In my last post, I detailed my objections to compacting complex consumer traits into simple acronyms (DINK, EMEA, etc.). In short, these acronyms trivialize consumers’ needs to the points where marketers find themselves communicating to stereotypes rather than actual people.
Marketers who do so, however, might counter that they need to simplify things because they simply cannot market to every last consumer need. And they’re right; they can’t.
However, why should we stop there when marketers can effectively communicate with greater nuance than a handful of consumer groupings? Why, you’d need some kind of machine to do that, wouldn’t you? (Disclosure: if you don’t see where this is going, don’t bother reading past the jump.)
Enter data. Strange as it may seem, all those impersonal ones and zeroes can do a better job of defining a marketer’s audience than the cleverest acronyms. Data don’t smooth out the edges or ignore the anomalies. If a 45-year-old Utah mom loves Nordic death metal, so be it. If the Harley-Davidson rider also subscribes to the ballet, it is what it is. Data don’t judge; data simply know.
On its own, no single data point really means much. Maybe it doesn’t matter that one mom in Utah likes apocalyptic music or that one Harley rider appreciates arabesques as much as Shovelhead engines. However, data don’t make these assumptions. Using basic cluster analysis and offer management tools, marketers can learn whether these data points--or any other data points--matter.
As an example, let’s look at what an online retailer could do with customer information. The retailer could conduct an analysis that includes customer characteristics (location, demographics, browsing history) versus purchases. As a result, the retailer would know what kinds of people buy which kinds of products. Of course, this cluster analysis would probably reveal some “duh” findings, such as that Minnesotans buy more sweaters than Floridians.
However, after weeding through the basics, the retailer might also learn about less evident relationships between customers and purchases. She might learn that people who live in large cities prefer green shirts and people who live in the suburbs like orange. She might figure out why, but it wouldn’t matter much.
Rather than trying to understand why, the retailer can instead see what it means to her business. She could create four cells in her email list, two with city dwellers and two with suburbanites. She could then send an email with orange shirts to one city cell and one suburban cell and one with green shirts to the other cells. The results would give her a good idea if the color preference observed in analysis translated into sales.
This example simplifies how data work, but only by degree. Really, it doesn’t take all the much work. In fact, this cluster-based approach takes a lot less work than having researchers survey hundreds of people and creating cute acronyms.
Moreover, many if not most organizations already have the data they need. Prices for consumer databases have dropped substantially over the past 15 years. I’d bet that more marketers have the problem of too much consumer data rather than too little. During my time at Acxiom, they boasted of having hundreds of data elements for nearly every household in the United States. Even marketers with zero data can start to build their own by recording interactions with customers in store, online and on the phone with relatively little difficulty. So marketers certainly have data at their disposal.
On the other hand, this data-focused approach works best in Internet marketing, where the costs for customizing individual communications approaches zero. Does this limitation mean marketers who work primarily in broadcast and print should ignore the approach? Two things to consider:
As long-winded as this post is, it nevertheless only scratches the surface of what marketers can do by taking simple steps to understand their data better. Great rewards await those marketers who leave acronyms behind and really try to understand their consumers.
Marketers who do so, however, might counter that they need to simplify things because they simply cannot market to every last consumer need. And they’re right; they can’t.
However, why should we stop there when marketers can effectively communicate with greater nuance than a handful of consumer groupings? Why, you’d need some kind of machine to do that, wouldn’t you? (Disclosure: if you don’t see where this is going, don’t bother reading past the jump.)
Enter data. Strange as it may seem, all those impersonal ones and zeroes can do a better job of defining a marketer’s audience than the cleverest acronyms. Data don’t smooth out the edges or ignore the anomalies. If a 45-year-old Utah mom loves Nordic death metal, so be it. If the Harley-Davidson rider also subscribes to the ballet, it is what it is. Data don’t judge; data simply know.
On its own, no single data point really means much. Maybe it doesn’t matter that one mom in Utah likes apocalyptic music or that one Harley rider appreciates arabesques as much as Shovelhead engines. However, data don’t make these assumptions. Using basic cluster analysis and offer management tools, marketers can learn whether these data points--or any other data points--matter.
As an example, let’s look at what an online retailer could do with customer information. The retailer could conduct an analysis that includes customer characteristics (location, demographics, browsing history) versus purchases. As a result, the retailer would know what kinds of people buy which kinds of products. Of course, this cluster analysis would probably reveal some “duh” findings, such as that Minnesotans buy more sweaters than Floridians.
However, after weeding through the basics, the retailer might also learn about less evident relationships between customers and purchases. She might learn that people who live in large cities prefer green shirts and people who live in the suburbs like orange. She might figure out why, but it wouldn’t matter much.
Rather than trying to understand why, the retailer can instead see what it means to her business. She could create four cells in her email list, two with city dwellers and two with suburbanites. She could then send an email with orange shirts to one city cell and one suburban cell and one with green shirts to the other cells. The results would give her a good idea if the color preference observed in analysis translated into sales.
This example simplifies how data work, but only by degree. Really, it doesn’t take all the much work. In fact, this cluster-based approach takes a lot less work than having researchers survey hundreds of people and creating cute acronyms.
Moreover, many if not most organizations already have the data they need. Prices for consumer databases have dropped substantially over the past 15 years. I’d bet that more marketers have the problem of too much consumer data rather than too little. During my time at Acxiom, they boasted of having hundreds of data elements for nearly every household in the United States. Even marketers with zero data can start to build their own by recording interactions with customers in store, online and on the phone with relatively little difficulty. So marketers certainly have data at their disposal.
On the other hand, this data-focused approach works best in Internet marketing, where the costs for customizing individual communications approaches zero. Does this limitation mean marketers who work primarily in broadcast and print should ignore the approach? Two things to consider:
- In the short term, this kind of cluster analysis may give mass marketers insights that they might not get from traditional research techniques. Moreover, comparing available data to purchases grounds insights in actual, not reported behavior. If nothing else, understanding purchase patterns gives interesting depth and direction to traditional research.
- In the very near future, broadcast and print will look a lot more like websites and email insofar as marketers will have the ability to target at the individual level. Every digital cable or satellite dish box has an IP (Internet protocol) address, just as your PC does. Since the cable or satellite company knows your address, it can hook into publicly-available household-level data. So, for instance, Toyota will be able send a Scion ad to a household headed by a 25-year-old, a minivan ad to a home with three kids and a Lexus ad to an affluent middle-aged couple.
Perhaps print will outrun TV for a change. Already this year, Amazon has received over a quarter of a million orders for its Kindle Fire media tablet. As more consumers read the news on tablets, publishers and marketers will have a greater opportunity to target them based on behavior patterns. Think of how Amazon might tie together browsing patterns and purchase patterns to sharpen their offers.
As long-winded as this post is, it nevertheless only scratches the surface of what marketers can do by taking simple steps to understand their data better. Great rewards await those marketers who leave acronyms behind and really try to understand their consumers.
No comments:
Post a Comment