Last month Webtrends released my whitepaper on Database Marketing and Web analytics – a whitepaper focusing on segmentation techniques and the use of both scoring and filtering for target marketing. I did a webinar with Webtrends on the same general topic last week and today I thought I’d talk a little about the webinar because it focused on a slightly different set of problems than the whitepaper.
The webinar was titled “Is Your Marketing as Relevant as it Could Be? – the Convergence of Web analytics and Database Marketing.” Which I know sounds a bit like a hyped up doctoral thesis. It’s almost long enough but not quite obscure enough to be a good PhD thesis. I’ve already talked about the broad thesis of the webinar in a previous post, but as I developed the content for the presentation I think it went in an interesting direction.
For marketers seeking relevance, it seems to me that the trinity of database marketing: “Offer, Targeting and Creative” is a practical approach to what relevance actually means. Database marketing has a long and proven history of success. It’s an approach that’s worked in a number of different channels and in nearly every industry vertical there is.
As successful as database marketing as a discipline has been, however, it’s got some issues. Principal among these issues is that it’s been most successful in channels that are in various stages of decline. Direct mail and outbound calling, along with the list businesses that drove them, have become far less relevant and important. The places where consumer live today, the web, mobile, interactive games, app stores, etc. are spinning off vast amounts of information about consumer likes, interests and preferences. They aren’t however, the channels that database marketing has really encompassed.
So you have a pretty fundamental divide between the techniques that have proven successful in direct marketing and the channels where consumers actually live these days. That can’t be a good thing.
The primary thesis of the webinar is that these new channels (and I deal mostly with Web as you’d probably expect) present real challenges to the discipline of Database Marketing but that there are some ways to surmount those challenges.
Here are the two slides that I think were at the heart of the webinar (is there some general rule that almost all webinars really only have two slides that matter? Perhaps I’ll try a 2 slide webinar someday…)
The three issues in the slide above: the anonymous nature of much of the data, the difficulty in attaching meaning to page-based data, and limitations and problems in the Web analytics toolset, have plagued database marketers seeking to use online data.
I tried to address each of these in the webinar:
Of the three, I felt like I had the most interesting things to say about the problems associated with finding meaning in page-based data.
The problems of anonymous data really resolve down into two basic solutions: the increasingly important efforts by many Web sites to de-anonymize visitors via opt-in mechanisms like registration, and the potential for using site personalization as the feedback mechanism for database marketing to fully anonymous visitors.
Tools remain a significant issue. But products like the Webtrends Visitor Data Mart and the growing availability of data feeds and advanced analytics warehousing solutions make me think the tool problem will solve itself and can be addressed right now by organizations with serious intent.
The difficulty in getting meaning from page-based data is a more interesting topic. Like most disciplines, database marketing has its own rich set of tricks and tools. Those tricks and tools have focused heavily on certain kinds of data (address enhancement/changes, geo-demographics, demographics, subscription lists, credit, etc.), none of which are generally available for online database marketing. Instead, there’s a stream of page events that, on the surface, appear to mean very little.
In the webinar, I suggest three avenues that Semphonic has found to be fairly successful in creating the type of meaning database marketers are seeking in online data. The first of these methods is the simplest, most obvious, and most important: the creation of a rich taxonomy around pages. It isn’t the fact of viewing pages that’s important for database marketing. Variables like number of pages viewed are almost never meaningful. What is meaningful, is what those pages are about – and that’s contained in the taxonomy.
Rich page meta-data is the single most important Web analytics infrastructure piece you can have – not just for database marketing but for analytics in general. One of the points I make in the webinar is that you shouldn’t think of this as a single taxonomy. That’s almost never enough. At minimum, a good Web site will encode taxonomies based on functional classification, product category, feature interest/concern, buying stage, and content type.
Functional classifications tell you whether the visitor spent time in real content and help identify the visitor’s buying stage. If the content was primarily informational, then the visitor is probably at an early-stage of buying. If the visitor moved from informational to convincer content, the stage may have been advanced. This is vital information for database marketing especially for targeting.
I imagine that the need for and value of product category taxonomies is obvious. Knowing what products, type of products, and mix of products a visitor viewed is, hands-down, the most important thing you can know about a visitor’s web experience. Product taxonomies are essential for every part of the database marketing canon: offer, targeting and creative.
The need for a taxonomy around feature/interest is less obvious but every bit as real. Most sites provide a set of detail pages or tabs around their product and services and it’s incredibly valuable to know whether a prospective buyer viewed the features, the pricing detail, the technical specifications, or all of the above when browsing. This type of information can really help tune both offer and creative.
Buying stage can be covered under a functional classification, but it’s often worth a separate taxonomic category of its own. We usually recommend that every piece of content be coded as to where it fits in the buying cycle – something that’s particularly useful for understanding B2B site behaviors when companies are looking at whitepapers and presentations that may have little or a very great deal to do with service offerings.
Coding these taxonomies makes it dramatically easier for database marketers to translate page data into meaningful categorizations of visitor interest and, from there, into effective targeting segmentations.
In the webinar, I talked about how Clustering is, in many respects, an alternative to one of the most disappointing Web analytics techniques – pathing. We all used to think pathing was a panacea and that if we could trace visitor’s paths, we’d really understand what worked and what didn’t on the Website. It didn’t exactly turn out that way. Paths turn out to be too complex and too varied to distill into easy analysis. Instead, we’ve found that clustering – particularly clustering that uses variables indexed to the behavior in those rich taxonomies I just talked about – is a much more fruitful technique for database marketing. Clustering creates a kind of basket of interests or scores for each visitor – and those just naturally make excellent filters for targeted marketing segmentation.
I’ve blogged on Use-Case analysis fairly recently and it’s become a cornerstone of our (Semphonic’s) approach to several key Web analytics tasks including Management Reporting. It may not seem like a natural fit for database marketing, but it actually is. In our use-case analytics process, we signature every visit to the Website with a visit intent. Based on that visit intent, we assign a specific success metric. That means for every meaningful visit (i.e. every non-bounce visit) to the Website, we can assign a use-case and success code. Those two codes are wonderful aggregations of a whole session and far more meaningful for targeting that the simple fact of a visit or a set of pages. In effect, the use-case analysis saves the database marketer from having to figure out which page events indicate which types of visits.
What all three of these techniques have in common is that they are ways to aggregate page events into meaningful visitor level codes or scores that are easy for a marketer to interpret.
As I said in the webinar, Taxonomy, Clustering, and Use-Case Analysis are the three best techniques I know of to make online data useful for database marketing. You can listen to the whole presentation for some more color (about 35 minutes before questions – 50 minutes in all) or drop me a line and I’ll send you the Powerpoint!