The Five Most Common Misconceptions in Personalized Marketing
According to data published by Caslon & Co. based on research by the DMA and PODi, personalized marketing content generates response rate 3x or more than static content for various marketing objectives, including ordering.
You can scarcely read an article, blog, or presentation from pundits on the future of marketing without hearing it. Companies who have managed to do it already know it. But what if you have thousands, tens of thousands, or even millions of customers? Retailers, e-tailers, consumer products and services companies, and B2B distributors have a hard enough time segmenting their customer base or finding appropriate customers for a given product, never mind trying to figure out “the” specific products that are the best to offer every single one of their customers at any given point in time. If they could do it, the results would be fantastic. But conventional wisdom puts this beyond all but the largest companies.
There is so much attention to this area now that things are changing. Check again. What you assumed true a couple years ago may not be so today. Here is a list of the most common misconceptions I hear:
1. We need as much data as we can get about each customer to make accurate predictions.
Conventional wisdom is you need to pull data together from as many sources as possible to discover the right customer attributes and behaviors that most accurately predict a customer’s loyalty or buying interests. That, by necessity, entails a huge data integration project before you can get started, but the basic premise is not true. Skip the massive big data project. With the right technology, you can get highly accurate purchase predictions from very little data about past purchases. A few fields will do it.
2. The best lift comes from finding the right customers for each product.
By far, the most typical use of predictive analytics for direct marketing is to determine which customers are likely to be interested in a given product and run a campaign to those customers offering them that product. This is partly because the company has reasons to want to move a given product (high inventory, better margin, a manufacturer’s promotion, etc.), but it’s also because of limitations imposed by conventional wisdom. Data modeling is expected of data scientists to figure out a way to group or “score” customers on their likely interest in the product. Many tools, algorithms, and techniques may be involved and it could take days. Probably okay for a handful of products, but forget about finding the best customers for “all” your products, right? And besides, who cares? Well, there are technologies out there that automatically score all products against all customers without anyone doing any modeling. And that information has the potential to change the way you market. Rather than product-centric campaigns, you can do customer-centric campaigns that make every customer a different, personalized product offer. In our experience, the best lift comes from those campaigns, far exceeding the product-centric approach.
3. It’s hard to execute individualized marketing campaigns.
Digital marketing is enormously adaptable, but many marketers still assume it’s too hard to individualize emails to every customer. In fact, all but the most basic email automation systems allow templated emails that contain areas for variable content. That content can be automatically pulled from a control file, a customer list that contains or points to the variable content for each customer, and inserted into the template on the fly. The hard part is building the control file, usually a delimited flat file that can be uploaded to the email system. But that’s getting easier too. There are services and do-it-yourself tools for building these files from the analytics. So, for example, a marketer could specify a list targeting fading customers where they will see four product offers, three they’ve purchased before and one they have not, based on their purchase probability and automatically output a control file in the format needed by their email system.
4. Print campaigns are too expensive and response is down.
In fact, print (postcards, catalogs, etc.) is considerably more effective than most digital media for getting a response. It is much more expensive per piece, however. In this case, the trick is to mail to the right customers, the ones most likely to buy, and not waste money on those who are not. Analytics ranking each customer by loyalty, likelihood to buy, and expected value of purchases can make print campaigns considerably more profitable. Add to that personalizing the content in the print materials, and you have potentially explosive returns. More and more printers offer this variable print capability.
Figure 2 courtesy of www.marketingcharts.com.
5. The most important job of marketing is to find new customers.
That is an important job. Debatably, however, it is more important to get the most value out of every existing customer. Most marketers know it costs far less to win business from existing customers than a new one. There’s no quicker, easier way to materially move the company’s top and bottom line than increasing the lifetime value of each customer (not just on average, but of “each” customer). There is no guarantee your “loyal” customers will continue being loyal. Personalized marketing to your customers makes a difference.
So there you have it. It does not have to be that costly or difficult to predict the products each customer is most likely to buy and use that to personalize marketing, whether in email or print as discussed above, or for online recommendations and ads. The more customers and products, the combinatorial challenge increases exponentially. The job requires advanced analytics. There is a body of “conventional wisdom” about how to do this based on predictive data science that has been around for decades. The fact that the inner workings of this arcane science are accessible only to data scientists who tend to be steeped in this conventional wisdom does not help. New ideas and approaches are challenging the norm, however. And with them, many of the barriers and costs are coming down too.