Is Your Data Deceiving You?
by Kathryn Roy • April 18, 2011 • Analytics, Marketing, Strategy • 0 Comments
Bruce Clay, the SEO guru, cited a remarkable statistic in a recent webinar: 40% of Omniture sites are set up incorrectly, with 20% of those sites having the Google search code inserted multiple times on a page. That will inflate results.
Think of all those well-meaning marketers who are rigorously collecting and using metrics but are being led astray. The darn reports – so professionally formatted – look sooo reliable. But even they could be lying to you.
I highlight this observation because it happens more frequently than you suspect. When I engage with clients, my 1st question is always: “what are the biggest current impediments to increasing sales” but my 2nd step is always to validate what we believe and identify key information we don’t yet have.
One client was operating under the assumption that marketing’s efforts around webinars, whitepapers, and other offers were pulling in leads at a satisfactory rate. But in the validation process, sampling these “leads” and investigating in detail disclosed that all these “leads” were all prospects in the pipeline. Marketing’s efforts may have been useful nurturing these leads but they weren’t attracting them. This insight helped spur new efforts to attract leads.
Another client was deriving comfort from the belief that they “owned” 40% of the market. (Geoffrey Moore in Inside the Tornado notes: “to dominate a segment typically means winning 40% or more of its new business over the past year to 18 months”). But my client was looking at units sold across all vertical segments, not on a segment by segment basis. They were in a weaker position than they realized, but had time to rectify the problem.
Psychologists call this tendency “cognitive capture” – when you focus on a metric instead of the larger context of how the metric is gathered.
How can you protect yourself? Regularly select a key metric and do a deep dive to validate that you are getting legitimate results. Look at both the data collected and how it is collected.
This post has been viewed 554 times.
