Web analytics with attitude

Jim Sterne, Producer of The Emetrics Marketing Optimization Summit, discusses how to look at web data to find out what customers really think of your site.

Measuring clickthrough, pageviews and revenues is revealing, but it's a bit like asking an in-store shopper how well they like your store based on the time of day they came in, where they walked and how much they bought.

Do they "like" the store? Different story.

I spoke to Roger Beynon, Chief Strategy Officer at Usability Sciences to get the inside scoop on how you can analyse integrated click stream and survey data to come up with a picture of how different people feel about a website.

Beynon refers to web analytics systems as DRIP - data-rich, information-poor.

"Web analytics can, for example, tell a site manager exactly how many visitors dropped out of a checkout process and the precise points at which they quit," he said.

"Web analytics provides little or no information, however, as to who those visitors were or why they left the process".

This reflects the feelings of Larry Freed, President and CEO of ForeSee Results, who said, "When you read log file or sophisticated analytics reports, you surmise where to focus your attention, but when you look at actual customer comments, they'll tell you where you need work.

"Then use the analytics to figure out how well you're fixing the problem".

"Web analytics captures no demographic information on who is coming to a site, or on why they are coming," Beynon continued.

"If a site doesn't know who's coming and why they've come, meeting visitor needs involves luck as much as design, forcing site management to make decisions on content, and redesign armed with incomplete information".

Beynon is not too keen on trusting surveys by themselves, either: "Visitor satisfaction surveys measure a site's effect on its visitors, but offer limited insight into the user experiences that produce the effect".

For that, he submits, you have to look to click stream data since it shows where visitors went on the site, and what they did.

If your survey asks questions about intent, then you can learn why they did it.

Beynon and Usability Sciences have come up with a way of visually reporting how people feel (like any good survey firm would) but they do it by overlaying attitudes on top of navigational data.

They call the result Attitudinal Analytics.

I've touched on this at past Emetrics Summits and I'm sure it's going to create a real buzz at the London event in May.

"Attitudinal analytics illuminates the entire user experience, affording clearer visibility into causality; in turn, more accurate understanding of causality makes possible the design of more accurate solutions, which lead to greater, faster improvement," said Beynon.

The attitudinal part of all of comes from asking people whether they were successful in accomplishing "their" goals, instead of focusing exclusively on your own (sales).

Those who were happy about their website experience show up at once end of the spectrum, those who were not, at the other end.

Then comes the fun part: What-if, in living colour.

How did people who had successful visits get from the home page to the shopping cart? Where did the unhappy folks bail out? Where did the unhappy people get flummoxed? Of those who deemed themselves successful, how many are likely to return and buy again? How did their clickstream differ from those who were successful but said they were unlikely to return? Did the happy/successful people click on something more often than those who left unfulfilled? The next layer of detail to throw into the mix depends on the questions you ask.

You can compare and contrast visitors based on demographics, dollar value, lead source, product selection, brand affinity, visit frequency and, well, you get the idea.

The point is that attitudinal analytics give you a little more view into the hearts and minds of your marketplace.

That's just what we've been hoping for.

Back to top