Guest author Leo Sadovy is the director of global marketing at SAS.
During the Cold War, the KGB was so good at identifying undercover CIA agents that officials worried that there was a highly placed mole in the agency. But as Jonathan Haslam, a professor at Princeton University, wrote earlier this fall, that wasn’t the case at all.
It turns out that the KGB was quite effective in mining data. The KGB gathered publicly available information on deployed U.S. Foreign Service personnel, along with observations and data from allied countries, analyzed it, and discovered how the agents’ housing and pay patterns were markedly different than those of the State Department officers the agents were posing as.
Data, Analysis, Insight
The KGB’s Yuri Totrov was able to find 26 independent indicators which invariably distinguished CIA agents from the genuine and otherwise harmless State Department field service officers, or FSOs.
- The CIA pay scale was significantly higher than for FSOs.
- FSOs could and typically did return home after a 3–4-year tour. Agents did not.
- When agents did return home, they did not show up in State Department listings.
- FSOs were always recruited before the age of 31. Agents could be older.
- Only real FSOs attended the three-month training session at the Institute for Foreign Service.
- Field agents might be reposted within a country. FSOs never were.
This wasn’t rocket science, and it didn’t require a high level mole as the more paranoid CIA chiefs suspected. No, the patterns and insights just popped right out of the data when the right analysis and investigative techniques were applied.
These same sort of insights are available to businesses, although they may be hidden somewhere in a pile of enterprise data. There are a number of descriptive analytic approaches, data mining and classification techniques, available to handily tackle this problem. The most well-known include clustering, market basket analysis, and decision trees, much of which can even be accomplished visually and without the need for specialized skill sets.
Let’s use customer marketing as an example. Concealed within customer demographics, purchase history, service calls, and product data lie the equivalent of Totrov’s 26 attributes and indicators.
- Which customers leave after the introductory period and which will renew?
- Which customers will upgrade to the next model automatically, and which will switch brands at this point?
- Which customers will buy directly online versus those that buy in the store after doing online research?
- Which customers are typically motivated by an online discount versus those interested only in particular features?
The Buyer Who Loved Me
Totrov used his knowledge of a few known CIA agents to extrapolate into the unknown. The same approach is available to businesses that have known customers with known behaviors and attributes, which can be applied to an unknown market.
Knowing that a certain age group, gender, ZIP code, length of service, method of purchase, method of payment, etc. is the key attribute signaling customer churn or an upsell opportunity allows sales and marketing to target various promotional investments more productively.
Here are a few examples of how organizations can take full advantage of customer data:
- A retailer can execute different marketing strategies based on a segmentation of seasonal versus year-round customers, maximizing the value of the latter while targeting, say, a Christmas-gift buyer at exactly the right time.
- A sports franchise can detect attendance and secondary market trends to identify which season-ticket holders are at risk (and lure them back), and which regular fans are the most likely candidates to become new season-ticket holders.
- An online retailer can use website traffic analysis to distinguish uninterested browsers from those whose online behavior suggests that they will become buyers with just a little timely nudge or incentive.
- A telecom provider can select the best second product to offer, understanding that customers who buy multiple products or services have a lower churn rate.
- A bank can use their customer data to recognize distinctly different paths to cross-sell and upsell based on the customer’s initial interaction with the bank. The next best offer to a first-time loan client is different from that of a savings account holder.
Totrov’s undertaking was the big-data project of its time. It may not have consisted of terabytes of data, but it was still a comparatively difficult task using the manual tools and techniques of the day. Today’s businesses, by contrast, have powerful tools that can mine far more data in mere hours.Customer and marketing insights won’t have quite the same intrigue as unearthing Cold War spy networks, and it likely will never be made into an action thriller starring Matt Damon or Daniel Craig, but it might help close a sale.
Still image via The Spy Who Loved Me