Website personalisation and contextual targeting represent two significant opportunities for companies looking to create more relevant, targeted website experiences. Using technologies like VWO and Google Optimise, marketers can now personalise website experiences based on a huge array of user and session-level data points.
In this post, I discuss how content personalisation can be used to improve UX and strengthen your Conversion Rate Optisation strategy, while also walking through how to set up a personalisation campaign in VWO. First, let's start with the the role of segmentation in web analytics, and its importance in relation to content personalisation.
Why segmentation is criticial to CRO
It was Avinash Kaushik who rightly said, “all data in aggregate is essentially crap”. Aggregated website data hides so much nuance; how visitors from different marketing channels respond to content, the influence of different devices on website behaviour, differences in user intent, and so on.
Segmentation in web analytics allows us to dig deeper into each of these layers to develop a more detailed picture of the types of intent behind each of these contexts. Typically, CROs then use this data to create hypotheses about a site's many users, their preferences, perceived intent and propensity towards certain behaviours.
One size doesn't usually fit all
Often when we conduct A/B testing, we’re looking to establish an overall winner. However when a clear winner isn't established, we might think the test has failed, or that we were wrong in our hypothesis. While we may not have found the result we'd hoped, there's so much more insight under the surface waiting to be found, and in many respects this deeper level of understanding is what underpins website personalisation.
Digging into test data at channel, device, or even keyword level enables us to unearth insights and often competing narratives within the same dataset. Instead of viewing this as a failing of the test, it may be an opportunity to develop different versions of a given page, based on what the test outcome has told us about segment behaviours.
For instance, Paid Search traffic might respond more favourably to a product page when pricing and discount information is foregrounded than visits from a Referral site, and vice versa. Using these insights, we can create different versions of the same web page, tailored to the specific needs and user intent signals of each segment.
How to set up a personalisation campaign in VWO
Step 1: Start with A/B testing
Before jumping into personalisation proper, it’s important to identify the specific designs which perform most strongly within specific segments so that you have the available data points from which content personalisation can be based. Therefore, conducting A/B testing prior to personalisation is a key part of the process.
Using a combination of VWO and Google Analytics, we can slice and dice test data to get a more detailed picture of how different segments respond to different iterations of a page, then use this data to inform a content personalisation strategy.
In the following section, we take a look at an example of an A/B test, the goal of which was to identify which variation of a page would be more effective in funneling users into the sales process. In this example, neither version outperformed the other to a significant degree - with each converting at around 20%.
However, using the segmentation feature in the VWO interface, we can see that Organic Search traffic responded more favourably to the Variation:
Dig deeper with Google Analytics
By pushing VWO data to Google Analytics via Custom Dimensions (or Custom Variables depending on your implementation), we can really begin to slice and dice the test data to generate more thorough findings. In this case, I was was able to see from within VWO and Google Analytics that making a wholesale change to the website didn't make sense.
Not only would it have done little to improve overall page performance, but it would have reduced the conversion of other key segments. However, the test data shows that SEO traffic performs more effectively on the Variation. While the margin may look slim, VWO's Bayesian statistics engine reports that the Variation could convert up to 19% higher than the Control - something not to be sniffed at.
In instances like this, or where a more pronounced Conversion Rate difference is present within a specific segment, it makes sense to tailor a version of this page to be served only to that segment of visitors – thus improving both Conversion Rate and user experience for those visitors.
Step 2: Define the variation and campaign parameters
VWO has a host of features that make it much more than simply an A/B testing platform, and one of those is its ‘Personalistion’ feature. This feature allows you to serve different versions of a given page to different segments of users, and you can be incredibly detailed in your campaign setup, as alluded to in the chart below:
Setting up a personalisation campaign in VWO is actually very similar to setting up an A/B test. To set up a personalisation campaign in VWO follows these basic steps:
- Enter the URL of the page you want to test
- Use the visual editor to create the necessary page variations (informed by your previous A/B test outcome)
- Define the target segments for your variation (in this example we would want to Variation 1 to all Organic visitors using Mobile devices)
- Set up a Goal URL, or other parameter you want to track as a Goal (such as a form submission or button click)
- Name your campaign and enable heatmaps for more granular campaign analysis
- Launch the campaign!
Step 3: monitor the results of your campaign
Unlike A/B testing, there is no ‘goal’ or variant that we want to establish as a winner, as that’s already been taken care of through careful testing and analysis. However, as with any CRO campaign, we’ll want keep a close eye on campaign performance. Consider benchmarking the performance of your test page prior to implementing your personalisation campaign, or, setting the campaign to show the original ‘Control’ page to a small percentage of users so that you have a point of comparison.
As with all CRO work, the proof is in careful planning, testing, analysis and iteration. Website personalisation may be the next big thing, but it will only ever be effective if the right level of preparation and thought has been applied.