We are still living through the aftermath of the Tower of Babel, though the main language of choice in the marketing, data and analytics industry remains English. Outside of the U.S., I speak at conferences and events in Korea, Brazil and the U.K. Even when I presented in Korean — with a PowerPoint presentation consisting entirely of English — I called data “Data,” though the pronunciation is more like “dei-tah” there. Korean business people love to say “Big Data” in English, though the meaning is quite different from what I am accustomed to. They use it with a much broader meaning than we do in America; they literally imply anything and everything related to data activities, small or big, raw or analyzed. Conversely, I have encountered groups of people in America who have a very narrow definition of it, whether it be about literal size, complexity or even specific platforms, such as Hadoop. I am sure each of you has a different notion of the word.
Recently, I participated in a retail conference in London regarding “Personalization.” I was a panelist, and I noticed they spelled the word “personalisation.” I didn’t want to argue about how funny that spelling looked among folks from a country where the English language was literally spawned, but what is the point of having the letter “z” in the alphabet if they are not using it for a clear “z” sound? In any case, they too seemed to be searching for the meaning of the word in marketing, as the very first question to the panel was “What does personalisation mean to you?” Not surprisingly, each panelist provided different answers.
Since then, I have been attending marketing and technology conferences quite diligently this season. While a great many panel discussions, industry tracks and keynote speeches were about personalization, I found that literally everyone meant different things by saying it. Unfortunately, some presenters were as confused as their audiences, and some were downright clueless (more on the subject of useless conference tracks in future articles). Yes, all of that popularity means “Personalization” is the next big thing after “Big Data,” and it truly reached the buzzword status. And that is really too bad for the users of data, technology and analytics.
Why? Because many users end up thinking that they are doing a good job at it, while in reality, they are only touching the surface. Such an attitude leads to investment in the wrong places, while other vital steps could be missed completely. It is not much different from patients in a placebo group thinking that they are taking the real trial drug. It is even worse than that in marketing, as users may have paid a good sum of money to check off that little box called “personalization.” The first blame should be on the service providers who overpromised the effectiveness of the toolset (as in “All your problems will be solved if you buy this!”), but the users must be more educated about it, too.
So, what does personalization mean to you? Allow me to list a few possible answers:
- Addressing your customers by their first names?
- Suggesting more of the same products that they just purchased through collaborative filtering?
- Collecting explicitly expressed preferences and reacting to them?
- Keeping in touch with your customers all of the time?
- Customizing emails and landing pages based on customer preference?
- Knowing when to contact them and through what channel?
I think we can safely agree that calling someone “Dear Jane” in an email isn’t the end of personalization. Suggesting more of the same products? Such practices, joined with “keeping in touch with customers all of the time,” often leads to “personally annoying your customers,” not necessarily personalization (refer to “Personalization Is About the Person”).
I happened to have caught a rather technical presentation (with a title that includes “personalization”) by a reputable provider of a personalization engine, and I was quite impressed with all of the complex and ingenious algorithms they applied to the effort. I am not a mathematician, and I do not mean to criticize those brilliant scientists about their efforts. But I must say that three out of four their steps were about products, not people, though they left a step for behavior-based segments. Presented segmentation methods and variable sets were not by any means at the level of as-good-as-it-gets, but adding behavioral segmentation is a very hopeful move, indeed.
Regardless of the complexity, stringing up related products together, using collaborative filtering, popularity hierarchy and/or clever methods to harness unstructured meta-data are still more about the product, not the consumers. People have an uncanny ability to smell machines, even through remote channels. Personalization definitely requires some human touches (or at least illusions of it), and that come from understanding the target’s current and past behaviors (refer to “Data Atrophy”).
So, what do marketers to do, if even the most advanced kind of personalization engines are still more about products, not people? We need to fill in the gaps with data and analytics. To get there, let’s first break down what personalization is made of:
I am a firm believer that every personalization (or any type of 1:1 messaging) must start with data. But for the purpose of being pragmatic, I reversed the order here.
Simply, if a marketer doesn’t have enough content that matches different types of customer demand and their personas, the effort will be pointless, even with an ample amount of data. Contents — literal and graphic — must be created with potential targets in mind, and they should be properly managed through DAM (Digital Asset Management) systems. We are talking about something far more organized than some memory sticks sitting in a desk drawer in a creative agency. For many marketers, this is “the” personalization effort, as content creation is an age-old marketing function, and effectively managing it is at the heart of digital marketing.
Then, the marketer needs to acquire the ability to show different contents to various types of customers. This is where all of those commercial solutions come into play. If it is about the website, is it modularized, so that various parts of the pages can be customized? If it is about email campaigns, can each email be tailored with different offers and feature products? If it is about offline campaigns, how flexible can versioning be? There are already supermarket chains that customize almost every coupon book with different binding sequences and contents. The ability to deliver customized messages to customers and prospects is a must-have, not an option, for any type of personalization initiative.
Next, are all of these efforts data-driven? What types of data are being used? Just product meta-data and product-level sales data? Or individual behavioral and demographic data? If so, are they just based on snapshot data of the present, or the person’s historical data, as well? Are product-, event- and transaction-level data summarized to an individual level for proper personalization?
That leads to analytics (and this “analytics” has many meanings, too). Are data converted to forms of segments or personas, or are the raw data still being plugged into the engine? The difference in effectiveness is huge, as even machines prefer clean and simple data. Further, even with ample amounts of transaction- or event-level data, we often find lots of huge holes in data when aligned around the person, as there is no way to know everything about everyone all the time. Such gaps should be filled with statistical models, while we often label those with different names, such as segments and personas. (This leaves yet more room for serious misunderstandings.)
Illustrated is a three-step approach to personalization, starting with installation of a commercial personalization engine. Then test-run the engine with simple segments, based on available data. After all, reacting to immediate customer needs and displaying different versions of content based on known explicit data is not simple or easy. That would still be more like “personalize contents only sporadically for some people through some channels.” To reach the stage of “personalizing content constantly for everyone through all channels,” event- and product-level data must be realigned around target individuals, and personas must be built to fill in the gaps (refer to “No One Is One-Dimensional”).
Personalization is definitely the most popular buzzword these days, though it means different things to a lot of people. What does not change is that this movement is here to stay in the age of information overload, as marketers must stand out, for their survival, with relevant messages to ever-distracted consumers.
What we simply refer to as “personalization” is made of multiple components, and that is why many of us are confused by it. Therefore, we must aspire to reach a true personal level with our customers through a step-wise approach, not a single giant leap. Let us not make the mistake of calling the mere first few steps the whole thing, when more important data and analytics steps are not even in play yet.