Coincidentally, or not, this week an investor, an entrepreneur, a vendor, and a writer each enthused about behavioral targeting. They seemed to all believe this was the future of advertising. But while I do believe in the demonstrated benefits of behavioral targeting, I couldn’t help but wonder whether believing in it was like believing in Santa — something you do when you’re a kid, but grow out of, hopefully, at some point.
Past Behaviors to Indicate Future Behaviors?
The premise of behavioral targeting is that past behaviors are good indicators of future behaviors. It’s something that seems easy to believe. But let’s peel back the holly leaves a bit.
Granted, research shows that the brand of car that the parents bought has a strong correlation to the brand of cars the next generation buys. That is brand purchase behavior. But the behaviors used in online behavioral targeting calculations are impressions, clicks, keywords, tags, click-stream (sequence of sites visited), etc. — seldom actual purchases. That’s because the data sets are entirely different and separate. The sites that carry ads or the ad networks that serve ads on them have data on click behavior — what content was on the page, the number of times ads were shown, how many clicks, etc. They may even occasionally have demographic information about users. But those sites are seldom the purchase destinations, so they don’t have data on purchase behavior. Click behavior and purchase behavior data sets are very hard to correlate without using identifiers like personally identifiable information, which cannot be shared without violating some privacy covenant with users. So, the conclusion that past purchase behavior can indicate future purchases doesn’t apply here.
Using past purchases to indicate future purchases is problematic in other ways as well. A customer may have purchased Pampers in the past and plans to buy Pampers again. But will she buy them at Costco, the local grocery store, or at Amazon.com? Who knows? And who should pay for the ads that drove the sale of Pampers at one of these destinations? Another example was a guy in the elevator who had a really cool, tiny Samsung Juke phone. I was about to make a comment about how cool it was, but someone else did first. But the guy immediately answered, “Thanks, but it is the worst piece of sh*t phone ever! Don’t ever buy it.” His past purchase may actually have hurt Samsung’s future sales of phones or possibly other electronics to him. And his sharing of his hatred may have tainted the other people’s future purchases too.
For other products that have long repurchase cycles, like houses, cars, or washing machines, no one knows when the next purchase will be. Perhaps the washing machine doesn’t break down. Perhaps the consumer wants to buy the same brand again, but the stores around him don’t carry the brand. Perhaps the consumer gets a job transfer and has to immediately buy another washing machine for his house in another town. The point is that past purchase behavior may still not be able to predict what brand, when, where, how, or why a user buys the next time around.
Dramatic Lifts in CTRs Still Yield Tiny Results
So, why are so many people festive about behavioral targeting these days? Perhaps it’s those dramatic, multi-hundred percent lifts in click-through rates (CTRs). Right. But don’t you wonder why behavioral targeting proponents only cite lifts and never the actual CTRs (define). Instead they hide behind standard “it depends” or “it’s confidential” excuses. Well, it’s actually because the CTRs are usually a rounding error to zero. Indeed, it is an improvement of 400 percent if the click rates went from 0.001 percent to 0.004 percent. See the growing body of evidence that Facebook’s mega-hyped behavioral targeting system — complete with demographics, behaviors, and even conversations — is not what it was cracked up to be (see the actual Facebook advertising metrics and benchmarks). The low click rates still yield tiny trickles of traffic. Then only a tiny fraction of those go on to actually complete a purchase, resulting in minuscule impacts on sales.
Then behavioral targeting advocates change the subject and talk about the virtues of increasing the likelihood of clicks. That’s fine. But let me put it this way. I’m not opposed to behavioral targeting. Indeed, having behavioral information for targeting ads is definitely better than having none. But that is like saying having demographic information is better than having none because knowing that the user is female allows advertisers to eliminate wasted ads (e.g., showing ads for cosmetics, feminine hygiene products, or Jimmy Choo shoes to guys). But beyond these broad strokes and targeting of large groups of customers, it gets more complicated…fast. And additional targeting parameters move quickly toward the point of diminishing returns. And when behavioral targeting enthusiasts start talking about the “dozens” of psychographic parameters like “dynamism, openness, bravado, and 59 percent more likely to own a Mac” that can be used to optimize click rates, I start to get really hot and bothered.
