Dmitry Spuntik

CEO, Founder eSputnik

Ideas for Using Artificial Intelligence in Email Marketing

Ten years ago implementation of artificial intelligence (AI) in large-scale business processes seemed to be unreliable, inefficient and noncompetitive with human’s performance. However, today AI really astonishes: it can paint pictures and guess what you wanted to draw from a few sketches. And that is just the beginning, as neural networks and bigdata open new possibilities for AI. That is why today it is essential to include potential use of AI elements in the future business models whether it is a manufacturing or consulting company.

Email marketing is one of the areas where AI is not just advisable, but obligatory for implementation. Great amount of incoming and outgoing data, real response of end users, the possibility of monitoring tendencies, making experiments and processing huge data arrays are excellent instrumental in AI implementation and learning. You may ask: what is the use of it? Just recollect how computers have changed our life. We are sure that AI mainstreaming in business processes will have an immense economic and progressive effect.

So, let’s consider the areas of AI implementation for email marketing optimization and increase of your business efficiency.

Artificial intelligence in marketing

1. Reactivation Recommendations

We often hear the question: What is the right moment for reactivation of clients or subscribers? The answer is always quite general without due regard to your business segment and your subscribers.

But in practice for smart reactivation campaign we need clear signs showing that this or that client is being lost, as every client requires individual approach and one client should be reactivated in a week, another one - in a month and the rest - in six months. In this case AI makes segments of clients, determines reactivation time frame and creates event “it’s time to reactivate”.


2. Frequency Recommendation Engine

Let’s imagine that there is a filter for passing emails on the basis of some criteria. These criteria can be absolutely different, for example, a client has already received this email, unsubscribed or marked email as spam. Any action which can be considered as an entry point may be used: previous activity, source (the channel of getting a subscriber), email subject and category, emails planned for this subscriber. The task of AI is to detect these criteria. 


Speed – no more manual strategy description. 

Efficiency – no need to do segmentation by client’s activity, subscribers stay engaged in campaigns and aren’t disturbed by irrelevant triggers.

More detailed information: Management of campaign frequency

3. The Product Recommendations

Personalized product recommendations is the most popular sphere of AI application for sales increase. AI makes product recommendations on the basis of user’s behavior in online store and in the Internet. Recommendations can be added as an additional block in any email type: regular promotional campaign, any triggered message (order confirmation, cart abandonment email, invitation to give feedback) or sent in a separate email at any time or in accordance with scenario.

Yves rocher email example


4. The Order Of Promo Blocks Suggestion

There is one more process where AI is much required. Imagine that you have a list of products for promotional campaigns and all of them should be used in any way. All subscribers have different preferences and few people will buy a lawn mower instead of tablet. AI is of great use here to monitor subscriber’s insights. 

Along with that, there are a number of other questions: what to write in subject line? It would be nice to list top 3 offers in subject line and put the other top 2 in preheader. What is more efficient: to mention brand name or product category in subject line? 

It is not a secret that this is a subject line which defines opening rate, however the offers work the best when they are put at the head of the list.


5. Marketing Qualification Recommendation

Defining the most valuable contacts as potential clients can be done on the basis of criteria and features collected, organized and assigned to each contact. It isn’t just about such simple criteria as email opening. Let’s make parallels with IT company, which requires managers, who can make best solutions with account of client’s needs, cooperate with big team having high staff turnover, know English, have more than 3 years of managerial experience and know one or several programming languages. The task of AI is to find the right people matching specific criteria.


6. Sales Qualification Recommendation 

This function is used in the same way as marketing component with the only difference: subscriber is ready to become a buyer. There are a lot of signs showing that a client is ready to make a purchase: view of similar products, reading reviews, questions in reviews, reactions to discounts and overviews of desired product. All this is a signal to sell. 


7. Customer Lifecycle Optimisation Recommendation

To promote a client in the sales process series of messages should be sent to different channels. At the same time, there are numerous important issues influencing on the achievement of the desired result, such as: when the first email should be sent? Which channel to use? What is the best pushing "message" in email? How long should we wait before sending the last email? 

This is really time consuming and labour intensive to draw a plan with all possible variations. At such volumes no time is left for client support. It would be much more convenient to make a general pattern (stages, email variants and their parameters, possible communication channels etc.) and the entire process of calculations and comparison will be done by AI with no involvement of marketing specialist. 


8. Build Segment by Personas

To build a segment by personas we use a template of an ideal client with a set of suitable characteristics. Further, we create a segment of clients similar by behavior and key characteristics. As a result, we get a data array about an ideal client, which is constantly updated and self-learning. 


9. Find Unique Group (Clusterization)

Segments differing from other segments are distinguished algorithmically by a set of characteristics and features (with the use of neuronet and by indistinct features). Examination of such segments allows to make special and more effective offers to groups of subscribers previously disregarded. 


10. Score Email Copy and Fine Audience

If you have prepared an offer or an article, you need a method which defines subscribers potentially interested in this topic, to whom sending this info will be the most efficient. You can indicate segment's size with account of your budget or offer's exclusivity. This method also helps to increase the chance of audience's reaction by changing some message parts on the basis of the results received earlier. 


11. Personal Content Generator 

The method is similar to the one mentioned above: you also have an email and several different topics or forms of offer submission in email and you need to make a focus on discount, product's popularity or characteristics in a personalized way. And this is the task of AI to define what should be focused for each client. 

Split test of email topics


12. Send Time Optimization (Day of Week and Time) 

Everything is simple here: AI is used to define when, to whom and through what channels a message should be sent within specified time frame to maximize the intended result.

All the same, even the smartest AI still requires human intervention. That is why we have created a whole department of mathematicians, who build mathematical models and estimate the chances, implement and test results.

Special Request Inline



3.8 from 5 based on 64 reviews

Dmitry Spuntik

CEO, Founder eSputnik

Comments 0