Yet some designers use this idea to avoid engaging with analytics data completely. It’s understandable – many bright people are uncomfortable with numbers – but it’s a shame, as analytics offers many insights to UX design.
Analytics Improves The Research Phase
Data provides valuable context for your research documentation. Adding pageviews to a content audit shows which areas are most popular. Plotting visit share on an information architecture shows whether it’s working. Recording drop-offs in a registration flow shows where users are having problems.
It’s not all numbers either – qualitative data is a great starting point for user research. Find words and phrases signalling intent in search logs. See which tasks people are doing, and how your site frustrates them, with the 4Q survey.
Analytics data also lets you quickly test your research conclusions. At CABE, user interviews suggested that people thought in subjects (‘housing’) rather than formats (‘policy papers’). We confirmed this with analytics. Being able to validate a single user’s opinion against a large dataset is very powerful.
Data Helps With Prioritisation
Whether you are putting together a wireframe or a prototype you need to make tough decisions to create a great visual hierarchy. Analytics can help. Historical data shows which interface elements people use most and current data shows how people interact with the live design.
At a larger scale, analytics can help you choose which design projects to actually do by quantifying their potential impact. Without this you risk wasting precious resources on projects that only benefit a tiny percentage of users.
Of course numbers never tell the whole story – there are still business goals, strategic aims and brand requirements – but they should always be in the mix.
Testing Reduces Risk
A/B and multivariate testing help you make design decisions because they take the risk out of getting it wrong. Instead of prevaricating for days over the copy or placement of elements you can make a best guess and test the alternatives.
Testing is great for building consensus too. Gather suggestions from colleagues in workshops or even placate the local HiPPO by including their preferred version. People invest more in the design process once they feel part of it, but prepare yourself for when their version converts better than yours!
Analytics Speaks The Language Of Business
Business people feel more comfortable with numbers than opinions, so using analytics is a great way to sell ideas to the people who fund your projects.
A proposal based on increasing unique visitors is more convincing than one based on intuition alone. Better still would be to quantify the change in revenue gained or expenditure saved by calculating how much the extra visitors are worth (assign a goal value, see how many visitors reach the goal and multiply).
Yes, it’s tricky to make accurate predictions but you get better with practice. And when you deliver the promised increases, people start to trust your design work regardless of whether you quantify the impact upfront.
Getting Started With Web Analytics
- Read Avinash Kaushik
Avinash is the god of web analytics because he focuses on the difference between analysis (actionable insights) and reporting (‘puking’ data). Read his excellent blog and devour both of his books.
- Try out some tools
Play with free tools like Google Analytics (traffic data), 4Q survey (qualitative feedback), Crazy Egg (heatmaps) and Google Website Optimizer (A/B testing) to get a sense of how they work.
- Focus on what matters
In particular, conversions/goals/funnels (how many people do things you want), audience segmentation (splitting data based on characteristics) and event tracking (logging clicks on parts of the interface).
- Avoid the red herrings
‘Time on site’ is a not measure of engagement. ‘Visits’ or ‘pageviews’ on their own are bad metrics (traffic costs money – you want conversions). Path analysis rarely leads to useful insights. Never use the term ‘hits’. Ever.
- Communicate the data properly
Insight is useless if no one ‘gets’ it. Read Show Me The Numbers by Steven Few (practical advice), Visual Display of Quantitative Information (theoretical classic) and Information Is Beautiful by David McCandless (inspiration).
Web analytics isn’t an answer to every question and it’s never a replacement for talking to your users directly. In particular, data-driven design may never be able to improve your site beyond a theoretical local maximum.
But if you’re avoiding web analytics because it looks too complicated, chances are that you’re losing out on some great design insights.
Let me know what you think about this – particularly if I’ve missed any obvious uses or if you disagree with anything – on @myddelton. And big thanks to Katarzyna Stawarz for asking the question that inspired this post…