The Most Important Post on Web Analytics You’ll Ever Read

When John Lovett joined Aurelie and I here at Web Analytics Demystified earlier this month an awful lot of people said, “Hey, nice job getting such nice guy on board,” “We love John, he’s great,” and “Man, what a great addition to your team!” Clearly John has the respect of the industry, but one thing that remained an open question in some people’s minds was “how will John make the transition from the ivory tower an analyst sits in to the ground floor where consultants actually do work?”

I admit, I wondered that too in a way, having made a slightly different transition myself years ago. It’s not easy to come away from a situation where you provide advice but are tasked with, honestly, doing very little real work. During my own tenure at JupiterResearch years ago I ensured my own connection to practical web analytics by writing my second and third books. But John had been an analyst for nearly 10 years … and so wondering how he’d hit the ground was a reasonable question.

Wonder no more.

While John has already contributed greatly to the businesses bottom line and helped out with one of our largest new retail clients, he absolutely floored me this morning when he published his post Defining a Web Analytics Strategy: A Manifesto. I asked him to elaborate on some comments he made at Emetrics where he essentially poo-pooed the use of so called “Web Analytics Maturity Models”, describing the almost religious zeal some people seem to have when talking about models and declaring himself as a “Model Atheist.”

Having written the original Web Analytics Maturity Model back in 2005, I have had first-hand experience with their failure to produce anything more than a generalized awareness that most companies simply don’t “get” web analytics, something that we more or less all know already. But honestly I was surprised when John took this position on the subject because, well, in my experience those that don’t do, teach, and models are a classic teaching tool.

I had assumed that as an analyst John was a teacher, not a do-er like I have been for years now in my capacity as a practice leader, consultant, and web analyst. Man was I wrong …

John’s “Manifesto” is perhaps the most lucid yet succinct explanation I have ever read detailing the steps required to make web analytics work for your business (as opposed to the other way around.) I almost asked him to edit the post for fear that he was opening our kimono too much, but if Social Media has taught us anything it has taught us that transparency is king. The fact that he managed to encapsulate what others have been trying to explain with long-winded speeches, tangential arguments, and downright rude behavior is a huge plus.

Some of you may read John’s manifesto and think “Gee, this seems to point to the need for outside consultants” which is a fair criticism. But before you react consider two things:

Consultants (like us) have a tendency to, you know, recommend consulting. Everyone’s perspective arises from their own personal biases, regardless of how many times they declare the contrary. We are consultants, consultants who want to feed their children. Forgive us our bias and we will forgive you yours …
Consultants in the Enterprise are like death and taxes, we are more or less inevitable. Often times an outside perspective is exactly what the business needs to actually start to act upon the message that otherwise great employees have been stating for years. Other times the business simply stops listening to their employees and won’t make a move until McKinsey, Bain, or Demystified come in and charge big money for insights that were already there. Either way, ours is the second (or is it third) oldest profession and it must be for a reason …

I would challenge you, dear reader, to spend some time reading John’s post and considering what he has to say. Think about how you could apply his ten insights to your business regardless of whether you turn to consultants for advice or not. Listen to your business partners needs, put away your models and roll up your sleeves, transcend mediocrity, establish your own waterfall and embrace change!

When I said “web analytics is hard” I meant it, I really, really did. But I wasn’t trying to box anyone in or establish myself as some kind of amazingly wonderful “guru”, I was simply telling you all the truth based on my dozen years of experience in the sector. Yes, getting started can be easy; yes, making Google Analytics do stuff can be easy; and yes, you can do an awful lot in an hour a day if you simply apply yourself to the task … but the problem is that within any business of size, complexity, or nuance — which is to say all businesses everywhere — the act of getting from raw data to valuable business insights that you can repeatedly take action upon is apparently so freaking difficult that almost nobody does it.

How is that “easy?”

You all know I love a good debate so if you disagree with my comments here please let me know. If, however, you have something to add to John’s manifesto, I would encourage you to comment on his blog post directly.

Happy Holidays, everyone.

