Posted by Andrew Edwards on April 22nd, 2011 at 7:32 am

The condition of web analytics today is a mystery wrapped inside of an enigma.

The superiority of measurable media, loudly proclaimed as the very reason why ads have gone to the web, and the reason why news printed on dead trees is itself dead or dying, and the reason why bookstores and record stores are going the way of the pterosaur; despite these loud claims, measurement itself is notorious across the land for its shallowness, its lack of accuracy, its general state of brokenness and dissatisfaction.

Maybe we have gotten ahead of ourselves. Maybe just knowing how many visitors you have is fine--because, rudimentary as that measure may be, it still beats sheer guesswork (the prior method of measuring audience). And maybe, despite the millions and even billions invested in on line venues, measurement is not necessary to justify any of that spend as long as we all have a little faith in the wonderful world of the web.

Not.

The real story is about lack of planning, lack of resources, lack of trained personnel, lack of understanding and perhaps most amazing, often a lack of simple responsibility. And then add in the woes expressed by all those companies--big ones, usually--that own analytics software as they experience rather shocking problems related to badly underpowered computing. Altogether you have a landscape strewn with failure and frustration at worst; and often enough, confusion leading to inaction.

The amazing part of this tale of underpowering analytics is that all of the problems would be very, very easy to fix if folks would internalize the fact that more resources--talent as well as RAM--need to be deployed in order to get meaningful results.

How is it possible for a company with hundreds of thousands of worldwide employees to have one or two analytics experts servicing the entire need? It is not possible--but at least one company is trying.

How is it possible for a company with an IT-based service model fail to post enough servers to its analytics farm? Strange but true.

How is it possible that a sophisticated organization should be hogtied by the very weak claims made by folks who don't know analytics that "it will take forever" or that "we have a code freeze and so that major realease will have no tracking"? Because developers and agencies don't seem much to care for analytics and often seem to do their best to derail analytics through methods I can only describe as akin to civil disobedience: non-cooperation, diffusion, slowness, deafness and blindness when it comes to following well-documented instructions; even outright refusal to execute their small part of the task in order that their customer may measure the success of their own web site.

Often I have wondered if there is a single overarching reason for this set of peculiar behaviors, and I am not sure there is a neat fit for any single explanation. But I will say that the promise of analytics itself is truth.

Will the truth set you free? Maybe not in corporate corridors it won't. If you're clueless about analytics, and you don't do any planning for it, and you don't do a good job in architecting your site or creating successful content (remembering all the while that "content without conversion equals zero" as Rand Schulman has said)--if you are weak in these areas, and don't embrace analytics as a way to learn to do better, then the truth (as delivered by analytics) may not set you free--except in the way a person who has lost their job is now "free" because (as Janis Joplin once sang), they have "nothin' left to lose". And I don't think this is the type of freedom most folks are looking for.

Is there a way out?

No doubt--and here are some pointers:

1 - devote a line item in all development projects for analytics

2 - build analytics in from the start: yes, in the Gantt chart just like everything else

3 - never let the tool or your own architecture stand in the way of measurement (with the right kind of expertise, you do not have to)

4 - make sure your effort is properly provisioned (for cloud solutions) so you don't lose data when you need it most

5 - make sure you have a robust, functioning analytics environment (for software) so analytics processing doesn't fail before completion

...and perhaps most challenging:

6 - don't let developers tell you when analytics tagging will take place; give them a deadline and hold them to it!

There are many other ways to improve analytics efforts and these are only a few. But the above suggestions do respond to challenges that, over the years, have proven common to perhaps even a majority of analytics environments. So one should not be surprised to recognize one's own organization somewhere in the above descriptions.

Which means it is time to power up.