When it comes to the pioneers of web analytics I’m not talking about Google Analytics not even AW Stats. At a recent conference address, my first slide showed who I consider the real pioneers of analytics. For the 3 seconds this slide displayed, the attendees had a puzzled look on their faces. I quickly explained that the images were of Noah, Moses and Augustus Caesar. I didn’t dwell on this slide and quickly moved the talk forward by 2,000 years to the counting of newspaper subscribers. I merely deferred the slide to a later hallway discussion or future blog post. A few took me up on the hallway discussions and I’ll assume most others preferred to wait for this blog post. So here it goes.
While it may be difficult for most people to see the correlation between these biblical and historical characters they in their own way set a standard for a specific type of analytics and to which many us still practice today in our interpretation of web analytics data.
Understanding Noah‘s (as in Noah & the ark) involvement in analytics appears simple. He counted the animals coming on board the ark. “Two zebras, two lions, two bear, etc. of each species.” This is what we’d call in today’s world exact counting or taking inventory. Noah new exactly how many animals were loaded onto the ark out of how many potential species.
This was not a simple task, but a basic check list and quick eyes made his job easier. Yet in today’s world on analytics, once we get beyond accountants and inventory bean counters, where does this play a role in web analytics? The people who practice the Noah version of web analytics are the ones still focused on absolute counts (even when it is nearly impossible to achieve). They’re focused on counting page views and visits as if an increase or decrease in these numbers has a magical effect on the companies bottom-line. While this may be an appropriate measurement for a site that makes its money solely on display advertising (sold by thousands of page views CPM), it isn’t applicable to most modern websites today and needs to be shelved.
Moses’ training in analytics was more than his counting of sheep while in exile from Egypt, while he led the Jewish people for 40 years on at least two separate occasions he took a head count of his followers (as commanded by God). Yet instead of following Noah’s lead of an exact count, he counted the heads of only males “From twenty years and above, all those who are capable of fighting, you should number them, you and Aaron" (Numbers 1:3). It’s from this count, that he was able to extract the total size of the population and not just how big an army he was amassing.
In modern times, we call this a poll or audience sampling. This is very close to how modern analytics are calculated when we only count (and track) people who accept cookies. Modern web analysts know that they can’t count the entire population, but by taking a high sample count (typically around 80%) they will get statistically relevant data on user behavior, what’s working and not working on the site, and can easily identify areas for improvement. Yet even with this large sampling, bosses still want the absolute count just like Noah did. So how do you manage requests like this? Do you tell the boss “If audience sampling was good enough for Moses and God it should be good enough for you?”. Of course not, if you like your job, instead point them in the direction of political polls. For example, in the USA during the 2008 presidential campaign, leading newspapers and both political parties were taking polls of typically 2,000 – 3,000 people out of an estimated electorate of 150 million (0.001%) and accurately reflect the outcome of the election (within their margin of error of 3% 4 out 5 times).
What was actually done with the data from this census is no longer known, yet it established a basic principal which we all need to follow to properly optimize our websites. This is the principal of segmentation. When and where possible by segmenting our web analytic data, we get a better understanding of user behavior when we know, where they are coming from (both from a referring site perspective and geographical location), who they are (male vs. female, age, etc.) and where they are going (why did they come to the site), then just measuring exact counts (the Noah method), or basic audience sampling (the Moses method).