Web 2.0 Expo: Web Analytics 2.0: Rethinking Decision Making in a "2.0" World by Avinash Kaushik

One of the highlights of the entire Web 2.0 Expo in New York City was Avinash Kaushik's talk (given on the afternoon of Wednesday, Sept. 17): "Web Analytics 2.0: Rethinking Decision Making in a '2.0' World." Filled with confidence (i.e. with loads of chutzpah), Avinash, a consultant currently engaged by Google, explained the idea behind web analytics. Again, what follows is my summary of his points which were flying too fast to be captured accurately. Do not take them as the absolute summary of his talk. I assume they are distilled in his book (free copies of which were distributed at the end of his talk.

In the traditional world, there was always a struggle to obtain more data. Today, there is more than enough data on user habits - in fact, so much, that we may not always know what to do with it. There is clickstream (i.e. recording the number of clicks on a web page) but this data does not give much insight into behavior. Why not? Because we don't have all the infomration we need from the interweb. Thus the birth of Web Analytics 2.0.

The aim of Web Analytics 2.0 is to measure more and different things from websites. All the current data is telling you what happened. But these tools are bad at measuring "how much"? Some explanations:

• The what: clickstream
• The how much: multiple outcomes analysis
• the why: experimentation and testing, the voice of the customer (missing from the web today!)
• the what else: competitive intelligence and insights

It's about competitive intelligence. If you're not using competitive intelligence data, it's as if you're driving in a car with opaque windows. At Google we use bounce rate. Search engines determine where you are going to enter a website. If you look at your top 20 pages you'll see which ones are the poor ones. It's a question of "sacrificing usability at the alter of sexiness." With Google Analytics you can understand what sends the top ten users and why lower traffic comes to a website. Traditional analytics are about collecting information whereas Google Analytics are about analyzing data and deriving meaning from it.

For example, take a comparison between the web and television: on television you can't track things. But on the web, a viewer's actions can be entirely measured. Indeed, this is how movie studios are producing movies - by seeing how people watch them on the web (as a product of web behavior, which is entirely measurable).

Let's talk about conversion rates. There is macro conversion rates (the transactions of an entire website) and micro conversion (transactions needed to create a action). These are ways to quantify value.

A website must have multiple goals and in analytics, must have many different goals. But you must track your goals. What about the metric of recency? Are you encouraging return visits? You can ask: Why do you exist? If you can say what yo uare trying to do with your website, there's no reason not to be able to measure it.

This gives you a way to understand "the why." It gives customers a real voice. The primitive way of doing this was with surveys - all of which had three questions: Why are you here, were you able to complete your task, and if not, why not? The goal was finding segments of discontent.

[At this point my notes are less focused - Avinash was going a mile a minute and it was hard to keep up with him, especially since I wasn't totally familiar with his terms.]

Avinash collaborated with iPerceptions to create a questionnaire.

Scalable listening. Experimentation and testing. Axiom: Hippos create bad websites. (Hippo = high paid person's opinion.)

Learn to be wrong. Or prove others wrong, fast.

It's irrelevant what web creators want. Let the customers tell you what works through use of analytics. Don't guess or impose - partner with them. Competitive intelligence will enable you to benchmark expertise. Instead of "ready, aim, fire!" you'll be able to say: "research, target, fire!"

Learn your targeted keywords. Study Google Insights For Search. But you need a holistic view to understand all of people's behavior with regard to a website. We're still far away from it.