Eskom on Monday evening took the unprecedented step of implementing Stage 6 load shedding. In an announcement on Twitter, that almost immediately received widespread…
User testing (UX) is a fundamental component of the web development process. It is conducted as a matter of course in the US, UK and other developed economies during the development of any large website. In emerging markets, the entire field of Usability is still relatively small, and user testing is infrequently conducted. It doesn’t have to be like this, and it shouldn’t be.
Let’s begin by taking a look at the kind of situations that warrant user testing, and then some ways of going about it.
Why test? To Plan
User testing should first be employed in the planning phase of a new website. Every website is destined to be used by a certain target market, and it is important to learn about this target market.
Factors such as
- the familiarity the users have with web interfaces
- the trust they have in ecommerce
- the average internet connection speed they have access to
- common sites and services they consume regularly
all have an impact on the way your website is developed.
In addition to this kind of information is the rich mine of psychographic data that can be discovered about your potential ‘average’ users which should serve as an invaluable guide in design questions ranging from colour, font and general look and feel, to more technical aspects of functionality.
Why test? To Iterate
At further stages in the development process, when there are functional versions of the website available to play with, there is immense value to be had in further testing. Before embarking on the arduous task of putting the finishing touches to the frontend design of a site, it can be an extremely wise investment to conduct tests, and run through some improved iterations of the functionality and general layout of the site.
If usability potholes are uncovered at this stage of the process, it saves an immeasurable amount of work down the road.
Why test? To Optimise
Your website may well be excellent, but there are very few instances where no improvements can be made. Learning more about the way users interact with your website can help you understand why some of them leave without doing what you want them to do. In turn, having this kind of knowledge can inform what kind of optimisations should be made.
Growth in conversion rate from 4% to 5% due to a minor optimisation may seem fairly small, but it is a 25% increase in conversion rate, and hopefully sales and profit too – an incredible return on investment for what can be an affordable and easy process.
How to test? Qualitative methods
User tests can be categorised in many ways, one of which is into qualitative and quantitative methods. Qualitative methods focus on fewer users but gather richer data about them. Often these kinds of user tests are employed in the planning and early development phases of a website. Jakob Nielsen has discovered through his research that one needs no more than five users to discover almost all the information one can gather from user tests of this nature.
The utility provided by further users is marginal in comparison to the costs of involving them. This is great news – performing tests with five people is usually very affordable and accessible especially in the face of the benefits it can provide.
Qualitative user tests can range from card sorting exercises, in which product, webpage or information types are written onto cards and test subjects arrange them into categories that make the most sense, to stakeholder interviews and task oriented tests.
Task oriented tests involve asking a user to achieve some task on a version of your site, while speaking aloud any thoughts they are having whilst trying to perform the task. These sessions can be captured on video (both the subject and their actions on screen), and provide an in-depth perspective on ways to improve the user experience of a site.
How to test? Quantitative methods
There are some amazing tools available for gathering large amounts of rich data about the way users interact with a website, and for testing different aspects of the site to help the optimisation process.
Google Analytics (GA) is not often mentioned in lists of usability tools, but it can be one of the most powerful (free) tools if used correctly. A thorough GA implementation can provide visualisations of the conversion funnel, with the drop off rates at various points, even in such granular detail as users’ progression through a contact form on a site.
Other analytics applications include click mapping tools such as Crazy Egg, to provide graphical representations of the areas on a webpage users click the most, and remote feedback platforms such as Feedback Army which facilitates the gathering of small amounts of feedback from a large number of users about particular features of a site.
A/B and Multivariate testing involves serving different versions of a webpage to different users. Given enough traffic and enough time the difference in user behavior on the different versions of the webpage can become statistically significant, and significantly valuable. An example of an A/B test might be serving two versions of a product page on an ecommerce site, one with user reviews of the product, and one without.
The pages are served randomly and in equal proportion to the users, and eventually it might be possible to see that conversion rates are twice as high on the page with the user reviews, for example. Being able to test like this in a live environment with real users (without them knowing) can provide insights that are simply impossible to get offline.
To test or not to test?
An investment in user testing, if managed properly and by people who know what they are doing, will return value far beyond its cost. It may be in the form of hundreds of saved development and design hours, or improvements in conversion rates, and subsequent increases in revenue and profit.
Sophisticated and extensive testing can be conducted by usability and UX professionals, or, if budgets are low (or non-existent), there’s nothing wrong with getting granny and your friend with the severe caffeine habit around and running some task oriented user test of your own –- you might be surprised at how much you learn.