Artificial Intelligence

To help our clients improve their marketing results, eSputnik has implemented Artificial Intelligence (AI) to minimize the effort, automate all possible processes, and take direct marketing to a fundamentally new level.

AI is a tool that analyzes big volumes of information of various types, generates data-driven solutions, and predicts customer behavior. While processing new information, AI constantly learns and adapts to new tasks, living up to its second name - machine learning.

Based on analytics on the constantly changing and updating data, AI algorithms help answer the major marketing question: To which recipient what content in what order when to send?

What’s more important, machine algorithms are built in the system, and AI-supported analysis is automated, meaning you don’t have to manually configure tasks each time. All you do is set up the conditions for analysis once; the system will run based on these conditions.

Below, we’ve listed 11 processes in the eSputnik system that are supported by Artificial Intelligence. Note, however, that these are already incorporated, currently available solutions. You can set up for analysis any data and eventually configure any custom AI-supported solution that would generate recommendations based on your particular business model and audience.

1. Content Display Hierarchy

Based on the customer website behavior, previous purchases, and most often browsed items, AI determines what product categories a particular customer is more likely to be interested in, and organizes the email content in the corresponding way.

For example, you’re a sports clothes shop that runs clearance sales with three discounted categories - cycling, boxing, and workout. With AI recommendations, the display order of the banners, product cards, CTA, or any other email block featuring the corresponding category can be tailored to each recipient. Thus, customers previously demonstrating most interest in cycling gear would see the cycling block first; customers who mostly used to buy workout clothing would see the workout block first. This approach applies to any email block.

Content Display Hierarchy

2. Product Recommendations

To come up with the personalized offers for such blocks as Recommended for You, Also Bought, You May Also Like, Staff Picks, Our Favorites, etc., AI analyzes both online (web tracking) and offline data: browsed categories/items, often viewed categories/items, last purchase; email click rate; online/offline promo code or bonus card usage, etc.

Based on it, it automatically fills the blocks with the content relevant for each particular recipient. You can add product recommendations to any campaign (bulk, triggered, transactional) or send them as a separate promo email.

Product Recommendations

3. VIP Segmentation

Based on the purchase frequency, average purchase check, and purchase categories (luxury, VIP), AI helps approach each customer with a personalized pricing policy by generating product recommendations within the individual price hierarchy. For example, customers spending $50 per purchase on average would be recommended products within the $50-80 range. Customers spending more $150 per purchase would be offered products from categories with a higher price.

VIP Segmentation

4. Recommendations Based on Social Media Activity

AI can analyze customer behavior - likes, shares, comments - on your corporate social media pages (Facebook, Instagram, Twitter, etc.) and use it to build more precise and accurate recommendations. This solution is especially efficient for businesses who run their main marketing activity via social media and not via the website, or don’t have a website at all.

5. Reactivation

Any business segments customers based on their shopping activity - active customers, sleeping customers, inactive customers. Based on the consistency of the website activity, AI algorithms can predict the pattern of customer migration between these segments. As a result, you’ll be able to drive sleeping and inactive users back by sending them the corresponding reactivation campaigns at the right time.

6. Signal for Sales Offers

Based on a number of metrics (increased website visits, growing browses/views, reviews and feedback exploring, questions in a chat, clicks on discounts, average time of decision making, etc.) AI allows to figure out a good time for sales offering. The system predicts when a customer is ready to make a purchase and only needs a little encouragement to opt for your product. At this very moment, you can send a little incentive as a promo code, discount, free shipping, etc. to win this client over competitors.

7. Building of Unique Segments (Clustering)

In the eSputnik system, AI can detect unique segments based on certain behavioral patterns typical of particular groups of customers. For example, you have a general segment of customers who buy only sports gear. However, some of them prefer only a certain brand, say Patagonia, and never buy anything else. Some make the purchases with the biggest average check at the end of every month. Some always choose free shipping and never click any other incentives. Some may do all.

To send the most relevant and response-generating offers, you can build separate segments for each group of customers and approach each with a specific marketing strategy. Instead of general campaigns, send Patagonia fans always opting for a free delivery campaign with the corresponding brand recommendations and free shipping offer. Send customers who shop only at the end of the month campaigns in between 25 to 30.

8. Exclusion of Random Behavior

To come up with precise behavior prediction and product recommendations, AI analyzes a lot of different data we’ve mentioned above. However, there happen random acts (untypical purchases, social media activity, buying frequency, etc.) that fall out of the general customer behavior pattern.

For example, shopping history during holiday seasons can be a deviation from the established buying paradigm. In December, people may make several purchases intended as Christmas gifts but they aren’t interested in them in general and won’t respond to the corresponding offers at all other times beyond holidays.

AI tracks such random and accidental behavior and excludes it from the overall behavioral pattern to avoid irrelevant and inappropriate recommendations and offers at the usual time.

9. Customer Lifecycle Optimization

Each company calculates customer lifetime value (CLV) based on its business model and audience types. AI helps lead the customer through all stages of the lifecycle and make your relationships as long-term and productive as possible. It recommends the most optimal content and sending time for campaigns at each stage and indicates the most efficient communication channels.

10. Send Time Optimization

Based on the average open time of each customer, AI will send the campaign at the time when the recipient is more likely to respond.

For example, for a segment of 100,000 contacts, you schedule a regular promo campaign to be sent on any day of the week within an 8 AM - 7 PM time gap. Some of these recipients prefer to open emails only at the weekend; some only on weekdays. Some open only in the morning, some during a lunch break, and some only in the evening. Moreover, these preferential patterns may fluctuate or change completely.

The system analyzes all the data and will automatically send the campaign at the time that is the most optimal for a particular recipient at the particular time of sending. You don’t need to build additional segments or configure conditions: the process as well as all other AI-supported processes is run on its own.

11. Annoyance Level Optimization

AI helps determine the sending hierarchy for contacts included in several segments to avoid bombarding people with numerous messages at a time. For example, a contact has made a purchase and received a transactional email; soon after that, on that very day, they received a regular newsletter and another two hours later - an anniversary greeting. Such a schedule is too tight and may result in subscriber’s confusion and annoyance.

To avoid this, in the eSputnik you can set up an annoyance limit. Each contact will be sent messages based on the type of the email and the number of segments the contact is included in.

Annoyance Level Optimization


As you can see, AI can become a powerful tool in your marketing arsenal. It helps analyze big customer data, predict behavior, build advanced segments, optimize content, and personalize sending time. It also saves time and effort of your marketing team as all machine learning processes are automated and don’t require manual configuration.

The more diverse your product range and audience, the more assistance AI-supported recommendations can provide. Use them to address each customer with an individual communication strategy and targeted offers. Such an approach will generate a bigger response, turning one-time customers into the loyal audience, and boosting your sales.