At the recent CMA conference on Marketing Accountability, I had the pleasure of being on a Web Analytics panel with Scott Crosby, representing Google Analytics (Scott is Head of Urchin Web Analytics, Vertical Markets Group). Facilitated by Richard Boire of BoireFiller Group, our topic was Web analytics: converting online metrics to ROI measures.
We kept it informal and digressed to answer questions when necessary. My main focus was on KPIs - that good KPIs are indicative, actionable and support the attainment of organizational goals; how coming up with the right KPIs requires a clear understanding of organizational goals as a starting point; with improvement and optimization as key objectives, ratios and trends are typically more useful KPIs than absolute measures.
Scott used Google Analytics to demonstrate how campaigns can be assessed, how monetization is included in many GA reports, and how funnels can be used to troubleshoot dropout and low conversion. He also showed us an upcoming visual, non-tabular enhancement. The question of why GA is free was addressed up front - i.e. Google believes that if people can easily optimize their Adwords spend, they run more Adwords campaigns. Aligned with this position, Scott concluded by encouraging everyone to measure their campaigns, even if this was done by using one of GA's main competitors, which he listed.
It's a challenge to cover off Web analytics in an hour with any audience, let alone with a diverse audience of marketing managers and data mining specialists, so we allowed for lots of questions. Overall, feedback was very positive. Here are some learnings and areas that were discussed:
- With data miners present in the audience, we need to explain that "logs" are analogous to "records", but not exactly.
- Because data mining analysts often work with a transactional database, if a transaction has occurred, it's in the database. Whether or not a transaction happened is not an issue. So some of the potential inaccuracies of Web traffic data are a bit unsettling to a data mining specialists.
- Questions were asked about tracking of podcasts, whether it was now possible to measure how much of a podcast is viewed, etc. We couldn't answer this and suggested they might want to post a question on the Web Analytics Forum.
- Questions were asked about the prevalence of realtime customer predictive analysis and dynamic content production, sort of like Amazon's recommendations. We responded that walk before you run. Start by closing the feedback loop on current campaigns before trying to optimize in real time.
- Who would make a better Web analyst - a data miner or an IT person? My vote is for the data miner. Although either might be great analysts, problem solvers and are always learning, data miners have had exposure to customers, preferences and segmentation.
- There was quite a bit of interest in benchmark conversions. Accompanied by the usual "be careful how you apply this" warnings, we directed them to the Fireclick Index and WebSideStory's Statmarket as well as the Internet Retailer Top 500 report. Scott was asked if Google publishes benchmarks based on all the sites being tracked by GA. Scott responded that Google has no plans to publish aggregated benchmarks. Also, no data aggregated or otherwise goes over to the Adwords group from Analytics. Analytics has privacy very tightly controlled, which is a great thing to hear.
A jam packed hour, as you can see!