Using web stats for more

Paul Boag

We all use web stat tools like Google Analytics for tracking marketing campaigns. However, they can also be used to improve your site.

As I continue my efforts to finish the website owner’s manual, I have reached a section on web stats. What struck me was how little most of us use the power of web stats. They can do so much more than monitor your marketing efforts. In particular they can:

  • Help find and resolve problems on your site
  • Improve the quality of the content you deliver

Let us look at how.

Finding and resolving problems

When it comes to web stats one of the most popular figures to monitor is conversion rate. Conversion rate compares the number of users visiting your site to the number that complete a call to action.

If your conversion rate is low, this could reflect a problem with your site. This could be due to:

  • Usability – The user is unable to find the call to action due to poor navigation or other usability issues.
  • Accessibility – For example a particular browser does not render the site correctly and so users cannot complete the call to action.
  • Content – The site does not provide adequately convincing content to encourage users to complete the call to action.

But what consists of a low conversion ratio? That entirely depends on your call to action. For example, an ecommerce site could have a ratio anywhere between half a percent and eight percent depending on the sector and product. On the other hand, a call to action that does not cost the user money should expect a higher figure.

The best approach is to compare a conversion ratio against itself over time. As you make adjustments to your site does that harm or improve conversion rate?

Fortunately website statistics can also help establish what changes will improve your conversion rate. Start by looking at where users exit your site.

Dropout points

When looking at where users exit your site exclude those who only view one page. If you do not the homepage will be at the top of your list. This is because people click through from a search engine, discover this is not the site they wanted and leave immediately. Although this may indicate a problem with your SEO, it does not reflect badly on the site.

Once these anomalies are excluded take a look at the remaining pages. Why are users leaving at these points? Is the content relevant and clearly presented? Is the navigation usable? Are you suggesting a next step to the user or are these dead end pages?

Look at the history of users who dropout at a particular page. How long have they been on the site by this point? What other pages have they viewed? How long did they spend on the exit page before leaving? Does this reveal trends which help to identify the problem?

Sometimes the problem will be obvious, other times it will not. In such cases try usability testing. This will uncover potential issues. If usability testing is not an option try using a tool like Click Tales.

Click Tales picks up where traditional analytical packages leave off. It allows you to see what users do on an individual page. It record user sessions anonymously showing you what they click on, hover over and how far down the page they scroll.

Although a technology like Click Tales is impressive it cannot replace traditional usability testing. It does not provide you the opportunity to question the user. For example it will not explain why users abandon shopping carts?

When the website owners of ecommerce sites start examining their website statistics they are often horrified by the dropout rate experienced on shopping cart pages. They worry that there is a fundamental usability flaw. However, in many cases that is not true. Questioning users reveal they abandon baskets for a host of reasons ranging from ‘I was saving the items to buy later’ through to ‘I wanted to compare the price on another site’. Like the homepage, shopping carts will always have a high exit rate and no amount of statistical analysis can change that.

However statistical analysis will allow us to improve the content and products we provide on our sites.

There is a real benefit in understanding what users want from our sites. From what content they want to what products they will buy, understand users requirements allows you to mould the site to user needs.

Website statistics can help identify popular content but not in the way you might expect. Looking at the most visited pages will not provide answers.

Popular pages can be misleading for three reasons:

  • Pages can be visited by mistake
  • Page can be popular because their prominent
  • Pages can be popular because they are gateway pages to deeper content

The homepage is a good example of these problems. I have already explained that the homepage is visited my mistake from search engines. It is also a prominent page and used as a navigational tool for finding other content.

Looking at how long users spend on a page can help to weed out ‘false positives’. However ultimately this is a flawed approach and can only give a partial indication of the popular content on your site.

A better approach is to look at the search terms users entered into search engines to reach your site. Almost all website analytical packages provide this information and it helps define users priorities. However, this is only going to show content that already exists on your site. If a user entered a search term for content you do not have, your site would not have been returned. The user would never come to your site. What you really need is a way of identify content that you do not offer but users want.

This is possible by examining the phrases users enter into your own sites search engine.

Approximately half of your visitors will use internal search. Every time they use search they are telling you exactly what they want from your site in their own words. That is incredibly valuable.

You need access to these search terms and particularly the ones that return zero results. This is the area where you need to do work. Users are expressing an interest in a piece of information you do not have or your search engine does not recognise.

Once you have access to these search phrases, start tailoring content around them. If the content does not exist, add it where appropriate. If it does exist but is not being found, introduce the exact phrasing your users are searching for. Better to use the language of your users than sticking to internal jargon nobody understands.

And there you have it. Proof that website stats can be used for a lot more than just tracking marketing campaigns.