Marketing Beyond Predictive Analytics
"Predictions of socioeconomic variables are as dependable as the Horoscope so let’s not rely on them…Instead we will need systems that don’t fall apart when we make a mistake."
There has never been a better time in the history of the world to be a marketer. Yet there is a great crisis in marketing today. The crisis is quite different than what you are likely to read in the trades or encounter at conferences. From my perch, the calamity facing marketers is what they have come to believe is the great savior; predictive analytics [PA].
This belief has become more evangelical than the most fervent Sunday morning experience. It has caused marketers to forgo break-even metrics for future oriented lifetime value models. Why not? With PA, one can glean the future with the latest, greatest algorithm. But in reality, PA is a metastasized cancer that is eating marketers from the inside out. Let me explain.
As the quotation from Taleb suggests, we have a real problem of dependability. He was not speaking specifically about marketing, but the issues are the same. Yet everywhere we look we have more and more resources being allocated to PA without the slightest understanding that for the most part it is unreliable.
A quick look at the scorecard is in order. We see major online marketers who swear by PA losing money hand over fist in spite of supposed PA prowess. Really? Yep, even the giants like Walmart and Amazon lose money in their online marketing pursuits.* In fact, a cursory glance at the top Internet Retailer Top 1000 retailers finds marketer after marketer awash in red ink.
I have written elsewhere why PA is snake oil (called Behavioral Targeting before) and this is article is not going to dissect the algorithmic nonsense again. This is about what we marketers should be doing instead of chasing the string of the latest greatest analytical tool.
So what are the options for marketers rather than stalking consumers in the vain hope of discovering a magic elixir? The first and foremost thing for marketers to do is stop for a moment to reflect on the entire media ecosystem. Here is a relevant quote from Nate Silver’s, “The Signal and the Noise”:
“…in a market plagued by asymmetries of information, the quality of goods
will decrease and the market will be dominated by crooked sellers and
Online media is such a market for lemons. With online media fraud nearing 35%, bad media should be driving prices ever lower. Media pricing would fully collapse if not for the asymmetry of information marketers suffer. How often are you informed about media supply? It is a great mystery. One “savvy” media buyer told me he thought Facebook was 100% sold out. Have you ever received an email where you were tagged in an FB post? Notice that this has no advertising.
Once one looks at the media landscape, one begins to understand that the number of advertising options available is overwhelming for today’s marketer. The more advertising options available to marketers, the better it is for them. In fact, there is such an oversupply that it boggles the mind. It is taking advantage of this oversupply where the real action is, not algorithmic prowess.
In the US, online ad impressions are approaching 10 trillion per month and the increases in impressions are more or less following Moore’s Law by doubling every 12-18 months.
Another part of the ecosystem, the consumer, is also changing. Joe Smith from Paducah, KY visits 99 different websites each month. Yes, he visits Facebook, Google, Yahoo, AOL, and other top ten websites. But Joe Smith can be reached at dozens of other sites. He is the same Joe wherever he goes. What if I told you that marketers can reach Joe on one of those other non top ten sites and pay 90%-95% LESS than the cost of reaching him on Facebook?
OK, a little more information to ease the asymmetry. 90% of all online advertising cash buys goes to 10 websites! There are 232 Million ACTIVE websites in the world with about 90 Million ACTIVE websites in the US.
So if 90% of cash buys in the US go to those top 10 websites, this leaves 89,999,990 websites scrambling for the other 10% of the cash. Anybody reading see the opportunity here? Why compete on those 10 websites where the cost is so high? Advertise to those eyeballs at a fraction the cost somewhere else. It may be more work, but who told you that success was easy?
Now, another question for marketers to answer; what percentage of monthly advertising impressions remains UNSOLD? I do not know the exact percentage, but industry pundits proudly tout that online media cash buys are increasing at 12%-20% per year. That is impressive…until you understand that inventories are doubling every year. Online publishers are choking on their supply. So much of what media companies have available is not being monetized.
The lack of information on media supply has marketers chasing string with costly algorithmic solutions trying to make FB, Google and AOL more efficient. Marketers should be LOWERING their allowables rather than trying to find ways to justify raising them.
Before you spend precious money trying to predict what will be successful, crush down the cost of failure. A great place to start is decreasing the media rates you pay by 90%. You cannot imagine what a simple change from concentrating on “success” and switching to limiting the cost of “failure” can do for your business.
*Amazon reported a profit last quarter but this was due to Amazon Web Services (AWS) making up the loss for their online retail activities. Their predictive analytic models got the Fire Phone horribly wrong and lost hundreds of millions of dollars.
** Silver is distilling Nobel Prize winner George Arkelof’s “Market For Lemons” here