While some social media specialists pine away about online ranking systems, more experienced marketers don't care about online algorithms so much. Do you want to know why?
Everything they need to know about their customers is already online, offline, and proprietary. In fact, in places like Las Vegas, the ability to track customer movements offline has existed for better than a decade and even longer if you count some of the amazing things resorts did with coupon codes and player cards. (In some cases, Vegas invented data mining.) It's not just about one city. It's everywhere.
Real measurement doesn't track kitty vid clicks; it tracks how much cat food you buy and how many cats you own.
Anybody who read the recent Forbes article already knows half of it. Target's data analysis has become good enough that, based upon purchasing decisions alone, it has a remarkably high probability of predicting if someone in the household is pregnant (even if not everyone in the house knows it).
In some cases, I wouldn't be surprised if the data analysis is good enough to predict the probability of pregnancy before the future parent knows it. It wouldn't be hard to do. By culling a large pool of newly pregnant customers' shopping patterns and then analyzing their purchases just prior to becoming pregnant, any marketer can test for a statistical probability.
Match this probability against every customer based on site views, product purchases, etc. and there will be a pattern, assuming there is a pattern to be found. The same method could even apply to any number of life changes: employment, unemployment, engagement, political leanings, successes, failures, etc. When you study the psychology and sociology of purchases, patterns begin to emerge.
It's all relatively simple too. Any company with a guest ID program (whether online or offline via club cards), has a sophisticated data analysis program or the potential to make one. As soon as any customer makes a purchase, select information is dumped into a data bucket, usually purchases, names, credit cards, email addresses, phone numbers, and any other information people willfully give up.
Such data mining isn't foolproof, but it has gotten better. This is why online advertisements follow you around on the Web after you visit a site, why some marketers know when to send you a discount coupon, and how some sellers make reasonably good (but not great) predictions of whether you might like certain books, music, fashions, etc. And, even better, you don't have to be online to make it work.
Influencers are interesting, but researchers are powerful.
Naturally, the success of any company's data analysis isn't determined by the program alone. It's determined by the researchers — people who can identify trends and turn those trends into action.
Taking a second look at the Target story, statistician Andrew Pole was one of those people. His team created a statistical benchmark based upon women in the company's baby registry. They were buying larger quantities of unscented lotion; buying more supplements like calcium, magnesium, and zinc; looking at extra big bags of cotton balls and wash cloths; etc. All in all, Pole and his team identified 25 products as indicators.
When this information was matched to women not on the baby registry, Target had a reasonably good idea which customers might be pregnant whether or not they signed up with the baby registry. This is powerful information for any company that wants to create direct-to-customer communication, right down to specific messages based on which trimester.
Contrast the power of data mining against influencers who are actively attempting to appear influential by covering popular trends, soliciting traffic, using clever headlines, buying advertisements, gaming attention, etc. and the shortcomings become a little more apparent. Do you want to reach a percentage of an influencer's followers or do you want to reach people that you know are pregnant?
That's not to say influencers can't be useful. They can be useful for short-term prospecting, message reinforcement, amplification, and conversion (adding them to a database where the heavy lifting occurs). However, since numbers alone aren't nearly enough, it might make more sense to find out who your customers already listen to as opposed to any online algorithm.
The down side of data mining is always short-term creepiness.
Any time an article like the Forbes story breaks, it always feels a little bit creepy. But as creepy as data mining can be (and it is creepy), it's also a constant. Target might have been the company covered, but hundreds of companies have been tracking equally detailed data for some time.
They did it before social media and social networks too; virtually anyone with a credit or rewards card. And if you want a friendly reminder of just how much data is being captured, take an unexpected shopping spree and make some oddball purchases on your credit card.
If you trigger a fraud investigation, your credit card company will be able to tell you everything you did during the day, including travel routes. And if they wanted to, they could probably cull the data and tell you some pretty interesting things about you, your family, and where your next vacation might be.
