Data Mining burst onto the scene as an exciting database technology in the 1990s. Businesses with big databases, (big retail, big banks, big insurance, etc.) snapped up the data mining technology — statistical software by another name — so they could hunt for valuable correlations they believed to be hiding shamelessly in their gargantuan databases.
At the time, I remember talking to a senior IT guy who worked for Littlewoods, a UK catalog company. He told me his firm had been applying statistics to their data (i.e., data mining) for years and had indeed unearthed useful correlations. For example, they had discovered that women who buy black leather furniture also were likely to purchase red high-heel shoes. The company had assembled many such correlations and consequently was able to target customer promotions effectively and profitably.
Great Minds Think Alike
“There’s nothing new under the sun,” as it says in Ecclesiastes, and that seems to be the case with digital data and statistical techniques. As time passes, the same statistical algorithms re-emerge, often with new marketing handles. First, it was called statistics, then it was data mining, then micro-targeting, then machine learning, then AI, and mostly it was the same old math with a different hat on — algorithms all the way down.
When the Dot Com revolution took off, the commercial joys of Microtargeting were championed, and data mining/statistical software experienced rebirth under a new marketing banner. At the time (around 1999) there was a website that caught my eye called MovieCritic.com. It was free to use by anyone who happened by. In reality, the website was a product demonstration for the categorization and correlation software that a company called LikeMinds had brought to market.
To me, MovieCritc.com was a tour de force. It proved, much to my amazement at the time, how little information you needed to gather from one individual for mathematics to work wonders. Here’s how it operated: You registered to have an account. You then had to identify movies that you had watched and rate them on a scale of 1 (very bad) to 10 (very good). To make accurate predictions for you, you needed to rate 12 movies accurately. That was it. Just 12.
It would then be able to tell you which movies you would enjoy. And dammit, it was fiercely accurate. And here’s the truth, it knew more about what appealed to me than I did. For example, I have always disliked musicals, so I asked it to recommend musicals for me. It recommended Tommy (The Who) and Paint Your Wagon. And it was right. Despite my dislike for the genre, I’d seen both movies, and I had enjoyed them.
Sadly, MovieCritc.com disappeared from the Universe when Adobe acquired LikeMinds and took the website down.
The Power of Permission and The Need For Stats
At Algebraix we have built a permission-based ad market where content providers (advertisers) reward consumers for their attention. By the way, if you think advertising can’t be entertaining, go visit Superbowl-ads.com and watch a few ads, or a movie trailer or educational video on YouTube. Or listen to a song on Spotify. Pop songs are, and always have been, ads for the artist and the album.
We do not think in terms of ads, we think in terms of content. And there are many websites where users seek promotional content: Google Search, Yelp, Craig’s List, Groupon and others. We will be different because we will emphasize entertainment but the principle is similar. Indeed, we believe we can change the game. Advertisers will want to make their promotional content entertaining and educational. (I’ll say more about this in a later post — but if you want to get a hands-on idea, join our beta test program)
Permission marketing provides the foundation for this. Permission marketing will turn the tide. It will launch a new era where advertisers seek to engage rather than confront their potential customers. And it will work its magic with a little help from its friends: personal data ownership and data analytics.
We need:
- To take back ownership of our data (see Why You Need To Take Ownership Of Your Data for the big picture), and
- We need to deploy the heavy weaponry that data abusers like Facebook have traditionally deployed against us.
Providing the individual with complementary statistical firearms is more than a neat idea, it’s a necessity if we are going to level the playing field. The big aggregators continually harvest mountains of our data and comb through it repeatedly so they can inundate us with unwanted ads. We have become ready targets for the slings and arrows of outrageous algorithms.
When Spider Webs Unite, They Can Tie Up A Lion
We need to cooperate and collaborate. We need to build similar aggregations of (anonymized) data and massage them with our own algorithms for both our own individual and collective benefit. We can do that, by the way. We do not even require data that would identify us. We only need behavioral data (surfing the web, making media choices, making product choices and so on) and we need to aggregate it.
As Vanity Fair observed , “The more you use Facebook’s products, the more Facebook’s products get to use you, collectively knowing more about you than you know about yourself.” Let’s stand that on its head. Let’s begin by analyzing data about ourselves for our purposes. Let’s use the clever techniques of MovieCritic.com to discover what our true preferences are in every kind of choice we make. Let’s discover all our “red shoe correlations.” Let’s assemble a comprehensive set of profile information about ourselves, and then let’s choose how much to reveal to those who wish to do business with us.