Fact Combiner

When starting to plan their vacation, some customers have no idea where exactly they’d like to go – or their preferences are quite vague. Since they’re using the e-commerce platform instead of visiting the travel agency’s office, the algorithm, in a way, takes over the agent’s role. Imagine you feel an impulse to book a trip somewhere, but you’re not sure of your preferences. When entering the booking portal, some suggestions that will guide you through the purchase process may come in handy.
The more accurate they are, the better it is both for the client and the vendor. Our solution allows you to get to know your customers – by knowing their preferences and behaviors, you will find a perfectly tailored offer for them. The first spends less time searching for a perfect trip, and the second increases the revenue since the conversion rate is rising. There is a way to achieve that effect while making the best use out of the collected data. Meet Fact Combiner, one of Axabee’s projects!

About the client

Fact Combiner was created for Itaka – the largest tour operator in Poland and one of the largest in Central and Eastern Europe. As a travel agency with a global reach, Itaka uses an advanced booking and reservation system- Resabee (also implemented by Axabee) with many unique features. Since its flow and intuitiveness influences the sales directly, the company continues to introduce additional functionalities that may increase the conversion rate and improve the overall user experience with the platform. Fact Combiner is one of them.


The travel industry is one of the most dynamically changing sectors. In the last two decades, it has undergone a 180 degrees metamorphosis. The trips abroad and to other continents have become easier and much more affordable – and thus, gained a lot of popularity. As a result, the market got saturated with travel companies. At the same time, it became common to book trips online. That, together with rising competition, has generated a demand for seamless booking solutions that fuel sales without the intervention of the travel agent. Accurate recommendations are the key to it!
Important note: we’ve developed Fact Combiner as a response to Itaka’s demand for well-targeted recommendations, but it’s applicable to any other internal system, not only within the travel industry.

What did we work on?

Our project is an add-on to the internal systems of the travel companies based on a set of efficient algorithms. Its main purposes include:
1. Recommending products to users
2. Arranging the products in the search results in an efficient way
3. Getting familiar with our customers’ behaviors
4. Gathering the data for the automation marketing purposes
5. Creating marketing profiles
6. Creating real-time stats
All these aspects might come in handy in the optimization of sales funnels and profiling the marketing campaigns.


The data is modern gold, regardless of the industry. In many cases, even though the companies have access to it, they don’t take advantage of its full potential. If used thoughtfully, data can become a powerful tool for attracting and retaining clients. Via our data-processing solution, we wanted to help the companies generate better-targeted purchase recommendations and design effective automated marketing campaigns. At the end, it all comes down to an efficient algorithm!


Since the frequency of purchases in the travel sector is relatively low compared to other industries, the recommendation mechanism really makes a difference. The clients buy one or two tours per year – thus, it should be accurate enough to take advantage of this scarce opportunity. To make that possible, we chose an algorithm that finds similar users and bases the recommendations on their searches.

Due to the large amounts of data being processed, we had to make the algorithms as efficient as possible. How did we achieve that? Below, you’ll find our working process explained step-by-step.


For the e-commerce platform, customer-related information is the most precious asset. Fact Combiner enables making the best out of it on various levels. How? Let’s take a closer look at all its functionalities and the ways we developed them.

Recommending product to users

When entering the booking application, the user follows a particular path. We gather all these paths to make the project as intuitive as possible. In this case, to refine the recommendation system, we’ve researched associations between different actions and the booking. We’ve noticed, for instance, that when the users check the flight hours, they’re relatively close to finalizing the order. This type of information helped us polish the system’s accuracy.

Aside from this algorithm, we also use two others:

1. If we don’t know anything about the user, we recommend them the TOP options
2. If we have a little knowledge about the user, we use the association rules – an algorithm based on comparison which connects the dots and puts the products often browsed together next to each other.

The practical implementation

The results of our recommendation model manifest as the “recommended offers” bar on the main page and the product page. It also serves as a support system for mailing and search results.

We’re able to define the number of displayed recommended products. For now, it’s 3 – but it can change if there’s such a necessity. It’s worth mentioning, that we generate real-time recommendations for 60K users every day.

Arranging the products in the search results

80% of users enter the travel platforms with an intention to find the trip. While researching the user’s decisions, we’ve noticed that they constitute a relatively common pattern. The first three results get clicks, the rest is ignored, while the last one also gets some attention. Over half of the clicks go to the first and second products in the search results. These conclusions were helpful for polishing our solution.