Posted Tuesday, December 22nd, 2009 | 4 responses | Share, Save or Email

http://christopher-berry.blogspot.com/ Christopher Berry

Eric,

The assumptions I’m making:

1. Our understanding of the world is incomplete
2. The human world is complex
3. The human social world is even more complex
4. The production possibility frontier exists and forces us to make choices
5. Making a choice that involves a tradeoff within the PPF curve is a strategic decision

So, here’s my argument:

1. To grapple with complexity, an analyst has a choice: wild guessing or an algorithm of systemic learning
2. The consequence of that choice will bring the analyst closer or further from the PPF curve
3. Science offers an algorithm for learning
4. Science is Theory driven
5. Latent within a Theory are biases about which choices are important, typically, a model or many models
6. A Model an abstraction of the complex world that is coherent within a theory

So:

1. An analyst may use theory-informed models to make choices that bring them closer to the edge of the PPF curve

And, I’ll go further in saying:

1. The same process in use today by eScience and health researchers can be used by marketing analysts to assist in making optimized strategic choices

I’ll hypothesize then:

Analysts who use theory-based models to abstract reality (Ie. Science) will progressive make better choices, will advance faster, and will derive a sustainable competitive advantage for their firm than if they were guided by a different algorithm: namely, wild guessing.

—

There’s nothing wrong with analysts using models to derive competitive advantage.

If you’re saying that models might not be the best way to communicate with business managers – I say, “It depends on which industry” but, generally, in general, yeah, nobody has the patience to listen to us.

Results speak. Results matter. Show them the money, don’t talk about it. So, on that point, sir, we agree.

For us, for the scientist-practitioner, we need models. Sure, they’re abstractions. But they’re necessary abstractions because we’d never progress without them and it’s the best system we have. (Or that I know of.)

Is there a better way?
http://www.autotrader.com Steve Robinson

Eric – Most of my involvement with web analytics has been from the management perspective, but I am now becoming more hands-on with the discipline as my need to gather insights on my own has grown. After reading the recent postings from you and John, I am more convinced than ever that something you have implied in your presentations needs to be stated outright.

I come to web analytics from a long career in traditional business intelligence and I am struck by how closely the evolution of web analytics has paralleled that of BI, beginning around 25 years ago. Of course, this should not come as a surprise since both are just forms of decision support with their own specific tools and data. Those of us who were part of the BI emergence feel like we are seeing an old movie again, and we know how it ends. Nearly all of John’s points in the manifesto have been made in connection with BI in one place or another. Successful BI consultants innately appreciate the wisdom here. I say this not to downplay what John wrote, only to encourage those who want to understand what makes for effective decision support within their organizations to seek out the advice of those who are successful with traditional BI applications. As Eric likes to point out, competing on analytics applies across all forms of data-driven decision making.

I believe the convergence of web analytics and BI is inevitable as the organizational and data silos between the two practices are broken down and the vendors continue to consolidate. This will have significant implications for all the stakeholders, but that movie will be produced sometime in the near future.
http://www.webanalyticsdemystified.com eric

Chris: Thanks so much for your comment. I don’t particularly disagree with you but I think the most important statement you made was this one:

“Results speak. Results matter. Show them the money, don’t talk about it.”

The issue I have with the model-making and navel-gazing is that our industry is suffering from a long list of problems that need real solutions (e.g., “Results”.) While it’s interesting to see community outsiders wax philosophical about what is broken in web analytics I, for one, don’t have the luxury of time to sit on the throne of “thought leadership” while our clients wait for results.

Do you?

You ask “Is there a better way?” to which I can only answer, “I certainly think so.” How are you successful in your practice? Are you using models? Or are you doing the real work with your sleeves rolled up? While your argument for the former is duly noted, I suspect that like many of us you are actually doing the latter.

Steve: An excellent point and one I certainly agree with. Have you read the white paper I wrote about the “Coming Revolution in Web Analytics?” You can download it here and I suspect you’ll find a lot to agree with.

All the best to you both.
http://www.christopherberry.ca Christopher Berry

@Eric

I appreciate the emphasis on doing over theorizing.

To your point:

Are Grand Organizing Theories of analytics necessary?

Well, I think they’re inevitable. There are as many Grand Organizing Theories as there are analysts.

There are as many models as there are people.

And that goes for our clients too. If you talk to a Branding guy – they’re going to have a totally different mental model from a Direct woman. And do those mental models influence which subset of metrics we select for our strategy? You betcha! That’s the A-league, baby!

To your question:

My Grand Organizing Theory of analytics is Science, and, recently, abductive thinking. Inherent within Science is the use of models. The Effect are results.

And I’m with you on the need for the demonstration of Effect.

If you’re effectively annoyed by ‘outsiders’ and their Theories, keep out of social media measurement. ;) Phonies ahoy!

What really changed for me in November and December was the realization that I could expend energy on discrediting alternative versions of reality, or, simply, expressing what I believe and focus on innovation. I can’t change the community. I can only incorporate the good ideas I see and change out the bad ones within my own mental model.