Say the words ‘site re-design’ and if you listen really, really hard you can probably hear a collective shudder from IT and marketing teams around the world. Anyone who has ever been involved in a site re-design will know it’s a huge project and requires coordination of a number of parties involved. With resources constantly under pressure, testing can often get forgotten. However, this can actually cause more problems in the long run!
Ultimately, without testing, a company is potentially sailing into choppy waters. For a start, a total redesign has a much higher probability of not performing as well as the original version – you never know how a customer is going to react to change. Moreover, when you’re in re-design mode, it is easy to become locked into a certain idea and theme, and start to lose the wood for the trees. A more complex design, even if on just a single page, can go wrong in many more ways than making one or two simple changes.
Many big names have redesigned without testing and lived to regret it. The story goes that 5 years ago a major UK pharmacy chain underwent a major site redesign without the use of optimisation solutions, and the result was a significant decrease in online performance. The website launch release ended up being rolled back with, no doubt, significant costs to the company due to additional work the team had to put into understanding what the issues were, re-redesigning, and then implementing… in other words, a complete headache!
Mitigation of risk, then, is the number one reason for optimisation before, during, and after site re-designs. Given potential risks of site re-designs coupled with today’s increasingly competitive online markets (not to mention the risks for a purely online retailer), optimisation can be key to a website’s – and company’s – success.
So now I’ve convinced you why it’s important, how can you implement testing without causing too much extra stress? Firstly, it’s important to have experts in the field to provide guidance in testing by digging into the detail of how best to generate learning and value, and prioritise accordingly. Impact potential, degree of difficulty, development effort, and area of the page are all features which should be considered during optimisation efforts. The type of testing deployed will be dependent on a company-to-company basis, but the following stages can be useful to consider:
- Using A/B and MVT to feed ideas into site redesigns
- Using A/B test old vs. new
- Releasing a new site or part of a site to a % of visitors to gauge performance, and ramp up
- Using A/B and MVT to drive performance and conversion on the new site as soon as it is live
The above steps are relatively simple to implement in real life, and can be combined with each other to produce optimum results. I’ve outlined below some handy tips and examples for each approach to show how it can be done:
Option 1 – A/B testing allows you to test multiple homepages to identify the most favourable version. It was something Urban Outfitters chose to apply when with Webtrends performed when they rolled out a re-skin of their website in 2011. By feeding ideas through these testing strategies it can show if you’re on the right track.
Option 2 – using A/B testing to compare an old and new version of a site – is an overarching strategy which could be followed by an MVT enhancement on the winning version. This is the best method to understand the impact of what you are going to implement, as the pages are likely to be radically different. Of course, you may not wish to go back to the drawing board in the event that the Control version of the page actually continues to be the top performer. In this scenario, you could have an MVT planned and ready to go for the redesigned version as soon as you decide to deploy this to your production environment. If you go ahead with this approach, you are able to mitigate the risk of just delivering one version of the page. Webtrends has practised this with Moneysupermarket, a company with a number of different channels, including travel and insurance. In each channel, we tested old against new, starting at around 5% and slowly increasing in order to reduce all possible risk…which leads us on to option 3.
When using option 3 – releasing a new site or part of a site to a percentage of visitors to gauge performance – it’s likely that not all pages will need to be tested before the re-design goes live. There are, however, a number of key pages that need to be considered, such as the Homepage, Booking Page, Search Results, etc. A/B testing would allow you to test the effectiveness of one funnel against another, or test multiple homepages to identify the most favourable version. We used this approach with Barclays and Holiday Autos (part of Lastminute.com) when helping with the redesigns of their homepages. If you can isolate individual learnings on an individual page, and compile them, you can regularly feed these lessons into projects to show where redesigns have gone astray.
Companies can get carried away with re-design and forget to test their sites. Sometimes it can seem too late for optimisation – a company might be in the process of re-designing its website, realise it doesn’t work, but not have the time or resources to overhaul the new edition. Yet, even then, testing can help by trying to control some of the new elements, or by testing new versions of key pages as soon as a new version is out (option 4).
The takeaway is that testing shouldn’t be seen as separate to re-designs – they should work in tandem to produce the best possible website and, as a result, business success. With huge resources poured into site re-design projects, testing can help to ensure the project is on the right track and the end result will deliver enhanced site performance and, finally, conversions!
Ultimately, without testing, one is essentially making the choice to take a considerable amount of risk.
