Are you running upper-funnel A/B tests but not seeing a sales lift in your test results? Applying an attribution model may be your answer.
I hear this often and have experienced it myself. In fact, it’s rare to run an A/B test on your websites home, campaign or product pages and see a direct correlation in increased sales. You’re likely to see a lift in engagement measures such as clicks, bounces, and progression rates. However, it’s usually the tests further down the funnel that show statistical improvement on conversion rates.
Upper-Funnel Testing is Important
It’s important that your experiments at the top of your funnel are getting the kudos they deserve. These pages are key to your website’s success and are often used to influence prospects to move further down the funnel.
If your websites’ upper-funnel pages don’t address your visitors needs, then your bounce rate will likely be high and any advertising spend will have a low return on investment (ROI), wasting you precious budget. The most effective way to increase your advertising ROI is to optimise your landing pages, aligning the user experience to your audiences needs.
Additionally, you can focus your efforts lower-funnel experiments, but if your visitors aren’t progressing past your landing pages then your efforts are wasted.
Therefore, when experiments are most commonly prioritised based on revenue impact, we need to look at new ways to assign value to these upper-funnel tests. This is where CRO attribution models are proving to be valuable.
What is a CRO Attribution Model?
Attribution models have become common practice for marketers, where last click attribution doesn’t show the true value of each touchpoint. This practice has now started to move into CRO, with a few leaders now applying attribution models to their optimisation programs. This ensures experiments run throughout your funnel are prioritised based on their attributed value.
Firstly, you need to understand what a percentage increase to your conversion rate is worth to your business. In the most basic form you can multiply your website revenue by 1%. For example, if your websites annual revenue was $1 million, then a 1% lift would attribute $10,000 additional revenue.
Secondly, consistent measurement of your experiments is critical. It’s recommended to not only measure the progression rate (the percent of audience that progresses to the next step in your funnel) of an experiment, but also ensure that the progression rate of the remainder of your funnel remains consistent (i.e. no change to subsequent steps). If this is true for your experiment, then you can attribute the correlating revenue to the progression lift in your results. This CRO attribution model can also be used to estimate and prioritise your experiment backlog.
This is the most simple attribution model. There are many factors you can introduce to refine your experiments attributed value. If you’re looking at ways to apply this directly to your business then get in touch with one of our optimisation experts.
Words by Tracey Reed.