Ecommerce optimisation tips part 3: Combining web analytics with user research

This is the third of a series of six posts where I’m uncovering fundamental UX principles for designing great ecommerce user experiences.

Last week we discussed the importance of designing for desirability and how to avoid feature creep.


This week we focus on the importance of marrying web analytics with usability testing.

Understanding both the 'what' and the 'why'

During the last couple of years the proliferation of analytics tools within the marketplace means that website owners are getting access to masses of data they can use to inform their design decisions.

We’ve seen how our clients find it hard to make decisions from this type of data, and that many marketeers are literally drowning in the stuff and don’t know where to turn to interpret it.

The problem with analytics tools is that they cannot tell you why the recorded behaviour is happening.

The best web analytics strategies consider how to understand the 'what' and then how to explain what is going on by understanding the 'why'.

The analytics tools clearly give you the 'what' and you can start to understand the 'why' by testing pages or groups of pages that are causing problems.

This approach makes sense because it means that you target your UX budget at the areas of your site that need to be optimised.

By combining tools such as Google Analytics with guerrilla usability testing techniques, you can begin to get a clear idea of where your site is failing and why, for relatively little cost.

Usability testing can be a quick way of uncovering how to improve your site

It then becomes easier to justify spend on your website because you can be reassured that it is being diverted to the areas of the site that need it the most, and also that will yield the greatest ROI.

Don't just focus on conversion

When viewing analytics data it is all too easy to focus on conversion when in reality that is not the only task that users will be visiting your site to fulfill.

Analytics data means a lot more when it's viewed in context with user behviour

Some purchases will require significant research and so multiple short visits from a user may be a very positive thing that could, of course, lead to a purchase.

This research behaviour, when viewed within an analytics tool with no context, may cause concern unnecessarily whereas the reality is that the site is working brilliantly because it is actually supporting users buying behaviour.

It is important to understand the reasons why people are visiting your site and to then consider your analytics data with these in mind.

We are being asked to do more and more work around task modeling so that we can understand how people really behave, both online and offline, when making purchases or planning holidays.

The decisions that you make based on your analytics will be better informed if they are grounded with an understanding of what it is your users are trying to do. When we have created a task model, our clients can work with us to better understand how they can make their site function more optimally.

By spending time with users you can understand the reasons they are both visiting and leaving the site that will help to rationalise the behaviour that you are seeing within your analytics.

Use site search to learn more

At the beginning of new projects I'm always keen to get my hands on search logs as well as analytics, to get a feel for what people are really looking for on my clients' sites.

This data analysis can uncover all manner of gems. It tells you not only what people are looking for, but teaches you about the language that users will use to try and find what they're looking for.

You can also use this data to see which searches are resulting in no results. This can help to identify content gaps and offers a pragmatic way of prioritising what content you should be providing your users with.

Armed with this knowledge you can enrich your sites' information architecture with these keywords which contributes to search engine optimisation as well as making it more intuitive for your users.

This data also can be used to inform your meta data and thus becomes an essential feed to nourish the development of any controlled vocabularies that may be being used.

The area of analytics and UX is a rapidly growing and fascinating area. I would recommend the following resources to get further insight into these topics.

Lou Rosenfeld

Lou is a key figure within the Information Architecture and UX domain. He has been focusing on search analytics and you can view his workshop slides on slideshare.

Avinash Kaushik
Avinash works for Google and is the author of 'Web analytics, an hour a day' which I would whole-heartedly recommend. Avinash talks about combining the 'what' with the 'why' within his blog Occam’s Razor.

James is responsible for leading user-centred design projects across all industry sectors, and also runs cxlabs. He has written two UX books, speaks regularly at international conferences, and co-founded UXBristol.