Skip to Main Content

Jul 17, 2008 | 8 minute read

Webinar Recap: Web Analytics Unleashed

written by Linda Bustos

avinash kaushikWe were thrilled to have Avinash Kaushik join us for 3 Things to Die For: Web Analytics Unleashed today. It was an information-packed hour with a brief question and answer period, but Avinash has a reputation for being very helpful answering questions via email and has invited you to get in touch with him to provide your feedback at ak AT Also, do not hesitate to post your questions in the comments of this post.

Even though this was an ecommerce webinar, everything Avinash talked about is useful to anyone with a website. If you missed it, want to bookmark the replay for future reference or forward this webinar to a friend, client or colleague, the 3 Things to Die For: Web Analytics Unleashed replay will be posted ASAP. You can also catch all our previous webinars in our archive.

(The winner of the signed copy of Web Analytics: An Hour a Day will be announced in our RSS feed, have you subscribed to Get Elastic yet?

Context (framework)

Data, data everywhere.

In the past, there was a quest to acquire lots and lots of analytics data. If one found himself unable to make decisions, it was from a lack of data. But in reality, we have access to way more data than we know what to do with. Companies will not die from lack of data!

The goal is not to collect more data – it’s about extracting insight from this data. Clickstream data is great to help understand the “what” – which products sold, what people do within the cart, where did visitors come from. But clickstream doesn't address the "why" of visitor behavior. Many believe "if only I had enough path analysis data, I would understand the why." The reality is, the web is a personal medium. Clickstream data alone is insufficient to explain the “why.”

Solution? Web Analytics 2.0

Reports doesn’t measure outcomes. So it's key to expand the number of sources you have access to. Avinash lays the framework like this:

THE WHAT: Clickstream data
HOW MUCH: Multiple outcomes analysis (measuring micro and macro conversions)
THE WHY: Experimentation and testing (A/B, A/B/C and multivariate), voice of customer (surveys)
THE WHAT ELSE: Competitive intelligence (understanding whats going on outside your business, useful for benchmarking)
THE GOLD: Insights

This webinar's focus is on The How Much and The Why. As wonderful as clickstream data is, ttesting and things you can take back to your business.

Insightful Outcomes from Measuring Multiple Goals (The How Much)

Most ecommerce folks have an obsession with conversion rate. This is a classic mistake, as it's only one way to measure the direct value added by your website. reports the average conversion rate for ecommerce websites as 1.74%. That means 98% + are doing something else on the website. Many of these things are actually micro-conversions such as:

  • Attracting leads for your business, or offline conversions
  • Improves customer satisfaction
  • Establishes you as a thought-leader or information resource (blog, articles, reports, guides)
  • Attracts human resources
  • Et cetera...

Every ecommerce and non-ecommerce website can learn more about how your website is truly adding value / impact to your business by focusing on the "non-converting" visits and measuring successes from them. This subtle shift in thinking will also help you value your website a lot more.

Avinash cited an example of how one company believed its website was adding $13 Million of value before measuring micro-conversions, and after realized it was really "worth" $43 Million to the business.

Assigning "Fake Dollars"

Even though Avinash’s blog is not an ecommerce site, he measures "All Posts" clicks, as it's a stepping stone to becoming a subscriber (in the sense it measures deep interest in his content). He also tracks "About" page views - that makes him happy. He also measures page views for his Speaking Engagements page. With each goal assigned a dollar value, Avinash concluded his blog accounts for $19,000 of value overall.

Measuring this and trending it over time is key to showing decision makers the value of the website and almost always leads to better questions.

Avinash used Fidelity Investments as an example. Since Fidelity does a lot of business in its physical locations, measuring the sue of the zipcode search box can show you customer intent by geography.

Repeat Visitors. Loyalty is certainly a measure of success. You should look at distributions of data – perhaps 40% visit once, and 2.80% visit 201+ times per month. You want to understand what kinds of content your loyal visitors are consuming. What kind of cross sells can you show them? And so on.

Visitor Recency. Another measure of success is visitor recency. Who comes every day and why? How can you better meet their needs?

Pay attention to visit number, time since previous visit, depth of visit (page views) and length of visits (seconds).

So, it's very important to figure out what are the multiple goals, and ask yourself "am I doing anything for the 98% that are not converting?" And, "how do I extract value from that?" Measure outcomes first, then look for reasons and things to fix.

