4 ways machine learning can enhance your email marketing
Email marketers have been successfully using multivariate and A/B testing for years, but it’s a manual, time-consuming slog.
A classic customer relationship management (CRM) dataset for example, can hold millions of objects. This personal data is broken down into numerous fields covering name, age, gender, buying habits, email addresses and so on. To test a variety of email subject lines against all this data ‘by hand’ is a major marketing challenge.
The beauty of machine learning (ML) is that it has the potential to do all this and much more with much larger datasets, in far less time, with far better results.
Machine learning developments are potentially overshadowed by the fast-approaching General Data Protection Regulation (GDPR), coming into force 25th May 2018 for businesses operating inside the European Union, but more on that later.
Here are four ways your email marketing can benefit from machine learning.
Predict the best images for emails
The right image says a thousand words – but trying to figure out which picture or photo to include in your marketing email often comes down to gut instinct or creative decision. Fortunately, ML advancements are taking the guesswork out of this.
Adobe is using its artificial intelligence (AI) product Adobe Sensei to suggest relevant images for emails, and calculate an image’s conversion and engagement potential by analysing huge swathes of data to see how customers have responded to similar visuals in the past (it has an asset library of three million plus to call upon!)
Subject line and body copy email optimisation
Crafting the perfect email subject lines, body copy and calls to action is an ongoing struggle for many marketers – it’s hard to generate statistically significant results and multivariate test hundreds of options while maintaining correct tone of voice. Now, thanks to ML powered platforms, marketers can let the machines work out what resonates best with a specific target audience. This can even be done with a virtual audience to reduce the risk of real world hiccups eroding your database.
We all know the importance of good copy. Get it wrong, and your email ends up in the trash. Get it right, and your email is opened, read and hopefully – generates positive action. Leaders in email AI are using natural language technology to create human-sounding, on-message body copy and calls-to-action too. Many of these platforms claim to dramatically outperform content written by humans, successfully optimising copy that incites positive interest and action.
Predict customer churn
Adobe Sensei is also being used to analyse changes in customers’ online engagement such as a decrease in email opens or use of online services. This will help brands identify customers who are likely to churn in the near future.
Armed with this insight, marketers can step in and re-engage with these customers before losing them.
Optimise delivery time
Optimised delivery time is not a new concept. But as with a lot of the email marketing developments, the more data you can use in the process, the better the result. Essentially, email delivery times are modified according to when recipients are most likely to engage with their messages. Machine learning enables marketers to segment by finely tuned personas and run tests on huge datasets, producing statistically significant results.
So far, so great but where to next?
Yes, the application of machine learning advancements in email marketing has improved how marketers craft and send emails – but an optimised subject line is still just a subject line. Real wins come from data pulled from video views, social activity and other such unstructured data. Manually created data models can’t handle this stuff – but AI can analyse it and pick patterns.
However, the most important question a CMO needs to ask right now is simply: is my data actionable? Marketers have a tendency to try and store every single data point they can in ‘data lakes’. This is where the GDPR is going to trip them up. In fact, GDPR is going to make all direct marketing activities – not just email marketing – harder, and will certainly hinder ML and AI innovation.
If data can’t be actioned then it needs to be deleted. Before getting over-excited by AI marketers should focus on making small data improvements. In other words, they need to pay increased attention to the data they can use – rather than the data they want to use.
Dirk Wybe de Jong is VP of Digital Marketing at Celerity, a data marketing agency.