I recently had a conversation with the founder of a small skincare brand while she was going through her email dashboard. I had promised her a while back that we would go through her email reports, and as we were going through them, she kept apologizing for her numbers. They weren’t terrible, and in fact her open rate was 38% which is actually pretty solid for the skincare industry. However, she had read somewhere that with the use of AI, email marketing would solve all of her problems and she felt like she was falling behind because she wasn’t using it properly. That conversation ended up being the reason I wrote this post.
My previous article about trends in email marketing also related to the same topic and helped to develop my thoughts about email marketing trends in 2026, which are less about doing more and more about having better judgment and doing less but with better results. Most of the trends, which are described in the articles, are related to the use of more tools.
AI Isn’t Writing Subject Lines Anymore. It’s Making the Calls.
I remember about two years ago AI in email marketing could be written as a service to assist writers in composing emails. Later those programs evolved to generate subject lines. And now, AI decides what to write in an email and even if to send it.
The second difference is that today’s AI is not just deciding on the wording of a particular email but is also deciding on the sequence of emails. As an example, a home goods brand that I am consulting for had a traditional abandon browser flow in place. The old flow consisted of three emails sent out on a fixed schedule (i.e. one hour after abandonment, one day after abandonment, and three days after abandonment). These rules were based on simple if/then logic and anyone could have set up such a flow in an afternoon.
We ran the browse-abandonment campaign using an automation tool, that uses the same logic as for sending email campaigns to individual customers. Instead of fixed 1-hour, 24-hours and 72-hours after add-to-browse-list, the campaign was now triggered by various events: opening on mobile at 11pm at night, clicking on a link to product page, or clicking on link to product details page, and so on. Some of the settings for the automation tool were not optimal on the first pass, for example, the tool kept sending out discount codes to returning customers who didn’t need them. But after fine tuning the settings, the resulting campaign had revenue per recipient of around 18% above the campaign with fixed settings.
That’s the actual shift going on, less AI writing out your emails and more AI deciding when the right time is to send out that email to the right person.
Personalization Got Quieter, and Honestly, That’s Better
Personalization used to mean simple things like putting the first name of the customer in the subject line of an email and then later on other personalizing elements would be added to create a very ‘personal’ experience for the customer. Today, personalization has gotten quieter. No longer do emails shout out ‘WE PICKED THESE JUST FOR YOU, SARAH’. In fact, today, personalization has become so much a part of most of the emails that are sent that it’s become almost invisible.
So when I say that “people got tired of being told they were being personalized to”, I mean that being told that you were being personalized to in order to be brought into the fold of the personalized world started to feel like a sales ploy. In other words, it was no longer seen as being personalized to, but as being treated in a very particular way in order to try and get the customer to part with their hard earned cash.
Recently I’ve been working with a small independent bookstore and we’ve tried a few different ways of using information about a customer’s previous purchases in order to suggest other books to them. Initially we made a big deal of the fact that the books were being recommended to customers on the basis of information about their previous purchases, and we used a lot of Performative Language in order to draw attention to the fact that the recommendations were being provided on the basis of the customers’ previous behavior. However, this approach had a number of problems, not least the fact that it meant that customers were being subjected to a lot of similar messaging, which in turn meant that the overall impact of the emails was to make customers feel as though they were being bombarded by emails. In the end, therefore, we took a step back from the approach and started again from scratch.
This time, instead of making a big deal of the fact that the books were being recommended on the basis of customers’ previous behavior, we simply included a few books in the email and hoped that they would be of interest to the customers who received it. In the event, the approach worked extremely well. In fact, so well that the click-through rate on the books that had been suggested to customers on the basis of their previous behavior was nearly double that of the similar block of books that had been included in previous emails and had been recommended to customers on the basis of their previous behavior using a lot of Performative Language.
In other words, customers didn’t need to be told that the books were being recommended to them on the basis of their previous behavior in order to be able to tell that was the case. In fact, the less said about the matter the better. Which is perhaps the main point that I am trying to make in this section. Sometimes the best personalization is that which goes unnoticed by the customer. It is sufficient that they feel that the brand in question has been paying attention to them, without any need for further explanation.
Nobody Wants to Hand Over Data for Free Anymore, So Brands Are Asking Nicely
The privacy landscape is changing fast and recently Apple’s Mail Privacy Protection has been the major factor in the recent changes to tracking in email. On top of that, there are now even stricter regional data laws that are having a profound impact on email tracking and even open rate reporting. As a result, by 2026, many brands will have given up on trying to report on open rates and instead will report on click through rates and even most importantly revenue.
