Where AI Can Make A Difference In Email

There is a lot of buzz in the industry about artificial intelligence (AI). Every week email technology companies make an announcement about a new, revolutionary application of AI to email marketing. 

When asked  where they might apply AI to email marketing, most marketers I talk to focus on a small handful of applications: optimizing send time, personalizing offers, and optimizing subject lines. 

Having spent parts of the last five years working on applying AI to email, I believe that there are many more areas where AI can be productively applied in email. I’ve listed just a few here.

Acquiring and maintaining a better subscriber base. As I’ve said in previous columns, about half the the performance difference between average and best in industry email marketers is based on the quality of the addresses that are being mailed to. Building a large list full of active-with-email subscribers is key to driving improved performance. AI can play a big role here in two areas:

  • Using AI to acquire more active addresses: AI techniques are great at optimizing for a single metrics, or even across several metrics. These techniques can be employed to change the mix of acquisition sources and (more likely) the kinds of offers/CTAs  that are used to acquire email addresses in order to optimize the level of activity of the client base.
  • Using AI to “clean” a database: Simple machine learning techniques can help marketers determine when a dormant address has passed the “point of no return,” when the address is highly unlikely to ever be active again.  



Optimizing creative. Thoughthis subject has been getting a lot of interest in the email marketing industry, there are a few areas that haven’t gotten as much attention as they deserve:

  • Multivariate testing for batch and segmented batch mailings: Email marketers as a group don’t test as much as they probably should. Those that do tend to use traditional A/B testing. While true multivariate testing technologies require more variations for each dimension of the creative (offer, CTA, etc.), it is much quicker to find to optimal open, click, and conversion rates. Multivariate testing is particularly useful when sending “batch” and “segmented batch”—still the staple of many email marketing campaigns.
  • Content creation beyond subject lines: Natural language generation techniques have been widely used for several years now to generate and optimize subject lines. AI can go beyond the subject line to create and optimize text in the body of emails.

Optimizing audience: A lot of the focus in the industry has been on determining when to send email. Optimizing send time for individuals works particularly well for batch retail campaigns. However, AI can also be used to optimize who you should send mail to and how many messages to send to individuals. There are several use cases:

  • Improving inbox placement: By carefully choosing whom you mail to, whom you don’t mail to, and the order in which you send mail, you can improve inbox placement rates while minimizing lost opens and clicks. This is an area that I believe is ripe for innovation.
  • Increasing lifetime value of an address: I’ve discussed this concept before as minimizing “list fatigue.” By using UI to carefully choose whom you mail to, what you mail to them, and how frequently you mail to each address, you can reduce unsubscribes, subscribers becoming non-responsive, and complaints, while minimizing “lost” opens, clicks, and conversions. This maximizes the lifetime value of each address. 

It will be exciting to see what the future holds for this area. What other applications do you see for AI in email?

Next story loading loading..