Web Analytics 2.0

I really like this book; its written in an engaging manner. Some of the explanations and definitions are perhaps a little too easy and simple, but then that’s also a reason for non-technical folks to give this book a shot. And, if you’re generally into analytics like I am (for example computer performance measurement) then this book may spark a number of ideas and shift some of your thinking around. This review is really a collection of notes I took on the text, things I wanted to remember from the book in the future.

The author’s goal is to update early web analytics with more modern ideas.

I liked Kaushik’s differentiation of degree of engagement, which can be measured directly by web measurements, with kind of engagement, which cannot be (for example, its not possible to tell if a given session had a positive or negative impact on the client without external information). The author does suggest some standard ways to add kind of engagement information using other means, for example exit surveys.

Web metrics have four key qualities:

  • Uncomplex – as corporations become more democratic (matrixed) everyone needs to understand the meaning of the measures
  • Relevant – every business is unique, your metrics need to measure something useful to your business
  • Timely – you need to be able to gather and analyze the metric as quickly as you can make changes, but no more quickly; sacrifice real-time for quality and meaning
  • Instantly useful – everyone knows what it is and what to do about changes in value just by looking at it

Couple of great suggestions:

  • Don’t wait for perfection, this is the web after all, and
  • Don’t worry about real-time – you can’t make decisions fast enough

Related to the lack of need for real-time analytics, you might also consider data decay; Kaushik provides 6 reasons that historical information over 6 months old is mildly interesting, but only mildly.

The four simple questions to ask first about your site:

  • How many visitors are coming to my site?
  • Where are visitors coming from?
  • What do I want visitors to do on my site?
  • What are visitors actually doing?

Three most important exit survey questions:

  • What is the purpose of your visit to our website today?
  • Were you able to complete your task?
  • If you were not able to complete your task today, why not?

Create a testing culture – I believe this is critical for almost all areas of endeavor, but especially for web development and analytics; run experiments and gather information; create hypotheses and prove or disprove them.

There’s a great section of the book on comparative analysis, which you need to do if you want to compete in a niche on the net.

The author spends most of a chapter on the special measurement of blogs; typically blogs have different goals than for example a product website, so traditional measures are inappropriate. He urges you to calculate an ROI for your blog, and suggests specific things to measure for inclusion.

There’s a lot of good material around analyzing the performance of videos on the web. This includes clicks, where people stopped playing a given video.

The Data Quality Cycle

The Data Quality Cycle

Five rules for high-impact dashboards:

  • Benchmark and segment
  • Isolate your critical few metrics
  • Don’t stop at metrics – include insights
  • The power of a single page
  • Churn and stay relevant
Automated Behavior Process

Automated Behavior Process

Web Analytics Evolution

Web Analytics Evolution

Kaushik discusses search strategies at length, and suggests Google’s Search-based Keyword Tool <> for creating a successful long-tail strategy. I also learned a lot from his discussion of upper funnel keywords.

And on p. 400 he lists a number of great blogs to read that deal with web analytics.

And if you want to hear the author talk about his book: Podcast from Duct Tape Marketing: Analytics from a really smart guy.

Go read this book, its really fun, very interesting and I think there’s tons of useful information inside.


There are no revisions for this post.

Leave a Reply