We solved the optimization problem of pricing thousands of short-term rental apartments daily with a dynamic pricing model.
Forenom’s goal was to increase customer satisfaction, not just get an optimal price per venue. The result is an online reinforcement learning model that enhances profits and makes both visitors and Forenom happier.
*In the locations selected for a/b testing.
Benefits of the dynamic pricing model
Forenom is a rapidly growing serviced apartment provider specializing in business customers. They offer 7,500 serviced apartments, aparthotels and hostel rooms in all major Scandinavian cities. Over 200 000 guests stay at Forenom locations every year.
To achieve great customer satisfaction, it’s imperative that customers get the best price from Forenom’s own online booking system.
The Forenom revenue management team needed clear pricing guidelines. Before the project with Fourkind, the team offered the best market price based on the old pricing system. Updating the prices required hours of manual labour and prevented the team from focusing on developing their business.
“Fourkind were not just tech vendors, they helped us understand the role and possibilities that data has in our entire business.”
“One of the greatest benefits of our cooperation was that manual pricing work decreased.”
Director of Revenue Management
An ambitious pricing strategy calls for an ambitious pricing solution
Forenom was developing a new enterprise resource planning (ERP) system, as the old system was unable to support their growth. Development of the new ERP made it possible to make use of a dynamic pricing model that would require little manual labour.
Standard rule-based dynamic pricing software was quickly eliminated as an option, as Forenom wanted to execute their own‚ ambitious pricing strategy. This called for a cutting-edge dynamic pricing solution would be based on real data and would become more accurate by the day.
Besides customer satisfaction and operational efficiency, Forenom was keen to maximise revenue per available room. This means finding the optimal price – sometimes higher, sometimes lower – for each room, every night of the year.
The process began with a deep-dive into Forenom’s unique business challenges. By interviewing key people from revenue management to sales, we defined the most important aspects and measures that should influence the new dynamic pricing model and determined which functions needed optimizing. The main goal was to increase customer satisfaction, not just to get the optimal price per apartment.
A pricing model that gets better with every booking
The new solution is based on machine learning and was tested in two destinations in Helsinki. After just two months, the first version of the new dynamic pricing model was A/B tested against the old pricing model.
The new model was already performing significantly better than the old one, and we knew the results would only get better. The reinforcement learning-based model gets feedback from each booking, continuously learning to optimize prices. The most important piece of feedback used by the model is whether or not a specific price offered led to a booking.
After a couple of months of A/B testing, new rules were established. For example, Forenom didn’t want popular smaller rooms to cost more than the bigger, seldom booked apartments. Pricing should also follow the same rules across channels.
The intelligent system uses a myriad of data points to arrive at ideal prices. Data points and volumes vary in different locations.
The project was not only about building a technical engine, but about improving the pricing operations of a growing company as a whole. The dynamic pricing solution supports the ongoing development of Forenom’s business.
From the get-go, Forenom’s goal was to increase customer satisfaction, not just get an optimal price per venue. More customers now book their visits on Forenom’s own webpage since they can trust they’ll find the best price there.
Since apartments are no longer priced individually or manually, Forenom can now move towards implementing a comprehensive pricing strategy. The last A/B test showed a significant 13 % increase in revenue per room in the group of locations (23% of total capacity) that were included in the test. The increased revenue is based on either higher booking volume, higher pricing points or longer booking times.