Our Fact Combiner gathers data on user’s preferences, which later serve for perfecting the recommendation system. Observing their moves, we’re able to see if they were interested in the products we’ve displayed. Knowing that the products on the top receive the majority of clicks, we test the offers, placing them in different positions.

Important note: Part of our users sorts the products by price. Instead of discouraging them from doing so, we mark the promoted products which don’t fall under the “sorted” category. We checked that for 60% of searches, users use the target filter, which is browsing popular trips. However, 33% of users choose to sort products by price. There’s also an option of adding them to whitelists and blacklists, if necessary.

Getting familiar with our customers’ behaviors

Our tool enables gathering detailed data on our users’ actions. Thanks to that, we’re able to measure the efficiency of affiliate campaigns. What kind of data does it collect? Everything that may matter in the marketing context – like, how many people were looking at the particular trip, or their geolocation data. It helps us with building the big picture and optimizing processes.

Gathering the data for the automation marketing purposes

Most users are anonymous, but some log into the system – in this case, we have access to their mail address. It can serve the purposes of marketing automation. Based on the data gathered by Fact Combiner, we’re able to generate and send reminders that encourage users to finalize their purchases. The mail’s structure depends on the path the customer passed through and the stage at which they abandoned their cart (if they did).

Creating marketing profiles

All the users’ actions – whether they’re browsing anonymously or logged in – can serve us for creating marketing profiles. These are extremely useful tools for optimizing sales campaigns and many other purposes – but only if based on reliable and meaningful data. With our system, you can access any statistics you need – individual for each accommodation and the aggregated ones. Since data goes straight to the CRM, operating it is much easier.

The profiles can be filled with information such as: – the first browsing session
– the frequency of entries
– the browsed offers
– the booked offers

Currently, we have about 600K behavioral profiles of registered users in the second half of 2021. Plus over 4M anonymous profiles from the last month. All updated in real time.

Real-time stats

Using real-time data is a great way to encourage users to finalize their purchases. Seeing that someone else has bought the same products just a few seconds ago, they may be more willing to make their decision right away. You can introduce it to an e-commerce platform in the form of real-time stats that report, for instance, how many people saved the offer and when was the last booking. What’s important, our system includes offline bookings, too. That provides the highest accuracy of the indications.


The use of Fact Combiner on the e-Commerce platform has provided insights into customer behavior that were previously difficult to learn. Since the start of using Fact Combiner, Itaka has personalized trip search for individual customer preferences. Thanks to that, implementation of our solution has contributed to increased sales. Fact Combiner also has a positive impact on the customer experience when browsing through offers from tour operator. We track nearly 120M user actions each month, allowing us to simplify the entire shopping process for users and customize the site for their needs on an ongoing basis.
For one of our main projects – a mobile app for Itaka, we also applied Fact Combiner action. To provide the same experience in searching for tours on a mobile device as on the website. Currently, Fact Combiner action improves the creation of the list of searched tours and allows, based on the customer profile, to find tours in which the customer may be potentially interested.

Our project is a complex but expandable IT solution for travel that will most likely grow in the following years. We can integrate it on different levels, depending on the client’s demand.

What to expect if you decide to introduce Fact Combiner?

The effects after implementing our solution are visible in increased revenue. You are able to see how the path to purchase on your platform has changed and how sales have increased. Thanks to the use of algorithms that, based on customer behavior and preferences, find the next products that will be perfectly matched with your customer needs.

The implementation of our solution can take up to 5 weeks, depending on the complexity of the client’s system. Note that it requires some interventions in the frontend and backend. Thus, we may need some support from your development team. If they cooperate with us and share their know-how, the whole process will most likely become shorter and much more efficient.

You want to give it a go? Reach us out to talk about the details!

Get in touch with us!

Let’s make something great together.

    Web development


    Our product is a complex IT solution for travel & e-Commerce companies that want to improve sales through algorithms that suggest products according to individual customer preferences.

    See more

    Mobile app development


    Specially designed to fulfill the needs of tour operators, our application will help to plan and administer guided tours.

    See more

    Web development


    An automated system that allows you to sell and send online vouchers for customers who wish to give your products as a gift.

    See more