A leading travel company that has over half of its passenger sales coming via its online-site approached Rumble looking for a lucrative optimization strategy to increase sales of packaged deals.
In executing the optimization campaign, the client selected the key performance indicator (KPI) of the Conversion Rate, with the goal of increasing the total packaged-deals sold. The company’s online average conversion rate hovered at 3.3% and was a bit lower than the industry average. Using Rumble Px platform, they were aiming to increase it by 20% or higher.
Rumble Px in Action
Among the different Business Tools provided by Rumble Px, the client decided EasyNav Floater™ would be the most suitable tool to increase the sale of its packaged deals. When customers were navigating through the site, the tool prompted a “Book Early” discount on a packaged deal that matched the customer’s interest – increasing the likelihood to purchase.
Our client had identified the segments it wished to target. As it was mid-winter, they opted to target single professionals with medium to high household income. Additionally, Rumble Px had the ability to further segment the target audience by historical and real-time engagement data. Rumble Px then determined the optimal segmentation mix at each moment in time during the optimization.
The customer has defined several packaged deals, which was given to all users for a short period of time using EasyNav Floater™. This helped us collect data to train our machine-learning model. During this time, we made sure to collect the query parameters made by the user when searching for a hotel/flight/car rental. This way we could understand the user need.
From that data we discovered that the majority of the users behave in a certain way: First order a flight, and only than take care of the accommodation and transportation.
The next phase was to figure out the different micro segments under the constraint of “single professionals with medium to high household income”. We used K-Means Clustering algorithm with Sweep Clustering (an Azure ML feature that finds the best algorithm attributes) to segment the users.
This was the clustering experiment.
Taking the prior information we had and some of the knowledge from the client we trained a model that ranked the probability of a user to convert with a specific packaged deal. We than deployed the model to production – serving 90% of the users with the EasyNav Floater™, and the remaining 10% users as the control group.
Shortly after optimizing the delivery of the packaged deals using EasyNav Floater™, the client saw +17% uplift in conversions. By week two of the campaign, the conversion rate increase reached a steady 23% – resulting in a significant increase in both sales and revenue.
The following graph shows how the group of customers that interacted with EasyNav Floater™ registered a higher conversion rate (+23%) than the control group.