42 Degrees - Business from a different angle

Good analytics are vital to successful e-commerce yet some companies just don't bother. But if you listen carefully it's not difficult to hear what your website is trying to tell you

In the early years of the internet, e-commerce was mostly a volume game. The more traffic you generated on your website, the more money you made. End of story. Even if you wanted to, you couldn't find out much more detail about your customers – who they are, where they come from, what is working on your site and what isn't, because the tools for such analysis were both primitive and expensive.

These days, things are a little more complicated. It's no longer sufficient to have a website simply because it's new and exciting. Like any other area of business, you need a strategy.

That means you have to decide exactly what you want from your website. Do you want to build up a database of sales leads, communicate with existing clients, develop long-term relationships with customers, provide customer information or is it simply an online trading opportunity? Are you trying to help your retailers with more product information, are you trying to improve after sales care? The list of possibilities is huge.

But let's suppose you know exactly what your website is for. The next obvious question is, "How will we know when it has achieved its purpose?" One of the great attractions of the web as a marketing and sales channel is that everything can be measured: who is visiting your site, how long they are staying, which parts of your site they linger on, what they are spending, where they come from, even whether they go straight to your biggest competitor when they leave.

There are all sorts of software tools for monitoring whichever targets you have set yourself. Many of them are free and come included in your web-publishing package. AWStats, for example, tells you how many visitors you have had, how many web pages they viewed, how long they stayed and which country they were from. But these are not really designed for marketing analysis.

Microsoft has just launched an analytic service called Gatineau and Yahoo also has a free analytics tool. But industry consensus seems to be that the Google Analytics package provides the best introduction. It works by placing a computer code on your website that tracks your visitors and generates impressive reports.

"Google Analytics is excellent because it ties in with all the other programmes available," says Pam Vick, managing director of Caspia marketing consultancy. "That's why I and most of my clients use Google," she adds.

"It is certainly adequate," agrees Tim Leighton-Boyce, an independent web and e-commerce analyst. "You need to spend tens of thousands for the next level up, where you get sophisticated features such as integrating with information from your other systems."

It is relatively simple to use and allows you to define the basic information you need to evaluate your site's success. This includes number of visitors registered, how many went to your trading page but didn't buy anything and how many clicked on a particular page after a while. It should also provide reliable indicators of visitor numbers, geographical location and page views per visit.

It may be free, but it takes time and effort to get the best out of it, warns Leighton-Boyce. "With Google Analytics, the only investment, once you have set up the reporting system, is your time in looking at it," he says. However, for a variety of reasons many firms simply don't bother. "In my experience of working with different sized merchants, it is extremely rare to find companies who are spending real time working with this information. Many have not even got round to configuring the system so that the reports are easily readable on their sites.

"It's very common, for example, to find people who have not yet used quite basic features such as Google Analytics' ability to 'Exclude URL Query Parameters' to strip out redundant information, such as session identifiers, from their page addresses. It's a simple step but, without it, many of the reports might as well be in Greek.

"But the main investment in time has to be the ongoing analysis and taking action. It should be a never-ending process of improvement."

Key variables include 'pages per visit', which shows the extent to which the visitor invested time and effort in viewing your material and is a good indication of their level of interest in your company. The 'conversion rate' tells you the percentage of visits that result in a goal – a sale. Another very good one is 'average spend per visit' – that's not average spend per 'order'. This provides an accurate snapshot of the state of health of the site and promotional activity – a temperature gauge.

'Bounce rate' is the rate at which visitors come into your site and then leave it again immediately. But, in its raw state, this is not informative enough. To get the best out of your data you need to analyse it in as many relevant ways as possible.

Leighton-Boyce divides analysis of bounce rate by source: where has a visitor just come from before visiting your site (was it from your own ad campaign for example, or from a search engine)? That way you can monitor your promotional campaign and find out whether your products page is offering what visitors are looking for.

But he also recommends looking at the type of keyword used in the search – was it a generic brand or industry term or was it product specific? You should also look at the landing page (where visitors enter your site). Was it your home page, a product page or a deep information page?

"Each of these can suggest people with different purposes in mind or at different stages in their relationship with your company. Only by splitting them up in some way can sensible patterns of behaviour start to emerge," says Leighton-Boyce.

He says he often sees sites with an enviable conversion rate when viewed overall. But segmenting the traffic reveals that these sites often have a large group of very loyal returning customers who would arrive at the home page from sources such as a brand search on Google, or a clickthrough from an email.

After some analysis, the true picture often turns out to be very different: "Focusing on the other traffic that had been searching for generic terms and then landed on a product page – some from unpaid search results and some from paid search adverts – revealed much higher bounce rates and much lower conversion rates than appeared at first glance."

But that shouldn't be the end of your analysis: the conversion rates for the individual product pages will also vary. Compare the bottom performers with the top performers and try to work out why, advises Leighton-Boyce. "It could be that the poor pages are fine and the issue is price – in which case, web analytics will be providing a useful clue about your position in the market," he explains. "But it's often the case that the poor pages are just that: they have much weaker sales copy, poor photography or something along those lines that is easy to spot when the analytics encourage you to compare them against a 'better' page."

With these tools, you can get an indication of whether the visitors coming to your site are using it in the way you intended. If they are not, then you can choose whether to modify your site, or – harder – attempt to persuade them to modify their behaviour.

It may sound complex – especially if you are not a mathematical type. But it is the sort of analysis you need to be carrying out if you want to be able to fine-tune your e-commerce operation to get the very best out of it.