Your ecommerce website is not a one night stand. Understand how many visits are made before conversion with Visits to Purchase and Days to Purchase. Then, segment by acquisition channel to understand the differences in behavior from visitors who arrive through affiliate links vs. PPC vs. organic and so on. Paid search traffic (research mode) may take a few visits to convert. Pre-sold affiliate traffic may convert right away. Understand why your goals are or are not being accomplished.

Your Top Landing Page reports are a great place to start, then segment by traffic source.

Capturing the Voice of the Customer (The Why)

This is about giving a real voice to people on your website – not a fake voice. Give site visitors a channel to communicate and they will. The best way is to use surveys on-exit (not to interrupt the visit).

Avinash says the greatest survey in the world has 3 questions.

Why are you here? Your survey could have the following fields:

  • Researching
  • Looking to purchase
  • Product / service support
  • Check product order status
  • Register a product
  • Update information
  • Other

Were you able to complete your task? Yes / No

This is important. What if 44% of your customers report that they were looking to purchase and couldn’t complete the task?!

If you couldn’t complete, why not?

Pay attention to the feedback! Look for “segments of discontent” and really understand what are the things that are broken. Not your web usability team pontificating on what might be wrong, but really using that customer feedback.

Avinash and iPerceptions have created 4Q, an on-exit survey that uses these key question. Oh yes, it's also free!

Experimentation and Testing

You may have heard that many decisions are the result of the HiPPO, or Highest Paid Person’s Opinion. (Did you know this was coined by Avinash's former employer, Intuit?) Perhaps the HiPPO is responsible for the home page design and content, and it's hard for you to argue against the HiPPO!

A tip: Learn to prove HiPPOs wrong and quickly. Then you can create an "ideas democracy" where
everyone in the company has an equal voice and are allowed to try ideas. Google Website Optimizer can do A/B/C testing for free, or use a paid tool like Offermatica. But use something! Small, big or medium sized business, anyone can get their hands on a tool and get testing in as little as 6.5 minutes!

The simple A/B or A/B/C test is running 2 versions of a page, with one variable on each page. You are testing to see which version of the variable (like home page image, or headline) performs better against the "control" version.

MVT or "multivariate testing" is a bit sexier. Catch the webinar replay to find out if an endorsement from Oprah can sell more popcorn than a plug from the Today Show. Thanks to MVT, a popcorn e-tailer tested these 2 endorsements with various text promotions at the same time: Spotlight Gift vs. Spotlight Product vs. Customer Favorite, vs. Customer Favorite with Image A (Bags) Image B (tubs).

In this case, Combination 11, with 99% confidence, improved conversion by 24.9%.

Surveys and testing are essential to achieving True Customer Centricity. Involve customers in your website decision making process and you can make better decisions.

One more thing...

Avinash has a word of wisdom he calls the 10/90 Rule: of every $100 you have to spend on web analytics, spend $10 on the tool and $90 on great brains (your web analyst, and not someone who will just "puke reports." Invest in good people, and even hire a consultant for a few months before bringing someone in house.


1. Should you implement surveys at the beginning, in the middle, upon exit or post-transaction?

Avinash has experimented with different approaches and prefers on exit pops up when people leave your website. The advantage is you don’t interrupt them in any way (better response quality, better participation rates). Another way is serving pop-under surveys to visitors who engage a bit more with your site than the typical visitor. The pop-under will be noticed when they close the browser.

2. Surveys and analytics – what to test first if you have limited resources?

Avinash would pick surveys because, even though there is a lot of data in analytics, but if you’re not giving your customers a voice, it’s very difficult to be customer centric. Give your customers a feedback channel. Then you go fix those problems you have on your website.

However, this doesn’t have to be either/or. Both analytics and surveys are available free through Google Analytics of 4Q.

Start with A/B testing - it's straightforward and easy to do. Make dramatic changes and test them. Then tackle multivariate tests, with small tweaks here and there. Start with high Bounce Rate pages in your testing.

Further Reading

Avinash also sent through some links to information for some questions that didn't get a chance to be answered in the short 60 minutes we had available. Thanks, Avinash!

A Web Analytics Primer (Do These Five Things First).

Experimentation and Testing: A Primer

The Three Greatest Survey Questions Ever

Surveys Done Right. Tips, Best Practices, Recommendations

Six Rules For Creating A Data Driven Boss!

How To Excite People About Web Analytics: Five Tips.

Again, you can contact Avinash at:

ak AT