Today’s post features marketing history, mystery and generational oversimplifications. The question before us is this: What the heck happened to marketing data?
30 odd years with some of the biggest compilers and aggregators of marketing data provided me with a lot of insight into what I’ll call the Old World of marketing data and some knowledge of the New World. Today, I’ll talk about the key differences, and offer a few ideas about what’s fueling the change.
Customer Data is TRANSACTIONAL
Items, Categories, Prices, Dates, Discounts, Promo/Effort Codes and many others.
3rd Party Data is mostly Small, Powerful, Stable and DESCRIPTIVE
Gender, Age, Income, Presence of Children, Home Ownership, Hobbies, Interests, Cars, Marital Status…and maybe 1400 more.
Data is used for simple selection, segmentation and predictive modeling. EVERYTHING is matched to an Individual, Household or Living Unit.
Old World Targeted Marketing
Today’s Marketing Data is Big, Fast, Weak, and BEHAVIORAL.
Real time Web Browsing/Navigation data, Where you’ve been, How you got there, Where you are (both virtually and physically), Everything you’ve posted on social media (analyzed for direction of sentiment, intensity, keywords, context and more), Your photos, Profiles, everything you’ve bought with a card, and much more. Each individual data point (such as the length of your last online session) is weak in and of itself. In aggregate, the data is quite strong.
Data informs a set of algorithms which generate an ever-changing profile which may or (more likely) may not be tied to your offline identity.
While there are many other uses of data, we’re mainly concerned with how marketers put the right message in front of the right consumer at the right time. For this purpose, and because the number of impressions (the placement of a message before a set of eyeballs) is still a cost driver, both approaches need to have predictive power. Digital advertising is now offered on a Cost Per Action e.g., cost per click thru, cost per sale closed, cost per inquiry, etc., rather than the old Cost Per Thousand of the print world. The more success you want, the more it will cost. The more precision you want, the more it will cost. The new paradigm is almost entirely performance based.
Let’s say you are marketing fashion for young women. If you mail 100,000 women’s fashion catalogs to random living units, you will get some response, but your level of uncertainty is 100%. Same with web marketing. If you were able to place your ad for women’s fashion on 100,000, random devices , you’d get some clicks and make some sales, but how many and how much would be unknowable in advance. Marketers earn their keep by reducing uncertainty - by stacking the deck, if you will.
Traditional database marketers use sales data to segment their existing customers. They typically split them into buckets by Recency (of purchase), Frequency and Monetary amounts (RFM). Generally, a group of customers who bought more stuff more recently, and who have bought more than once will buy more than a group of customers who bought one item over a year ago and have never bought since.
They can stack the deck with prospects by confining the rental or selection of these to make an intuitive match with better customers. If Best Customers as a group tend to be women between 25 and 35, selecting these names from the prospect pool is going to yield much better results than a randomized approach.
The data used to stack the deck and reduce uncertainty in the traditional (print) world is:
Small. Each attribute takes only a tiny amount of storage.
Powerful. Just a handful of attributes determines the great majority of our buying patterns. Think about Age, Income, Gender, Household Makeup, Home Ownership and Education. It’s a tiny amount of data and all readily available for most consumers.
Stable. If I know your Date of Birth, I’ll always know your Age. Income varies over a lifetime, but usually only temporarily within the sweet spot of low-mid to low-upper income. Household Makeup and Home Ownership change over time, but not frequently for any household. You get the idea.
Descriptive. This is the big key. The traditional approach identifies unique individuals and attaches a defined set of attributes to them. These may or may not be used to form a profile or to group individuals by socio-economic strata or “personas.