Related Reading:
• Social Media Key Influencer In Multi-Exposure Purchase Path by eMaketer
• FBI Seeks Social Media Data Mining Tool by CBC News
• How A Smartphone App Can Detect How Fit (Or Fat) You Are by Forbes
Everything they need to know about their customers is already online, offline, and proprietary. In fact, in places like Las Vegas, the ability to track customer movements offline has existed for better than a decade and even longer if you count some of the amazing things resorts did with coupon codes and player cards. (In some cases, Vegas invented data mining.) It's not just about one city. It's everywhere.
Real measurement doesn't track kitty vid clicks; it tracks how much cat food you buy and how many cats you own.
Anybody who read the recent Forbes article already knows half of it. Target's data analysis has become good enough that, based upon purchasing decisions alone, it has a remarkably high probability of predicting if someone in the household is pregnant (even if not everyone in the house knows it).
In some cases, I wouldn't be surprised if the data analysis is good enough to predict the probability of pregnancy before the future parent knows it. It wouldn't be hard to do. By culling a large pool of newly pregnant customers' shopping patterns and then analyzing their purchases just prior to becoming pregnant, any marketer can test for a statistical probability.
Match this probability against every customer based on site views, product purchases, etc. and there will be a pattern, assuming there is a pattern to be found. The same method could even apply to any number of life changes: employment, unemployment, engagement, political leanings, successes, failures, etc. When you study the psychology and sociology of purchases, patterns begin to emerge.
It's all relatively simple too. Any company with a guest ID program (whether online or offline via club cards), has a sophisticated data analysis program or the potential to make one. As soon as any customer makes a purchase, select information is dumped into a data bucket, usually purchases, names, credit cards, email addresses, phone numbers, and any other information people willfully give up.
Such data mining isn't foolproof, but it has gotten better. This is why online advertisements follow you around on the Web after you visit a site, why some marketers know when to send you a discount coupon, and how some sellers make reasonably good (but not great) predictions of whether you might like certain books, music, fashions, etc. And, even better, you don't have to be online to make it work.
Influencers are interesting, but researchers are powerful.
Naturally, the success of any company's data analysis isn't determined by the program alone. It's determined by the researchers — people who can identify trends and turn those trends into action.
Taking a second look at the Target story, statistician Andrew Pole was one of those people. His team created a statistical benchmark based upon women in the company's baby registry. They were buying larger quantities of unscented lotion; buying more supplements like calcium, magnesium, and zinc; looking at extra big bags of cotton balls and wash cloths; etc. All in all, Pole and his team identified 25 products as indicators.
When this information was matched to women not on the baby registry, Target had a reasonably good idea which customers might be pregnant whether or not they signed up with the baby registry. This is powerful information for any company that wants to create direct-to-customer communication, right down to specific messages based on which trimester.
Contrast the power of data mining against influencers who are actively attempting to appear influential by covering popular trends, soliciting traffic, using clever headlines, buying advertisements, gaming attention, etc. and the shortcomings become a little more apparent. Do you want to reach a percentage of an influencer's followers or do you want to reach people that you know are pregnant?
That's not to say influencers can't be useful. They can be useful for short-term prospecting, message reinforcement, amplification, and conversion (adding them to a database where the heavy lifting occurs). However, since numbers alone aren't nearly enough, it might make more sense to find out who your customers already listen to as opposed to any online algorithm.
The down side of data mining is always short-term creepiness.
Any time an article like the Forbes story breaks, it always feels a little bit creepy. But as creepy as data mining can be (and it is creepy), it's also a constant. Target might have been the company covered, but hundreds of companies have been tracking equally detailed data for some time.
They did it before social media and social networks too; virtually anyone with a credit or rewards card. And if you want a friendly reminder of just how much data is being captured, take an unexpected shopping spree and make some oddball purchases on your credit card.
If you trigger a fraud investigation, your credit card company will be able to tell you everything you did during the day, including travel routes. And if they wanted to, they could probably cull the data and tell you some pretty interesting things about you, your family, and where your next vacation might be.
Related Reading:
• Social Media Key Influencer In Multi-Exposure Purchase Path by eMaketer
• FBI Seeks Social Media Data Mining Tool by CBC News
• How A Smartphone App Can Detect How Fit (Or Fat) You Are by Forbes