There is probably no journalistic tradition more annoying than the inevitable flood of “best-of” pieces highlighting – in every medium – the finest work from the year past. Of course, nothing pays tribute to your best work quite like lazily stitching it together into a hodgepodge quilt and serving it cold for the holidays.
Like many great traditions, however, it persists for no worse reason than it’s easy. So I present to you a special edition…drum roll please…of the best of 2010 from my blog…
I started the year deep inside a set of posts that turned out to be one of my favorites: a long series that delved into Tactics in Visitor Segmentation. I’d actually begun the series in 2009 but many of the posts came in the initial months of 2010. The first post in January presented a case-study in geo-segmentation for a real-estate site. There are a surprising number of cases where identifying the user’s geography and then tailoring site content is productive. But this post also illustrates a couple of really important segmentation lessons: building segments based on a first visit criteria and studying subsequent visits, and the value in studying sample use-cases when it’s impossible to study a long-tail of behavior.
Interestingly, I flipped this around in a much later post that was of my favorites from the whole year. This post showed how a problem – an A/B Test inside search results - that was nearly unsolvable in traditional Web analytics tools (precisely because it involved thousands of use-cases) was easily studied using SQL.
It’s clear, looking back on my output, that 2010 posts concentrated heavily on data warehousing and visitor segmentation. That visitor segmentation series also included a pretty detailed look at time-based segmentation and a Discover-based look at creating Control Groups to evaluate content effectiveness. Using Control groups is a core database marketing analytics concept that should be, but isn’t always, widely used in online measurement.
It’s easy to see, in retrospect, the emerging theme from my end-of-year posts - the convergence of Database Marketing and Web analytics - woven all through this series. Every one of these topics is, in essence, a database marketing technique re-applied to Web analytics.
Those initial posts were pretty Omniture focused, with quite a few detailed examples of Data Warehouse and Discover segmentation. I tried to balance the scales a bit with a look at some NetInsight segmentation techniques. My favorite of those posts is probably this one: about creating segmentation profiles. Segmentation tends to work in two directions. Sometimes, you start out with a set of rules based on some pre-defined notions of the visitors you want. But sometimes you are looking at a population and trying to figure out what makes them interesting or distinct. For those cases, having a generalized approach to profiling a segment is really handy and the techniques are largely independent of tool.
The data warehousing and Web analytics debate raised its head explicitly in several posts. I like this one best, though it’s really more of polemic and probably not as carefully constructed an argument as, for example, this post.
That first post is, I think, a pretty devastating critique of the idea that there’s something about Web analytics data that makes it unsuitable for data warehousing. The thing is, I’m not sure anyone actually believes that Web analytics data is unsuitable for warehousing – so I worry that this was a case of a thorough dismembering of a, largely, straw man.
Speaking of critiques, I wrote a couple of posts that attacked approaches to testing that focused on site-wide (all visitors) multivariate tests. This may not seem like part of a broader database marketing theme, but it really is. Testing is a core part of a broader database marketing methodology and the more I see of how companies are approaching this – based largely on advice from the testing tool companies – the less impressed I am. I suggest an approach that is much closer to traditional database marketing methods and that makes, I believe, far more sense.
All of these posts, in a way, came together in my recent webinars with Webtrends and Quantivo and this post on Web analytics and Database Marketing. I really think this is a fundamental direction in the industry and a compelling way to think about your Web analytics. You have this set of powerful techniques and disciplines all concentrated on customer channels that are becoming ever less important (e.g. mail, phone). You have these new online channels that are generating huge amounts of customer behavioral data.
The big problem is that online data hasn’t been easily tackled with database marketing techniques. At Semphonic we’ve become increasingly focused on solving exactly that issue. It’s a huge deal - a fundamental transformation in the utility and direction of both disciplines – and something that ought to be top of mind for serious online marketing execs.
Was the whole year taken up with database marketing and data warehousing? Not quite. A pretty large number of social media measurement posts reflected the rapid evolution in our thinking and practice over the course of 2010. I think this post – based on a presentation I did this Fall, captures our focus on using social media data as a research vehicle not just a campaign measurement problem.
If those were the big themes, some of the analytic posts on specific topics also stood out as particularly useful:
Customer Support Content Analysis
Econometric Models for Predictive Analytics
Measuring Content Effectiveness
I liked each of these very much and if the topic is germane and you missed them the first time around, give them a read!
So there it is…my bow to a tired but oh so convenient tradition.
“At least you didn’t do a 2010 Year in Review”, you say – that other weary old journalist’s friend. Never! Well, not in print anyway. Say, have I mentioned our really excellent “BWA 2010 in Review” podcast? Coming soon!