Advertising as an Automated Task
A warm welcome to our new algorithmic overlords.
A warm welcome to our new algorithmic overlords.
Archeologists have evidently, found ads in the ruins of ancient Arabia. They’ve found ads on the sides of pottery and on papyri that date back thousands of years. There’s a particular urn from ancient Greece depicting a horse, along with the text (and I’m not making this up), “Buy this and you’ll get a good bargain.”
Well, that’s all about to change. The urn and papyrus painters, as well as the TV spot creator and the banner ad designer, are about ready to meet some stiff competition in the form of long, complicated, nearly-impossible-to-penetrate arithmetic problems. We’re beginning to see the emergence of a brand new form of advertising in which human beings are in the back seat and the technology tools are up front.
Quick: what’s the bill-rate of an algorithm?
From Spot Runner to Blade Runner
This scenario, in which humans stand back while algorithms run the show, is going to happen. I can accurately predict that the Sun will rise on this brave new world sometime between a week from next Tuesday and the very, very, way-far distant future.
You heard it here first.
Hyperbole aside, I don’t truly believe any of us are out of jobs just yet, but we are starting to see the technology’s role emerge from the shadows of measurement into the bright light of execution. While we, as advertisers, have always used technology, traditionally it’s been solely to make ourselves smarter.
We’ve used research services to perform data dives to identify segments, viewing habits and purchase cycles. That helped us make smart decisions about where we’ll buy media, as well as the messaging we might use.
I talked to someone the other day who noted the target of his campaign way over-indexed on CSI-type shows. His product is pretty techie in nature. This led him to develop messaging focusing on the methodical problem solving style found in those shows.
On the back end, we’ve used technology to tell us how well a campaign performed. With the Internet, we’ve had no shortage of performance data, from impressions clear through to conversions. There’s a number if you want it.
Execution? That was in the hands of the people at the agency. They were the ones making the calls on not only where and when, but also (and this is the big thing) what. By this I mean, “What will we say to these people to get them to want to buy our product?”
We’re starting to find ourselves in a strange situation. A group of very smart people who decided to study engineering and math have become interested in advertising. Or, as they’d say “the problem of advertising.”
The “Problem” of Advertising
The “problem” with advertising is defined quite differently by this group. To them, a “problem” is a thing to solve. Anything can be solved, provided the problem’s clearly defined. That is, with enough data, you can figure anything out. Like how to clone a sheep. Or how to get to Mars. What neutrons are made of. Or what gets someone to click and buy a new laptop, mortgage, or vacation home.
Thanks to the Internet, these people now have data. Lots and lots of data. They can look at that data and begin to find patterns. The Internet is the world’s largest data collection outfit, and the scientists who found the problem of advertising are putting on their helmets and going spelunking.
Consider this: how many ads are running, right now, on the Web? Thousands? Way more than that. Each one was built by a clever mind with his or her best guess on what will motivate a consumer. Now, how many are for a similar product, say laptops? A decent number. Let’s say there 100 distinct campaigns are running. Most of will fail, of course, and a small number will be successful.
That’s an enormous amount of generated data. The one missing piece from this data set is the creative. You can tell from data alone which ad was most successful, but you can’t tell what was in that ad (unless, of course, you program that parameter in).
Did I mention Google (where a whole lot of scientists who are focused on the problem of advertising work) just launched the beta of their Web Site Optimizer? This is a tool, similar in concept if not in scope, to offerings from companies like Offermatica (which pioneered the space). It lets you see what bits of creative are most effective. Yahoo has similar functionality, too.
I, for one, welcome our New Algorithmic Overlords. This means the last piece of the puzzle — creative –has been converted into a bit of data, making the problem of advertising that much more solvable.
As humans, we’re not finished. I really, really am not suggesting we just click a few applications and let the software do the work. You can always tell when an ad contains the light of consumer insight and brilliant creative direction. Such ads will always win.
We must, however, begin considering how technology can help in that execution phase. There are patterns to human behavior, and algorithms can help to sort them out. It certainly is an opportunity to increase the relevance of our ads.
And yes, we are still the ones in control.
Cue: diabolical laughter.