Travy ta Kavy – AI personalization for e-commerce
Recommendation system + AI pop-ups that work like a live salesperson
AI
E-commerce
Machine Learning
Client
Travy ta Kavy – online store of premium teas and coffees
The store approached us to increase conversion and average checkout. The client understood that users needed help choosing from a large assortment of products, but hiring live consultants for online chat was not suitable due to the high cost and limited working hours.
Project team
Project manager
Backend Developer
Frontend Developer
ML Engineer
Workflow
Integrating AI technologies into e-commerce: from data analysis to personalization
Problem
The store had a large assortment (over 200 items), but conversion remained low because users could not quickly find what they were looking for. Standard filters and search did not solve the problem, as many customers did not know exactly what they were looking for, but simply wanted “something tasty” or “something for a gift”.
Approach
We decided to implement a comprehensive AI solution with two components: a recommendation system and dynamic pop-ups. The recommendation system analyzes user behavior (products viewed, time on the page, added to cart) and offers personalized selections. AI pop-ups appear at key moments in the buyer’s journey and help make a decision.
Decision
The implemented system consists of several components:
- Recommendation system: uses machine learning models to create personalized product selections, trending items, and hybrid scenarios (behavior + popularity)
- AI pop-up at first visit: soft engagement with an explanation of the store concept
- AI pop-up on product pages: product comments, every third one with humor to increase engagement
- AI pop-up in cart: personalized recommendations and unobtrusive upsell based on cart content
WHAT IS AI Recommendation?
AI Recommendation is an automated tool built on modern machine learning (ML) models that continuously learns the actions of each customer. ML algorithms deeply analyze behavioral factors in real time: pages viewed, time on the site, abandoned carts, and shopping history. By learning from this data, the system independently detects hidden patterns and predicts the user’s next steps with high accuracy. This allows you to completely abandon manual settings, as predictive ML models automatically adapt offers to each visitor, creating unique product selections and making interaction with the store as personalized as possible.
System by itself
Studies the behavior of each client
Forms Individual product selections
Works without your participation every day
HOW IT WAS BEFORE
Managers had to spend hours analyzing tables and manually hanging static “top sellers” on products. It took a lot of time, and most importantly, the storefront was the same for all buyers, regardless of their tastes.
Marketing works blindly and is not based on data: all site visitors are shown exactly the same products and offers. Due to the lack of an individual approach, customers do not find the right things and quickly leave. Owners have to spend a lot of time manually updating storefronts, but the store still cannot adapt to the tastes of a particular person.
NOW
Artificial intelligence has taken over the routine. Now the system analyzes every click in real time and forms dynamic selections on its own. Instead of blind “tops”, each customer sees what he is most likely to buy.
The recommendation system turns the store into a smart platform: algorithms continuously analyze the behavior of each customer and his browsing history. Based on this, artificial intelligence automatically forms unique showcases and product selections. The process is completely autonomous – the system itself shows the person what he needs, keeping his attention and increasing sales without manual work.
What we implemented
Personal recommendations on the site
The system automatically collects and analyzes the digital footprint of each visitor: views, dwell time on specific positions, interaction with the cart, and history of previous purchases. Based on this data, the algorithm builds a personalized storefront in real time, which adapts to the client’s tastes after the first minute of being on the site. This allows you to shorten the path to purchase and increases conversion by an average of 18%, since the user sees exactly what he is looking for.
Similar products
The buyer sees logical alternatives or additions, not random products. Instead of randomly issuing, AI compares the composition, price category and purpose of the products to offer the buyer the most relevant alternatives. If the desired tea or coffee is not available or the customer hesitates, the system selects a logical addition that meets his request. This is not just “neighboring products from the category”, but deep similarity analytics that helps keep the customer on the site.
Often bought together
The system shows pairs of products that are often purchased together — this creates a sense of “set”. The algorithm studies thousands of successful orders and finds non-obvious connections between products, forming ready-made sets. The system offers additional items that perfectly complement the main choice (for example, a specific type of sweets to the selected type of coffee). This creates the effect of a professional consultant who knows the best combinations, which encourages customers to add more products to the check. This approach ensures an increase in the average check by 23% without any manual moderation from the store owner.
Smart AI Pop-ups: Personalized Guidance
These intelligent widgets work in synergy with products at every stage of the customer journey. Algorithms analyze user actions and automatically adapt messages—from a warm welcome to personal advice or a well-timed joke—gently nudging them towards a purchase.
Results
AI personalization that sells without human intervention
The implementation of AI solutions yielded tangible results in the first month of operation. The recommendation system and pop-ups created the effect of a live salesperson who knows what to offer each visitor.
- 18% conversion increase: Personalized recommendations and timely AI pop-ups helped more visitors make purchasing decisions
- Average check growth by 23%: Unobtrusive upsell and personalized recommendations in the cart stimulated additional purchases
- Increase repeat purchases by 15%: AI system remembers customer preferences and offers relevant new products on subsequent visits
The client was satisfied with the results and plans to expand the use of AI technologies to email newsletters and a Telegram bot.
We are always ready to help you.
Start automating sales now