As mentioned earlier, Zero-party data is not a new technical trend but actually a clever way of brands admitting that guessing is expensive and that asking is free. To get the best possible data, brands now ask for it in the most convenient way possible in return for useful information and services. The illustration above shows how a simple preference quiz was added to the welcome sequence of the coffee subscription brand’s email campaign. The two simple questions asked in the preference quiz are about the subscriber’s preference for roast strength and how often they drink coffee.
This simple preference quiz enables the brand to skip weeks of guessing and to send the most relevant recommendations from email one onwards. Day-30 retention for new subscribers has increased noticeably and the founder explained why she thought it worked so well by saying: “We stopped trying to be clever and just asked.”
Zero-party data isn’t a technical trend. It’s brands finally admitting that guessing is expensive and asking people for the information they need for truly relevant experiences is free.
The Inbox Is Starting to Act Like a Tiny App
Interactive email within the inbox that lets customers do things like reschedule a class or put products in a shopping basket within the email (without them having to leave the inbox) used to be the reserve of big retailers with big development budgets. But lately, with more email clients supporting things like forms, text input fields and more sophisticated things like calculating total cost of things in a shopping basket, interactive email is becoming mainstream and is even being used by smaller brands and retailers.
This shows how even very small businesses can implement simple interactive emails in order to make rescheduling of classes in their studios as easy as possible for their members. In fact, this has such an obvious impact that the no-shows at classes for this studio have decreased. This is because rescheduling of classes has ceased to be such a chore for its members that they avoid doing it.
Automation Is Starting to Talk to Other Automation
Previously, email, SMS, and app notifications had been run as separate systems, targeting the same customers. Now, in 2026, more and more brands are connecting the channels up together so that, rather than competing with each other, and sending out identical messages, the different channels of communication are working in harmony to provide the customer with the best possible service.
Behavior-based communication. Or simply putting automation to use. Recently, I was able to observe this in action with a subscription coffee provider. Instead of sending abandoned cart reminders via email, they were now being sent via SMS. Why? The system in question had been analyzing the previous behavior of customers in relation to email and SMS messages. It then used this information to decide which channel of communication was the best to use in individual cases. The result: a huge reduction in the number of messages being sent. And, unlike with most efforts to reduce the number of messages being sent, the number of unsubscribes even decreased.
Catching People Before They Quietly Disappear
Early warning signs that people are drifting away, and stopping them before they completely disappear from view, has become a significant part of my work. As I mentioned earlier, silent customers are much worse than unsubscribed ones, because at least with an unsubscribe you know something has happened.
One pet supplies retailer that I worked with recently added a check-in message to their welcome sequence. It flagged up customers who were showing a dip in open rates and an increase in the time between purchases, weeks before they would have appeared on a report of lapsed subscribers. It was a simple two-question preference quiz to determine whether or not they were still looking for supplies for their pets, and if so, how their needs may have changed. There was an easy way for customers to update their preferences. The responses to this check-in message were far higher than for their usual win-back campaign, and a number of the customers who had responded went on to make a purchase within a month of receiving the message. Importantly, though, none of them had been offered a discount. One important thing to note about re-engaging inactive subscribers is that you have to act on weak signals rather than waiting for someone to drop off of your list completely. By the time someone hasn’t opened an email in 90 days or so, it’s often too late for an easy re-engagement.
Where I’d Actually Slow Down
This reminds me of something that happened last year where I was using an AI tool to generate subject lines for a client’s holiday sale campaign. The subject lines were tested individually and all of them performed well. But as the days went by and the various reminder campaigns were sent, I noticed that the subject lines all had the same rhythm and were using the same types of language to create a sense of urgency. “Don’t miss out”, “Last chance” etc. It wasn’t until I got a few comments from subscribers (half-jokingly) asking if a robot had taken over the newsletter that I realized what was going on. It was as if the subject lines had been written by the same person but with slightly different wording. In the end it wasn’t a problem but it was a close call and it taught me an important lesson about using AI tools to automate aspects of your campaign. While they are great at optimizing individual emails, they are terrible at picking up on the bigger patterns and nuances that are going on in the rest of your campaign. And so, while they can help to free up some time, they are not a substitute for human common sense and the ability to make decisions based on your brand’s values and personality. Brand voice still needs a human ear.
Nothing too earth shattering here, just be careful as with any automation. The main point to note is that single email generation is simple for AI but a sequence of emails is a different story. It’s easy for the AI to miss out key points and things can go unnoticed until a human comes along to review. Therefore, brand voice needs to be checked by a human.
What This Actually Means If You’re Running Campaigns Right Now
None of this requires huge sums of money or a total reworking of the email programs already in place. All of this is starting to work its way into the various flows of email already in place. This means that rather than focusing on all the various technologies and systems in place to send emails, the best brands are simply paying attention. For my client with the skincare brand we only added one behavior based flow to her campaign and the numbers went up and even more important the lady seemed to be happy again.
Top comments (0)