoney baloney psychotic self delusions): Lawd,--Lawd? George, this is an excellent, logical column and not "over the line".

  3. Dominic Schmitt, October 23, 2009 at 12:13 p.m.

    Start w/good creative and let the machines teach you something about your audience. That is truth in advertising and, more than any platform or invention, the dividing line between 'new media' and 'old media' mad men.

Next story loading loading..