Market only to humans.
If you only change one thing about how you undertake your marketing in 2018, let it be this one.
According to the U.S. Census bureau, there were about 252.8 million adult (18+) humans in the U.S. as of June 2017. As of June 2018, there will be about 255.3 million humans, an increase of about one percent.
These are actual living people, with names and addresses, who live in a particular location and who purchase actual goods and services. It doesn’t matter how big the “audience” is at the DMP. It doesn’t matter how big the cookie pool is. It doesn’t matter what the match rate is.
There are about 253 million adults in the U.S. that you can market to. Period.
And unless you’re P&G or another FMCG company, you’re marketing to a smaller group of category buyers. Say it’s 35 million or so, which around the number of people that golf or take cruises.
If you are not starting with an actual list of names and addresses of those 35 million and instead are starting with “signal data” or “audiences” at a publisher, ad network, DMP, DSP, DWTF or whatever, you are getting ripped off.
Let me say that again. You are getting ripped off.
Why? That’s because you are working with spurious signals that are attached to a site visit–probably accidental, if even real–and then modeled via some unknown modeler, with knowable biases (they want you to buy their media, after all) and then “expanded” via further unknown and biased algorithms to increase the size of the audience.
The next time you’re tempted to market to an “audience” ask the following questions of the person trying to sell this to you:
- How many unique humans will I reach?
- How many of those unique humans will be in my category buyer segment?
- How do you know?
- Who created the audience model?
- Where did they go to school and can I actually speak to them?
- What are all the data elements in the model?
- Where did you get them?
- How did you verify those data elements?
Then ask them to prove to you, by examining and showing you a random sample of 1,000 of the cookies/IDs/matches that they are actually humans and that they are in a segment of your category buyers. If that can’t be done, or you get some version of “you don’t get it, man” or “so what, it’s cheap CPM,” walk away. You’re talking to a huckster.
Think those questions are too time-consuming or too embarrassing to ask? They were only asked millions of times by smart marketers at Time Inc, Reader’s Digest, Columbia House, Lands’ End, etc. to create, oh a few hundred billion in revenue (at least) and actual, real bottom-line profit. If you don’t ask those kinds of questions, you’re going to get whatever the algorithm spits out to you. And you deserve it.
Takeaway: Be successful in 2018 and market to humans. Start with names and addresses, not cookies and signals. And win.
Happy New Year!
Thanks for reading this far. A couple of thoughts on the future of digital marketing, since it’s the time of the year for us to make all our B.S. predictions. In no particular order, what I see for 2018:
Devaluation of data — Almost every app and social media service has, as part of their business plan, “monetize the data.” After all, when you don’t charge anything, you have to rely on surveillance and then resell that information to pay the bills. With more apps, more services, and easier-than-ever data scraping, the amount of data grows every day. The truth of the matter is that only 3-4 variables work in most models, and the number of marketers who actually want to sell goods doesn’t grow as fast as surveillance. We’ll see more and more data sets chasing the same marketers. Each of those sets is collected in a biased way, modeled in a biased way, and expanded in a biased way and sold in a tonnage model. As that supply chases demand, the price goes down. As does the quality. (This is why I like to start with name/address data, which is the best and hardest to fake.)
And by the way, why would you ever look anywhere else for data about what we buy and look at than Amazon? Not only do they know what we consider and not consider, they’ve got the purchase history of things we don’t buy at Amazon from their co-branded credit card. No more guessing. Not that they’d sell you the data, however. And do you think Amazon takes “you don’t get it man” or “so what, it’s cheap CPM” in their internal meetings?
IoT explosion pollutes ad tech — The predictions for IoT devices keep growing. We probably added 20-30 million Amazon Alexa surveillance devices to our homes in the U.S. last week, and added all manner of light bulbs, doorway cameras, etc. to add to our home networks. All made in China and most with some version of “admin/admin” of username/password on the firmware. Forget being DDoSed by light bulbs. Why not convert those IoT devices to botnets that view pre-roll at $10 or $20/M?
Video fraud explodes — It seems nobody trusts programmatic display, so the next move is buy digital stuff that looks more like what you’re used to buying. That’s TV. So even though YouTube and Facebook have proved to be untrustworthy in the inventory they sell us, advertisers continue to rush to that same inventory. The bad guys are rubbing their hands with glee. The only thing that might keep them slightly away from video fraud is bitcoin fraud. Although maybe they can combine bitcoin with ad fraud… Hmm, I’m sure somebody’s working on that in Belarus right now.
More OCEAN targeting — We’ve seen how successful targeting people based on the Big 5 personality traits (openness, conscientiousness, extraversion, agreeableness, and neuroticism) has been on social networks like Facebook. The Russians got their guy elected for just a few bucks. While I don’t agree with using people’s personalities against them, I suspect aggressive marketers will be using psychology-based tactics this year. If you think I’m wrong, read Bob Cialdini’s classic “Influence” and tell me if a better laboratory than Facebook was ever invented to test those theories.
Audio search explodes — It’ll be a while before Amazon makes the duopoly a troika. But I suspect that because yelling into an Alexa surveillance device is easier than typing, we’ll get more completed searches there, i.e. concluding with a purchase. As Scott Galloway predicts, search moves to Amazon. And Amazon will have all that data.
More ad fraud — Sadly, while I think Marc Pritchard’s call to arms last January was a good start, I think we’ll uncover even more fraud in 2018 than 2017. Most will be of the video variety, but we’ll still have the standard display fraud. The reason is that most marketers won’t heed my call to start with targeting humans and, in the face of exploding options for buying media, will become more reliant on algorithms to make decisions than ever before. The bad guys will be all over that failing.
GDPR will be a big success — And I don’t mean just for privacy. I think we’ll see a renaissance of proper marketing in Europe. Sure, the lazy idiots who just want to follow you around the Internet with more lousy and intrusive data will bitch. They’ll go the way of the dodo bird and proper marketers, with proper strategy, will win. We’ll see an explosion in quality creative that sells while people have less of their personal data collected by irresponsible and shady ad tech and mar tech players. We will will see growth in advertising and a (probably slight) decrease in unwanted surveillance. With any luck, people will take notice and we’ll finally get some movement toward something like GDPR in the U.S. Although I’m sure I’ll be dead and buried before out bought-and-paid-for federal legislators consider it. Maybe states like California and Massachusetts will take up the mantle of GDPR. We can only hope.