By now, you’ve no doubt heard that customer experience (CX) demands a high degree of personalization, and that achieving “deep personalization” in your CX requires data

What you haven’t heard — until now — is much detail on what data to use for personalization and how to connect that data to your digital experience delivery tools like your web content management (WCM) system. In this article, I’ll explore integrating data for personalization and share with you some limitations my company discovered with some of the leading WCMs when it comes to using data for personalization.

Defining ‘Data’

There are many types and sources of data when it comes to customer experience. Let’s briefly touch on the main categories of data; then, we can focus on using these data for CX personalization.

Mike Hay from digital marketing agency ROI DNA gives us a useful starting point. Hay categorizes data into four different types:

  1. Contextual: In this category, Hay includes geolocation information like physical location and time of day which can be derived from a visitor’s IP address; information about the visitor’s device, browser and other info in the user agent string passed by a visitor’s web browser; and whether the visitor is new or returning, determined by comparing the visitor’s IP address against your web server’s visitor logs.
  2. Behavioral: This is exactly what it sounds like: data about what a visitor does. While Hay doesn’t make this distinction, it’s useful to think of behavioral data as what a visitor does before, during and after a visit to your site. Before includes the referring web site (the site a visitor was on before visiting yours). During includes pages visited, time spent on each page and total time spent on your site. After includes revisiting your site and visiting other sites (you’ll know this if you’re using remarketing). Hay also includes off-site behavior, including interactions with marketing campaigns (think: email opens and clicks) and purchase history. To this I would also add other interactions a person is having or has had with your company, like IVR usage and call center contacts.
  3. Demographic: This category includes individual attributes like gender, age, profession, education and marital status. To Hay’s list, I would add name, email address, phone numbers and other personally identifiable information (PII).
  4. Social: Hay argues that social deserves its own category. I would argue that social is just another source of behavioral and demographic information.

Notice there’s a split between data you can collect on known versus unknown visitors. While data on unknown visitors can be collected by cookies and web analytics tools, data collected on known visitors must be reported — or at least permitted — by the individual visitor and stored in a customer relationship management (CRM) tool or other system(s) of record. And other data, like personal health information (PHI), blur the lines between behavioral and demographic data.

Using Data

According to eMarketer, recent research from VB Insight and a similar study by Econsultancy and IBM showed that most companies are using only a small fraction of available data for personalization. While some companies struggle with data collection, that’s not the problem for most. The overwhelming majority (96 percent according to the VB Insight study) cannot unify the data they have, and most companies — a full 60% according to research by Signal cited in the eMarketer article — can’t "personalize customer experiences the way they wanted to due to fragmented data/profiles." 

I believe there are two reasons for this:

Lack of Master Data Management

Chances are, the data you have are spread across multiple systems and databases. Statistically speaking, your company is likely part of the 96 percent of companies mentioned above that have not unified these disparate data sources into a single data layer. That implies to me that master data management (MDM) — linking all data sources into a single — is seen by most companies as being either too difficult and/or too expensive to achieve. Neither has to be true for you and your company.

MDM requires that you know where your data are stored, and that you can construct a single (master) list of variables that corresponds to the data. For example, “Customer Name” might be the variable name. This data might exist as “fname” and “lname” in your mainframe system, “contact name” in your CRM, “first name” and “last name” in your email marketing system, and so on. Ideally, you have at least one variable — a unique identifier or UID — that can link the data you have on each person across systems. Hint: Don’t use email address or telephone number, as these can change over time.

Once you have created your master list of variables you want to use for personalization, you need a tool that can connect to the systems storing your data. You’ll use this tool to “map” the different iterations of the same variable (see the “Customer Name” example above) and concatenate (i.e., combine) the different data sources into one record for each person. That’s not as difficult as it sounds, but it does take some time and tech savvy.

There are lots of MDM tools on the market. Someone, somewhere in your organization is probably already using one or more. Ask around. If not, there are several resources from Gartner, Forrester and other firms to help you evaluate and choose one. And MDM doesn’t have to be expensive. In addition to commercial options from IBM, Informatica, SAP and others, there are open source MDM tools available, like Talend and Teiid.

Limitations in DXD Tools

Over the past few years, the leading digital experience delivery (DXD) vendors have improved their ability to use data for personalization. That said, when evaluating DXD solutions, my company found that most have only scratched the surface of what’s possible when it comes to using data for deep personalization.

For example, most WCMs offer out-of-the-box integration with web analytics software, like Adobe Analytics and Google Analytics, and CRM systems, like Salesforce.com and SugarCRM. However, most WCMs are only using a very limited amount of available data for personalizing experiences because they are focused on personalizing for unknown users.

We found that most WCMs lack the ability to use data for the deeper personalization required for known visitors, including (and especially) existing customers. No WCM that we evaluated offered direct support for MDM tools that would provide the rich data needed for deep personalization. And most WCMs are architected in such a way that makes personalizing content using business logic and data from MDM tools very difficult, if not impossible.

Your Homework

First, if you’re not using master data management tools today, investigate them. In addition to personalization, they can be useful for predictive analytics and optimizing customer experience in real time. Use the links above as a starting point in your investigation.

Second, take a hard look at your digital experience delivery toolset, and your web content management system(s) in particular. Investigate whether or how well they support MDM tools. Most MDM applications can provide data via RESTful web services, and many WCMs can accommodate replacing variables using RESTful web services — that’s a good place to start.

Creative Commons Creative Commons Attribution-Share Alike 2.0 Generic License Title image by  jeff_golden