The Three Heads of Online Analytics

Analytics Competency Center

Cardinal Path articleWho would have thought that suggesting a new definition of web analytics would spark such a great conversation! As my friend Justin wrote in a follow-up post, maybe we need to radically rethink our approach. While Justin addresses the “tools automation” aspect, I would like to share the framework I use as part of the Online Analytics Maturity Model to evangelize the role and benefits of an Analytics Competency Center.

Most people see analytics as some kind of bipolar disorder, a constant fight to find the balance between two apparently unsolvable personalities: marketing and IT.

But I believe there is more to it, as you will see in this article, it’s more like a three-headed monster waiting to be tamed.

Online Analytics

Business: start with business in mind

There is little value in doing analytics if it isn't to bring actionable insights - otherwise it should be called "reporting". Analytics should always start with a thorough understanding of the strategies and business goals which are communicated as requirements (and constraints), preferably stated as S.M.A.R.T. online objectives. In a perfectly mature organization, those would come directly from the business stakeholders. The reality is, most often, you need a business analyst to "translate" and bridge the gap between business considerations and something analysts can work with.

Technology: understand the medium

For most marketing people, this is simply "IT" and often stated with a bit of a condescending tone… There needs to be people who understand the capabilities and constraints of the technologies involved in the solution. Be it HTML, CSS and Javascript or understanding how each social media differs, how Facebook apps can be measured, understanding the differences between iPhone, iPad and Android mobile devices, etc.

You need people who know the ins and outs of the website, its technologies, how it interoperates with the back-office and how the information architecture and page naming taxonomies will impact SEO and data collection. You need people skilled at properly instrumenting the website for whichever tools you are using and how the data is stored and can be leveraged. As you can imagine, this in itself is a critical aspect and can quickly become very complex. Ultimately, the goal is to provide the means, the tools and the data to the 3rd head.

Analysis: the Sherlock Holmes factor

This is where statistics gets involved, where the analytical mindset is put to work in order to identify patterns, trends, and understand correlations. An investigative mind, problem solving methodologies and abilities to synthesize information to their most effective communicational forms are essential. The outcome is clear and simple: to provide actionable insight and recommendations back to the Business head.

Analytics Competency Center

Taming the beast

Nobody can claim to be fully skilled at all three dimensions, thus, the importance of a multidisciplinary team. Every analyst I’ve ever met was strong at one dimension, maybe two at best, but weaker in the third. Yet, all three dimensions are essential to successfully develop an analytical culture and provide the highest value. Based on my work on the Online Analytics Maturity Model, ROI of analytics will stagnate until at least two people with complementary skills are teamed and supported by a proper level of authority, a bit like the concept of "pair programming" in agile development approaches.

Stéphane will host a full day workshop on the concept of Online Analytics Maturity during eMetrics San Francisco in March and eMetrics Toronto in April. He will also present introduction sessions at several events in 2012, see his complete speaking schedule or contact Stéphane for additional information.

Subscribe To Our Newsletter For Monthly Updates



Your e-mail will be kept private

Simon Roberts | January 2012

I could not agree more with your approach. Too often people look at different reporting outputs (Google Analyics, Omniture...) but rarely are able to put anything they see into action. Often the reason is that there are technical problems to put these changes into action and this was often our case that great insights got put on the back burner waiting for the next site design, which could have meant months or years. About 3 years ago our company http://www.cognesia.com decided to try and knock this on the head and thought that the only way for Analytics to really add value to the bottom line is to make it possible for the analytics to drive the marketing and and automate a way so that insights spotted by marketing and analytics professionals could be put into action with minimal involvement from the IT department. The two most obvious uses of insights are to better tailor site messages and open a true marketing dialogue with returning visitors and avert the "first date syndrome 99.99% of sites only offer). Therefore to deliver to different audiences a personalized set of marketing messages and to continue this dialogue through email marketing messages. Just as it will take a sales person 2-3 calls and different messages each time to win a sale the same needs to apply online.

Now Cognesia is able to allow marketeers and business professional see where improvements can be made and then put live optimised content using a seamless solution. It is very satisfying for a business to see changes quickly lead to improved conversion and reduced bounce rates and this then drives greater investment in that process, which is good for the whole industry. Sad to say but Google Analytics blocks the ability for anyone to extract or use this highly valuable customer behaviour data to trigger improvements to your site but other solutions exist.

S.Hamel | January 2012

@Simon: thanks for commenting! I think the ability - or I should say inability - to take action often stems from a disconnect between the “business” dimension and the two others. That’s why in my definition of analytics (see http://online-behavior.com/analytics/definition ) I was insisting so much on “optimal and realistic”. We could argue that a recommendation which doesn’t lead to action is either non-optimal, unrealistic or simply not aligned with the business.

As for your point about GA, I prefer to avoid leading the conversation toward “tools” arguments :)

April Wilson | January 2012

Spot-on, Stephane! You very clearly articulate something I've been in the trenches preaching for years. Our profession is largely seen as a dual personality like you describe - marketing and technology - like Bruce Wayne and Batman are part of all of us.

The reality is that what we do - and most importantly, what we MEASURE, needs to encompass all aspects of the business. I've even had the experience where your triangle was blown up into a pentagon, and Finance and Customer Service also had seats at the table. Everything we do in the digital marketing space has a trickle-down effect throughout the rest of the organization and integration should always be step 1.

Nice visualization and argument!

