Can machine learning help you build killer email campaigns?

Artificial intelligence and machine learning are two of today’s hottest buzzwords. We’re quite sure most of our readers have picked up the terms in conversation and can name a few practical examples from the top of their minds. But did you know that machine learning is likely to become a force to be reckoned with in email marketing? First, let’s look at the difference between artificial intelligence and machine learning…

Artificial Intelligence or AI implies that machines are capable of executing ‘intelligent’ tasks. Which gives them almost-human characteristics. It also means that they haven’t been specifically programmed to do everything they can do, but to a certain extent are able to adapt themselves to different situations.

Machine learning or ML is a sub-branch of AI which enables us to build machines that can process huge streams of data and ‘learn’ from them, without human intervention. Google Translate (Google Neural Machine Translation) and Google Images are two striking examples of successful machine learning. Within Translate, the ‘machine’ gets to process a large amount of data which it tries to interconnect to make the returned translations more accurate along the way.

Machine learning also offers an abundance of opportunities for email marketeers, as a one-size-fits-all approach has become outdated

Now: optimise with common sense

A lot has been done to optimise email campaigns in order to make them perform better. From A/B testing and the analysis of open rates to the exact definition of the best moments to send out campaigns. A/B testing allows us to compare 2 different versions of the same email and test how well they perform with a target audience. The ‘winning’ version will eventually be sent out as the definitive email campaign.

Machine learning is often used in spam filters at the client-side, to automatically detect which emails could be unsolicited and which are not. Gmail also uses machine learning to determine in which category a certain email belongs and whether or not it is important to the addressee. Doing a test run of an email with different subject lines will undoubtedly better your chances when it comes to deliverability.

Future: automated optimisation by machine learning

Today, optimising emails can be a tedious and very time-consuming job. However, the future looks bright, as machine learning is ready to take over some of the hard work.

ML will not only help us get better results from our email campaigns. It will also save us time and money, enabling marketers to focus on more important tasks. The analysis of data and results, for instance, will take considerably less time and effort.

Wouldn’t it be wonderful if, for every individual contact, we would be able to define the best moment to have them click through, look for information of buy our stuff? In the not so very distant future, machine learning could take over some of the hard labour.

Some examples

  • Subject Line Optimisation

You only get one chance to make a first impression. When the subject line and the recipient’s name of your email haven’t been optimised, you could be heading for failure. ML can help you generate and construct the perfect subject line. A careful analysis of previous emails and some A/B testing can lead you to subject lines that work like a charm.

  • Send Time Optimisation

Setting up an individual send schedule for every contact in your database is sheer impossible, you say? Machine learning to the rescue!  Know that some contacts like to plow through their mailbox during breakfast. Others use their breaks at work to catch up on their mails. Based on the data at hand, the system is able to predict the best moment to communicate with every contact. You’ll see your open rates and conversion rise significantly.

  • Copy Optimisation

We usually call in copywriters to write the perfect content for our email campaigns. But is that bound to change in the future? Tools like Persado and Phrasee AI already use machine learning to compile text, based on the tone of voice of your brand and the type of copy that works best with specific target groups.

These tools tend to become more accurate with each campaign. When a campaign closes, the tool will analyse the data, to tweak the copy of upcoming campaigns in order to get better results. According to the Pharsee website, 98% of their ML-generated content performs better than content created by the human brain.

  • Flow Automation Optimisation

In email marketing, machine learning would be capable to automatically adapt existing flows based on available data, which applies to the sequence of the emails as well as copy and content.

Do keep in mind that machine learning is still under development. In combination with artificial intelligence it might be able to put together, send and optimise email campaigns without human intervention in the future.

What are the main advantages?

Machine learning allows personalisation on a larger scale. It gives you better overall insights and determines the weight of the data, which automatically generates more return on your marketing efforts.

Emails that are tailored specifically to your target groups, will undoubtedly solidify people’s loyalty for your newsletters and emails. And more loyalty means that contacts will have more interaction with your brand or company, which heightens the chance of a purchase considerably.

And what about disadvantages?

Needless to say, it isn’t all roses. Today, machine learning is far from error-free, and there are quite a few ethical issues as well. It will take some more time and effort before we can fully harvest the fruit of the technology. Especially when the quality of your data hasn’t been optimised, machine learning will not produce the best of results.

And what about the dark side of machine learning? In the wrong hands, it could become a powerful tool to generate deliberately erroneous and damaging content or fake news.


Machine learning, when combined with artificial intelligence, does have an enormous potential in (email) marketing. But since the technology is still in its infancy, some caution is recommended. It’s best to keep the disadvantages in mind and use machine learning with an ethical mindset.

New call-to-action