The Giant White Elephant in the Room Is a Relic
This is all highlighting the problem of the giant white elephant in the room. Targeting is an activity necessitated by “old school” push tactics — from direct mail to television to online banner ads. Since no advertiser has unlimited funds to spend on mailing costs, media costs, or other distribution costs to push a message out at all potential customers, they had to find ways to identify smaller groups who are more likely to like their product and target them with ads. The fact that targeting has helped to eliminate obvious advertising waste is true. But there still is waste. A lot of waste. In the online space, the ability to track details about ads’ effectiveness from cradle to grave reveals an uncomfortable truth. Even with the best targeting — leveraging demographics, psychographics, segments, personas, and behaviors — success metrics indicate that almost all ads are still wasted. In fact, using a generous 0.1 percent click rate means that 99.9 percent of ads’ impressions are wasted. Behavioral targeting may indeed help with dramatic percentage lifts, but the correlation to actual sales impact remains light.
Then there are the related myths of ad networks, view-throughs, and demand generation that need some old fashioned debunking. Some ad networks aggregate billions of impressions and click data but refuse to disclose what sites are in their network, how they got the data, or where your ads are being served. Then they magically deliver lots of clicks and claim massive percentage lifts. Savvy advertisers know to stay away. Other ad networks are so desperate to explain why CTRs are so low that they have stooped to fungible mathematics called “view-throughs.” They claim that if any user saw any ad of the advertiser anywhere on their network and went to the advertisers’ Web site at any time in the future (the time period is negotiable), the ad network gets to claim credit for a click-through — or view-through because there was no click-through. Did anyone ask whether the person actually saw the ad even though it was displayed? Or did anyone think that the person went to the site on his own, not because of the banner ad? Finally, demand generation. Some still believe that advertising can generate demand (i.e., by beating lots of people over the head lots of times). Just like the fuzzy math of view-throughs, they claim that repeated impressions have an impact on clicks, brand sentiment, demand, or even sales. Did anyone think that the user bought the digital camera because his friend recommended it and not because of being exposed to thousands of banner ads he didn’t see?
Out With the Old, In With the New
If we toss targeting, including its most sophisticated and modern form, behavioral targeting, what do we have left? Is there a better way? Yes, there is. Will this new way solve the inabilities, inadequacies, and ineffectiveness of targeting? Yes, it will. Well, to be honest, I can’t tell the future; nor, I suspect, can most other people. But by looking at search terms, I can tell exactly what an individual user is looking for in the present. The exact term reveals what stage of the purchase funnel they are at — for example “what kinds of motor oil are there?” says they are in the awareness stage and just beginning to do research; “which motor oil is best for my car?” reveals they are further down the funnel, in the consideration stage; and “where do I buy?” and “what is the best price?” means the purchase is imminent.
By looking at search behavior, advertisers no longer need to guess which large group of customers may like their product; advertisers know who is looking for something about their product. Advertisers no longer need to guess when they are going to purchase. We can tell who is where in the purchase funnel and perhaps focus our energies on the ones that are near the imminent purchase. Advertisers no longer need to find who to target by using more and more esoteric (i.e., removed from reality) parameters and behaviors or how to “push” the message out to them. The “targets” self-identify and come to the advertiser — they “pull” (i.e., search) for the information when they want and need it; they aren’t offended when the information they’re seeking is delivered to them. This even solves the timing issue — when exactly is this user going to buy his next washing machine (and any privacy issues)? Users pull the information and advertisers don’t have to use personal information to correlate unrelated data sets for the purpose of “targeting.”
So, why target a group (of users) with approximations based on made up parameters when you can “target” an individual with precisely what he needs, when he needs it, where he needs it, and how he wants it delivered (e.g., on a mobile device)? I guess some people still believe in Santa.
Happy Holidays. Chat with y’all in the New Year. Peace.
New Top-Level Domains (TLDs) have become more popular in the last couple of years, so here’s everything you need to know about them.
Amazon Prime was launched in 2005 as an express shipping membership program and more than a decade later it has tens of millions of subscribers who enjoy a lot more than just free, fast shipping on millions of products Amazon sells.
Sure, some apps are doing personalized push notifications, but what happens when your users are in the app?
Since cloud computing first gained mainstream attention around 2009, its popularity has exploded. Promising increased efficiency, flexibility and cost-effectiveness, it was hailed as the ultimate business solution. But are users seeing the benefits?