The first gotcha is the analytics. Personalized email messages and offers can work like magic to lift returns, but only if those personalized messages truly interest the customer. Otherwise, to the customer, it’s just more noise.
So the analytics to determine what to say or what to offer each customer in an email has got to be accurately “predictive.” That means something much more than demographic or behavioral segmentation. It’s sophisticated data science, beyond the scope of most marketing technology platforms not built specifically for this purpose. The “ability” to personalize messages is not the same as delivering messages that motivate. Messaging is not always marketing.
The next gotcha is “personalized selling.” Most platforms that personalize email communications try to build up a profile of behaviors and attributes that may indicate preferences and interests -- so when those behaviors are detected, appropriate messages can be delivered in real time. This is where the action is in marketing technology today, with a wide range of complexity and effectiveness available.
There’s a lot of data to gather, integrate, and interpret, and a lot of rules and actions to configure, but it can enhance the experience of a website visitor or encourage a customer to buy, if and when the right behaviors are detected. This is personalized selling, messaging to one customer at a time in a given context. However, it’s not marketing. Marketing is messaging to a market.
So the last gotcha is marketing at campaign scale. If you are going to do marketing -- that is, message to a market of thousands or millions of customers at once to stimulate demand -- and you want to personalize it, then you think in terms of campaigns. If you have a lot of potential buyers to reach and hundreds or many thousands of products to offer them, how do you identify exactly the right product (or set of products) to promote to each individual customer in your campaign, the exact products they are most likely to buy? It’s an intractable problem for almost every marketing platform.
This is why most marketers campaign around specific products, and have someone do all the analytic modeling needed to find the best customers for those specific few products. Finding the best customers for every product -- or any product on a moment’s notice, or the best product(s) for every individual customer -- is cost-prohibitive, never mind incompatible with campaign deadlines.
However, there are some new approaches emerging that enable marketers to cost-effectively execute personalized marketing campaigns rapidly and at scale.
The secret has to do with one final gotcha inherent in most marketing technologies that personalize messages: they typically lose sight of the ultimate objective of marketing, which is getting people to buy your products and services.
Instead of directly answering the most important question -- what is each customer most likely to buy, and when? -- the analytics are aligned toward predicting answers to all sorts of other important and yet less-vital questions. For example: What communication channels do customers prefer? Who is a good brand advocate, and whose sentiment is turning unfavorable? And so on.
The secret is staying focused on predicting the main objective (getting people to buy your products or services), reducing the data complexity by employing more advanced analytics that can get the most out of only the most necessary data, and automating the entire analytic process for scale. This approach is making personalized email marketing truly practical for the first time.