Daniel Waisberg | January 2012

April, I like Pentagons, but I must confess I like Dodecahedrons even more! I think analytics should be everywhere:

  • Finance: helping doing predictions
  • HR: website numbers should tell where to focus recruitment efforts
  • Customer Service: help understanding how to reach out to customers
  • and others

But one special place that I have been seeing Analytics interest is the Product. I very often get hired by product teams trying to understand how the product is performing, how to improve experience, and what to develop next. These people are very interested in understanding their audience.

But I believe Stephane's framework above is a broader view of the ecosystem, focusing on the main elements rather than on company departments (right Stephane?) I believe that Analytics should be decentralized to be part of every single department, because all of them will be benefited from it.

Sounds reasonable?

S.Hamel | January 2012

@April & @Daniel: It really reassures me when I see comments like yours – shows I am (we are!) on the right track! I strive to make things simple – conceptual rather than specific – that’s why I mentioned “business” instead of “marketing” (or any specific business function). As Daniel pointed out, the framework I’m suggesting applies to every business aspects. The dangers of adding Finance and Customer support is:
a) Make the concept more complex and
b) Unintentionally suggest other functions like sales, engineering, HR, etc. are less important. The graph in this articles about Strategic Business Functions (http://wiqime.wordpress.com/2010/10/22/strategic-business-functions/ ) shows a good approach to pin-pointing areas where analytics can bring value. Once we understand the business context, goals and constraints, and good data, we can put our analytics skills to work and jointly, with the help of SME (Subject Matters Experts), bring the most optimal and realistic recommendations.

Christopher Berry | January 2012

Thank you for the post. I want to elaborate.

There are T's, S's, and A's. T's are technical analysts. S's are strategic analysts. A's are Analytics (actual analytics, not fake BI reporting) analytics.

Reporting is not insight generation. Whereas reporting has benefits as an output, it is not the sole output.

T's tend to become religious about their toolset. You know them. The Omniture Fan Boys. The Google Fan Boys. The Webtrends Fan Boys. They are defined by their tools, much in the same way that BI's allow themselves to be defined by their tools. And that's fine. Just know that they are not strategic because they can't have a choice.

S's are strategic. They not only understand business, but they are capable of expressing tradeoffs and building business cases. They're political. They're great storytellers. They're awesome communicators. They can be sociopaths. Strategic analysts tend to be the interface between the business and the T's and A's, but not always. T's get frustrated with S's because they don't understand the technology as much as they should. However, a good S is not an empty suit.

A's are actual analytics professionals. They know statistics. They know how to operate statistical software. They understand bias. They understand confidence intervals and confidence levels. They understand modelling. They know how to execute a regression. Analytics is science, so, it's here where you'll find the most scientists. They experiment. They execute experiments. They're the principle ones who generate learnings and actual insights.

Hiring a T is very hard in this economy. Hiring an S is damn near impossible, though, there are a lot of wannabes. Hiring real A's, not just reporting monkeys or squirrels or whatever the hell we're supposed to call them, is getting easier.

Hiring a competent T-A is exceptional. Finding a S-A is getting easier. Finding a good T-S-A is damn near impossible. Most of them are consultants.

Web analysts need to identify their weaknesses, be it T, S, or A - and move to vigorously eliminate them. It may take years to do, but regressing into dashboard generation is a race to the bottom. I don't believe that most are on very good trajectories right now, and it's having a huge impact on the industries retention rate.

Thanks again for the post. I see your framework and say, "it takes an orchestra".

S.Hamel | January 2012

@Christopher: "Wunderful!" - so we could say the formula to Business Value is BV = f(T,S,A). Since we're right into the new year resolution period, your point about learning our strenghts and weaknesses should inspire us. I found this article to be inspiring: "What skills will you develop in 2012? Focus on the stars, not the monkeys!" at http://www.bridging-the-gap.com/what-skills-will-you-develop-in-2012-foc...

Peter O'Neill | January 2012

Hi Stephane (etc),

I totally agree with the three heads/types of web analysts as described above. However I disagree with the need for A's to know and use statistics to do their analysis and produce recommendations. It is one possible approach but only one of many.

I would argue that one requirement to create actionable recommendations is creative/lateral thinking. And similar to how someone can't be all three types, that usually doesn't go hand in hand with hard core data crunching. Or maybe it is just me, I understand stats but I don't use - can identify the insights though data exploration and discovering the gaps.

Peter

S.Hamel | January 2012

@Peter: fair point - I think knowing basic stats (and in some cases advanced) is a must. When I wrote the series on the math behind web analytics I was astonished at the lack of understanding of even the most basic things (http://www.cardinalpath.com/math-analytics-box-whisker-plot-in-excel/ & others) - and I don't pretend to be a pro of stats myself, far from it!

That being said, when I present Online Analytics Maturity sessions & workshops I stress the only way to be successful is "to be creative in a continuous improvement process". There is rarely (never?) a single big thing you can fix to magically double your sales. The only way to improve is to pay attention to details and fix them, one by one. Thinking outside of the box and being creative contributes to bringing the "most optimal and realistic solutions", otherwise you can only "copycat" the same solutions someone else (likely your competitor) already implemented, you can only bring the obvious solutions.

So yes, I agree with your point there are other ways than "pure stats" to bring actionable insight.

S.Hamel | February 2012

For anyone interested in this conversation, don't miss the other posts from my friends:

Post new comment
The content of this field is kept private and will not be shown publicly.
Type the characters you see in this picture. (verify using audio)
Type the characters you see in the picture above; if you can't read them, submit the form and a new image will be generated. Not case sensitive.
Online Behavior © 2012