Businesses are more and more constrained on their marketing budgets, or at least facing more pressure to justify where their budget is spent, and quite right too.
In my previous post ‘Racing Car Setup Lessons For Marketing Effectiveness‘, I looked at how we need to make changes little and often. However the next step has to be about ROI. So where do we start and how do we prioritise these changes. In this post I will look at the prioritisation of UX changes and how we can determine which one is having the best effect on the business from a commercial standpoint.
Auditing The Customer Journey
The place we typically start is understanding where the customer journey can be optimised. A user experience (UX) team will look at the current journey on the site and identify what areas could be improved. This uncovers a number of potential improvements throughout the site. If we looked at a standard digital purchasing journey (below) for a consumer ecommerce business, the identified recommendations could be aligned to any part of the customer journey below.
Understanding The Potential
Once these potential improvements have been identified – and there can be many – it is then necessary to quantify the value of these changes to help us prioritise which ones we will focus on first. Firstly we need to go back to the funnel and link the recommended UX changes to the funnel. Once we have done this we need to look at the web analytics data to understand the basics:
- The volumes at each stage of the funnel – to identify where the biggest drop off occurs (and overall conversion rates)
- The average order value – to identify the current commercial value of each order
From here we can then look at each stage of the funnel to work out the commercial opportunity. In the funnel below we can see the drop out rate of each stage of the funnel. Assuming the funnel behaviour doesn’t change downstream we can isolate an individual part of the funnel based on the UX recommendations to see how much of an effect any improvement we make would have on total revenue.
So in the example above we can see the major drop out occurs towards the top end of the funnel. However, to truly understand the potential impact we really need to understand the likelihood of customers converting in that part of the funnel. If we only focus on driving more customers to the top of the funnel it does not necessarily mean we will get more sales, especially if there is a high customer bounce rate (those customers not visiting more than one page on the website).
Creating The Benchmark
If we looked at the original funnel and asked ourselves, “what if we got an extra 200 customers to each stage of the funnel?” we could identify the sales effect this would have on the funnel overall. Using an average order value of €50 we can see where the opportunity lies.
As a result we can clearly see that, in the funnel above, the real opportunity is in the third stage of the funnel. We have a five-fold improvement in this stage of the funnel by getting 200 more customers to “add something to their basket”.
There is certainly a bigger revenue opportunity towards the end of the funnel but the incremental improvement is significantly higher in “Added to Basket” stage than in any other.
Furthermore adding 200 visitors to the “Added to Basket” stage is only a 20% improvement whereas at the bottom of the funnel “Complete” it is a 100% or doubling of the number of customers, which is going to be more challenging.
Now, this is only a way of highlighting the potential, rather than guaranteeing it.
So we now have our first test to optimise sales performance on our website.
So where to next?
The final part of the process is to test the changes:
We need to make sure we have our benchmark; in this example we used the existing funnel performance.
Identify the changes we are going to make, but be mindful that changing too many things will mean we will not know what has had an effect. As a rule of thumb change only one item in each stage of the funnel.
Run the test on a sample of the audience ensuring some customers receive the old experience.
Execute the test for long enough to be certain there have been enough people who have gone through the new vs. old funnel to be sure of a conclusion (the volume needs to be statistically valid).