When Team Sky launched as the British professional cycling team in 2010 it had a big goal; for a British rider to win the Tour de France within 5 years. It seemed a distant dream; no Brit had ever won before. But coach Dave Brailsford had a new approach, a theory he called the ‘aggregation of marginal gains’.
Brailsford was convinced that instead of attempting a major revolution in how the team trained, the key to success was making tiny, marginal improvements in multiple areas, which would add up to a seismic shift in their fortunes. Every aspect of training, from nutrition and saddles to pillows and massage gels were assessed to see if improvements of just 1% could be made in performance.
Team Sky didn’t win the Tour de France in five years, as Brailsford had predicted. They won it in three.
The theory of marginal gains will sound familiar to anyone working with website optimisation. Tweak a button colour here, change a call to action there, make tiny improvements to a website, and your conversion numbers improve. But are you harnessing the full power of marginal aggregation? Could your approach be more systematic, more effective? Here are some ideas on how to bring about a Team Sky-style revolution on your ecommerce site.
Large-scale ecommerce website redesigns generally get results. If you base your redesign on solid user-centred research and design, your conversion rates will usually increase.
But will that improvement last? The trouble is, your customers and their expectations will change over time. Users spend most of their time on other sites, and they would prefer that your site works in the same way as all the rest. As all the other sites aren’t redesigning at the same time as you, that means there’s always a growing gap between your site and what users want. Those changing expectations are likely to lead to a gradual decline in conversion rates.
The key to combatting that decline is to continuously optimise your site in between those big redesigns, making marginal improvements to maintain or even improve performance.
But that optimisation has to be informed and systematic. Dave Brailsford didn’t just firefight issues as they came up or make random improvements – he actively sought areas that could be tweaked in a systematic process. So how do you develop such a process for an ecommerce site?
The first thing you need to know with ecommerce redesigns is how you are going to measure the impact of any changes you make. Your measurement needs to be more nuanced than simply a site traffic tally.
Some ecommerce managers invest heavily in SEO to attract traffic, trusting in the fact that certain search terms are popular as an indication of demand. But although getting people on to your site is crucial, there’s no point investing in boosting traffic if they don’t buy anything or buy too little once they land. Fail to deliver value to those visitors and you’ll have high traffic but a terrible conversion rate. What’s more, they’ll be disappointed and may not come back.
Instead of traffic, your key metrics when you measure changes should be:
The percentage of people who purchase from you divided by your total traffic.
The average amount someone spends when they order from you.
The amount of money generated when someone visits your website.
We specialise in user centred design for ecommerce and software products.Find out more
The next part of designing a systematic approach to marginal improvements is to establish robust methods of testing your results. A/B or multivariate testing is the most effective way to understand how certain changes to your design or copy impact your KPIs.
But before you make any changes or test what effect they have, you need a hypothesis – an idea of what will happen when you make a change. For example,
We believe that changing the text on our ‘Confirm and pay’ button to ‘Buy now’ will increase conversion by 1%. We’ll know this is true when conversion between the order summary and confirmation screens increases.
A hypothesis like this gives you something tangible to test. You could make the change without a hypothesis, sure, but how will you know what to test and how to measure success? Running an experiment without a hypothesis is like taking a road trip just for the sake of driving, without thinking about where you’re headed or why. Sure, you’ll end up somewhere, but there’s a chance you might not have gained anything from the experience.
A hypothesis can follow this simple format:
We believe that changing [thing] for [audience] will result in [measurable change]. We’ll know this is true when we see [KPI move].
But split testing is never the full story. Even if your split testing gives you a clear answer on whether a change improves conversion, you still need to find out why it does that.
If you don’t understand why you’re potentially missing out on a breakthrough that could help you deliver even stronger results.
Finding out the why can be done through regular user testing to help you understand how your users use and perceive your website. (Natural Interaction can help with this – we’re experts at finding out how people perceive design changes on ecommerce sites.) This on-going process can even identify other areas where people struggle or misunderstand your design and feed your next round of optimisation.
Creating a robust system for making marginal gains is an on-going task. Split tests should be carried out monthly and monitored daily, and your entire site user tested quarterly. It might seem like a lot of work, but the net effect will be an overall improvement in conversion rates that doesn’t slip back between overhauls.
It can make sense to offload this testing process onto an external team; at Natural Interaction we’re old hands at running ecommerce website redesigns and on-going optimisation programmes and can take the headache out of all that hypothesising, testing and monitoring.
In a highly competitive marketplace, those tiny marginal gains in your ecommerce website redesigns might well be the secret to out-manoeuvring your competition and keeping those conversion numbers ticking in the right direction.