The fantastic 4: Four things to get right about your Web Analytics project

The Web Analytics blogosphere is full of articles, questions and opinions about 2 things: processes and selecting a Web Analytics Suite or Vendor. I also have read and heard an incredible number of citations of Avinash´s 90/10 rule that stresses the importance of people. While I also believe these subjects are perfectly relevant, I would like to shed some light of why I think only the right combination of 4 things will get you to succeed on your way towards a successful Web Analytics solution and culture. The combination of all 4 things, not just one or two of them.

Fantastic 1 – process

The overall approach of revamping Web Analytics always has to include a strong look at the process of Analytics and Optimization inside the organization. How does the process of implementing requests or needs for analysis or reporting look like? Does it include a cycle that questions and implements their relevance towards business or unit goals? Are KPIs and web metrics comparable and consistent across all units? What happens next? Is there a process to suggest optimization steps? Are these monitored and re-measured?
All these parts can but do not have to be implemented at once, you can always start with smaller steps and get mature in each of these areas as soon as you have insight about their real life performance.

Fantastic 2 – data

Yes, data. Not customers, events, interactions just pure data. Everything else is developed from it. You have to work on the overwhelming part of the Web Analytics picture that still consists of data collection, tagging, fragmented content creation and delivery, passing additional parameters to tags via different techniques, content/event mappings, integrating other data sources (or even –oouuh- logs), looking at privacy issues when getting a clearer picture on customers than what cookies deliver and all that stuff. Don’t underestimate the effort to roll out all these on a bigger portal that may consist of several applications and suites that deliver content, user experience, and functionality to people. It is rarely as easy as: Just implement this code in your CMS footer and you are done. It is not always hard but be aware that it might be, anyway.

Fantastic 3 – competence and people

Should we establish a center of competence with analysts who know the hell of a lot about web data, understand the business goals of all our units and can make sense of the connection of both? Or should we try to build up analysts in each of our units who are also subject matter experts for their part of the business?
I still haven’t decided what might be the best approach. But either approach is hard and means more to your human resources development than just sending people to some vendor trainings. You need to hire, bring in or develop people who are passionate about the web, about analysis and about business and you have to keep them with you.

Fantastic 4 – a solution and even more competence and people

In the end, or in most cases unfortunately at the beginning you will have to figure out that you need a solution that is able to cover your process, your fragmented data collection and is able to tell you all you might ever want to know from that data. So the challenges of all 3 steps mentioned before cumulate in your solution and vice versa.
You will require assistance from the vendor in terms of Web Analytics consulting also. Take a strong look at their professional service team and make sure they are not only capable of implementing and supporting but also in developing your people with their expertise and to support and enhance your process. Consider an independent consultant if they don’t.
If you got the other pieces right, you have clearly moved on from reporting to web analytics if your solution vendor can deliver on these pieces. As soon as you got the first 3 right, you will need a solution that is strong in combining every possible dimension of data, in segmenting and in filtering. You will also need to have adhoc analytics capabilities because otherwise your optimization processes can not succeed because of the iterative nature of web analytics: Every KPI you get and every analysis or report brings up new questions and new needs for filters or segmentations or new combinations of web usage dimensions or attributes and customer data from other channels.

Perhaps you will need new analysts if you covered all these steps and built a “Fantastic 4” web analytics solution. Perhaps what you called web analytics before was in reality reporting or marketing performance optimization. Iterative Analysis and true online business optimization is way more sophisticated than web reporting. Deriving value out of segmentation and combination of trends and behavioral attributes is another league than looking at KPIs.

Ein Kommentar zu “The fantastic 4: Four things to get right about your Web Analytics project”

  1. Patrick

    Hi Oliver,

    as you know I’m rather new to the field (having spent most of my time learning about the web (mostly SEO), marketing and statistics in general). But I have to say I find this pretty interesting:

    “Perhaps what you called web analytics before was in reality reporting or marketing performance optimization. Iterative Analysis and true online business optimization is way more sophisticated than web reporting. Deriving value out of segmentation and combination of trends and behavioral attributes is another league than looking at KPIs.”

    I’m really into analyzing data, but unfortunately what I’ve seen and read about web analytics so far has been mostly the very basics..more like definitions of the basic metrics, what tool to choose whether java script or log files are the better option, etc.

    I’ve also read some stuff about multivariate testing and the way Avinash suggested to apply statistical control limits and statistical significance to web data, but unfortunately I haven’t really seen any examples of web analytics, that really got down to the number crunching/hardcore analytics part ;-) .

    I know analyzing data shouldnt be an end in itself and theres obviously no use using complicated data analysis techniques if it doesn’t help improve business goals…and I kind of started doubting if web analytics was really that much about actual analytics, but after that paragraph it seems to be that it pretty much is the way you describe it.

    I was really a bit surprised that when reading a lot of posts in the yahoo group I never really came across any posts discussing analytics techniques but mostly web related stuff.

    You don’t happen to know of any sources on the web where I could see some examples of the more sophisticated web analytics techniques?

    I guess what you mean by segmentation is another type of segmentation than simply “segmenting” traffic by search traffic, direct traffic, etc..and segmenting search traffic by PPC and organic traffic..and then “segmenting” by keywords but rather the type of complex segmentation applied in analytical CRM/database marketing campaigns? sort of the way Amazon uses data mining (or what one would like to call it) to gain a competitive advantage by suggesting appealing products to the client based on browsing history etc.?

Einen Kommentar schreiben: