Web Analytics Integration: Holy Grail Or White Whale?

A few years ago, when I spoke about "integrating Web analytics data with data from other systems," I was often met with quixotic looks from vendors and asked, "Why would you want to do that?" Practitioners, on the other hand, wondered, "how are you going to that" with the current tool sets available at the time. Consultants would say, "we can help you with that, as long as you purchase specific tool X and have your data in database Y."

I quickly learned that it was a real challenge to integrate Web analytics data in heterogeneous data environments. Most of the tools weren't able to load transformed data from other systems and expose it in the interface, nor were many tools able to feed data to other systems. Those that could dealt strictly with aggregate, visit-level data, and could typically provide an export in only very common formats.

Time marches on as relentlessly as companies innovate. Now when I ask the same question to analytics tool vendors, I get a different response -- more along the lines of "yes, we can do that, we understand why you would want to that, and we have a various options for doing so." The Web analytics industry has come a long way for sure. Vendors have actual products or services that enable integration (such as Omniture Genesis, Coremetrics Connect, WebTrends Open Exchange, and Unica's Partner Ecosystem) or they offer technology solutions that support direct data integration using off-the-shelf extract-transfer-load tools into their open relational backend databases. Still, the question remains "what do I integrate," which brings me to the point of this article. Several integration points are now possible to consider when integrating data to and from your Web analytics system:

Ad Serving Systems. By feeding data from Atlas, Dart, Burst, and a whole host of other ad servers directly into your Web analytics tool, you can begin to understand and analyze the behavior of customers and visitors at the campaign level or at the specific ad level. You can look at impressions, clicks, click-through rates from your ad servers and directly compare them against conversion rates and other custom measures in your analytics tool. The purpose here is to improve, optimize, or eliminate the lowest performing campaigns/ads based on your goals and increase your best performers.

CRM Systems. Sending data out of your analytics tool to your CRM mart can enable a corporation to deliver on some very powerful and profitable activities, such as being able to identify and record specific events that customers perform on a site (such as a white paper download or shopping cart abandonment) and automatically trigger actions that promote value (such as following up with an email to the customer offering other white papers in the same or related topic areas or a discount on products abandoned in the shopping cart.) You can also record key activities that the customer engages in on the site to enhance their customer record. The purpose is to improve your knowledge of customer activities on the site to enhance the value chain and inform the sales process (among other things).

Financial Systems. It's not uncommon to see proxy values loaded into a Web analytics system to provide an indicator of the financial value of a transaction or other value-generating event on a site. But why use a soon-to-be obsolete proxy value, when you can directly connect your analytics data to real monetary value? For example, you can join actual product price or margin data with transactional data on the site to let stakeholders view the revenue or profit impact on the business today, not when the next report comes from finance. The purpose is to better understand how the site is monetized and what products outperform others. Or you can even use these data at a visitor level to identify lifetime value.

Custom Data Warehouses and Marts. One of the more exciting integrations you can do with Web analytics tools these days is join view or visit data (and thus visitor data -- as long as you use the right tool) with dimensions in your data warehouse or custom data mart. For example, online publishers may not want to field values in tags or parse urls, so instead they join a dimension from their data warehouse with a view or visit table in their Web analytics tool to bring to light how a specific customer attribute or segment is performing against one another.

Third-Party Systems. Certain vendors have concentrated their efforts on integration with third-party systems, such as qualitative survey tools, email service provides, multivariate testing tools, and paid search tools. The benefit of deploying these integrations can be enormous when placed in the proper business context. By enabling these integrations, you can associate things like site satisfaction scores with the specific urls people viewed on a site and thus approximate the impact of user experience on satisfaction. You can compare various KPIs for different email campaigns or better understand the search terms that drive your specific site KPIs.

Without a doubt, integration of various data with Web analytics tool is no longer a white whale. Instead integration has turned into the holy grail for companies that want to automate business processes based on Web behavioral data, enhance their competitive advantage, maximize their sales opportunities, and understand how to best optimize the site against goals to improve the user experience. How have you integrated data from or with your Web analytics tools?

Tags: metrics
Recommend (31)
8 comments about "Web Analytics Integration: Holy Grail Or White Whale? ".
  1. Jeff Greenfield from C3 Metrics , July 24, 2009 at 5:33 p.m.

    Judah -

    Great post as a 'heads up' to what's possible ...but the reality for corporate clients with separate departments handling different channels - makes integration all of this data extremely difficult.

    I would have added to your recommendation that there are 'new' systems out there (shameless plug) that not only handle all of the integration - but also deal with all of the 'heavy' math involved when your are dealing with billions of impressions each month.

  2. David Pavlicko from AVISPL , July 24, 2009 at 7:16 p.m.

    Excellent post- this is something we've been struggling with at our firm for over a year. While there are certainly solutions out there for companies, most of the vendors remain out of our price range - leaving us ( and many others, I'm sure) to try and formulate a solution in house.

    The good news is that while it may be a bit more labor intensive at the start, once you've set up your data correctly and have a decent reporting dashboard - like Tableau (I don't work for them)- you can perform some pretty in depth analysis.

    Of course if you set things up wrong, you could wind up making some really bad decisions.

  3. Matt Hertig from Alight Interactive , July 25, 2009 at 12:53 p.m.

    The key to building a measurement strategy is to start with the end in mind. What ultimately defines success? Is it clicks, conversions, purchases, etc.? Knowing this, organizations can then define the role of each data source. Data sources include ad servers, email providers, ppc providers, and web analytics. These two steps are the foundation to any measurement strategy.

    Once this is complete, you are then able to define an Assembly strategy to bring these data sources together, build dashboards to Comprehend the results, and Empower stakeholders and clients to make better decisions.

    This isn't white whale stuff. Is it easy? No. Is it possible? Sure it is! Holy Grail? Some might say so.

  4. Akin Arikan from Unica , July 25, 2009 at 4:40 p.m.

    So good to see an update from you Judah, as one of our industry's sharpest observers.

    FWIW, I have been blogging a series of posts recently with examples of companies that put one of your integration buckets into practice: CRM systems.

    If readers are interested, here is that link:

    http://www.multichannelmetrics.com/multichannel-metrics-one-year-later-how-far-did-we-get/

  5. Brad Stewart from Adjoy , July 27, 2009 at 11:51 a.m.

    Articles like this are precisely why I subscribe to Mediapost. Thanks for sharing your experience, and helping some of us keep on the cutting edge.

  6. John Gutierrez from Kaiser Permanente , July 27, 2009 at 2:37 p.m.

    We are building a solution that does what this article suggests...we are trying to understand both offline and online and how they both attribute to business goals.

  7. Cesar Brea from Force Five Partners , July 27, 2009 at 8:59 p.m.

    Excellent article Judah!

    I'd be interested in reader reactions to the relative importance of several challenges to getting to this level of analytic integration. I set up a poll here:

    http://www.vizu.com/poll-vote.html?n=175356

  8. Robert Rose from Big Blue Moose , July 29, 2009 at 2:19 a.m.

    Judah
    Fantastic article. Let me add one other that we've seen great results for - and that's integration with Web Content Management. Using Web Analytics to automatically affect workflow on content - e.g. publishing an index or widget of the most read articles on a web site, or the most popular by ranking, or retiring an article from a section of the site after it receives a certain amount of traffic etc... is an excellent way to improve the relevance of your Web content. I'm sure you can see how you might also start to let your imagination run wild with the different things you might do with your Web content with this. Most of the good Web analytics (e.g. Site Catalyst, Webtrends, Coremetrics) have API's to allow this to happen very easily - and even Google Analytics can be made to do this with little trouble.