• Supercharge A/B Testing

Can Personality-Based Marketing Supercharge A/B Testing?

The three most important questions that any company has to ask themselves are: what is the problem we’re solving, who is it a problem for and what is our proposed solution? And these questions are not asked just once, but continuously over the life of a company. As we build a new version of our platform with native integration for marketing tools including HubSpot and MailChimp, these questions have again become particularly pertinent. We passionately believe – and now know –  that to increase customer engagement, loyalty and retention, it is necessary to understand the personality of customers and personalise content accordingly. Here’s our case for why.

The problem

A/B testing offers little insight for the future
Testing. In the dark. All alone

A/B and multivariate testing are great optimisation tools, but on their own, can’t offer much insight that’s repeatable for future campaigns. Let’s go through the different stages of A/B testing.

You’re launching your first experiment. You’re staring at the screen thinking “should I test this image? Or that image? No, wait – this one!”. It’s a scene many marketers will be familiar with, but unfortunately, A/B and multivariate testing can’t offer any guidance.

After overcoming this initial decision paralysis to create the different variations, more questions surface: is there a more compelling variation you’ve missed? Will your results be unreliable? As you can only test so many variations against so many data points, these always remain possibilities with A/B and multivariate testing.

With the experiment over, it’s time to take a look at the results. You can see that your audience prefers a short subject line over a longer one, blue buttons over orange, an image of a landscape over a group of friends. But why is this the case? And why do some customers find this variation more compelling and not others? Without this information, future tests can’t fully benefit from previous results, and improvements are limited to what works best for the majority. But again, A/B testing and multivariate testing can only look at you with a blank stare.

You now want to apply the learnings of your first experiment to the next. But did you interpret the results correctly? If you go away on vacation, will your colleagues have the same knowledge of what works and what doesn’t and be able to run campaigns effectively in your absence?

The A/B Tester’s Dilemma

  1. How can I decide which variations to test?
  2. How can I make sure I haven’t missed anything?
  3. How can I understand why something is more or less compelling?
  4. How can I make content more compelling for everyone rather than just the majority?
  5. How can I apply learnings in an automated manner?

Who has this problem?

At a high level, marketers and customer relationship managers on a mission to boost customer engagement, loyalty and retention. Our new ‘Free’ and ‘Premium’ versions of our platform – which integrate with HubSpot and MailChimp – open up the tech that has delivered fantastic results for the likes of Hello bank! Belgium and Tinkoff Bank to companies of all shapes and sizes.

Proposed solution

Understanding customers as individuals

A platform that uses AI to predict customer personality from first-party data and allows content to be personalised accordingly, at scale.

My personality underpins how I think, feel and act – including how I respond to different images, words, and colours. Not only does this mean that matching content to my personality can increase appeal but that my personality can be predicted from the decisions I make – such as clicking on a particular link or image – which is precisely what DataSine’s algorithms are doing.

Here’s how it works:

Our platform plugs into your CRM or marketing system. Customer personality is predicted based their engagement with past campaigns (plus, optionally, an analysis of other information you’ve collected from customers, such as transactional data). By using first-party data we overcome some of the quality, ethical and regulatory issues that can come with third-party data (e.g. from social media). Next, the platform recommends how different elements – images, words, colours etc. – could be made more compelling for different personalities.

The algorithms that predict personality and content appeal are trained on millions of data points and improve in accuracy over time. The machine learning underpinnings mean that one user benefits from the learnings of all, in addition to the findings of research conducted by ourselves and prominent psychologists such as Dr Sandra Matz and Dr Chris Soto.

The A/B Tester’s Dilemma Revisited

Through the personality-based approach outlined above, we now have:

  1. A starting point for which variations to test (recommendations of images and words that have been proven to increase appeal)
  2. Reduced uncertainty (appeal prediction algorithms are trained on millions of data points)
  3. An understanding of why something is compelling (e.g. ‘this image will appeal more to your extraverted customers because it’s bright with reddish hues and features people engaging in an active, outdoor activity’)
  4. An automated approach (all test results are automatically fed back into the algorithm and reflected in future recommendations)
  5. Content tailored to the individual  (individual differences in appeal come down to personality – if we can understand a customer’s personality and what appeals to that personality, we can achieve a much deeper level of personalisation)

Does it work?

In short, yes. Here is a selection of our results to date:

  • Hello bank! Belgium increased customer engagement by 80%
  • A leading French bank increased sales by 71%
  • A leading Russian bank increased the conversion rate of their credit card sign up page by 59%

The ‘Premium’ version of our platform is currently available for free. You can sign up to be a beta user here, and we’ll be giving out access over the coming weeks. This version of the platform integrates with HubSpot and MailChimp and provides details on the personality of your customers and customer base, reports on the appeal of your past email engagement and potential for improvement, and recommendations of images that can deliver greater appeal. See the full feature list here.