Aware of the scenario and customer demand, Compass UOL developed, in partnership with AWS, a personalized model to determine the ideal fares in airline ticket purchase requests
Compass UOL applied a Learning Language Model (LLM) to contextualize the recommendations, making it possible to offer personalized and informative suggestions according to the specific needs of each client.
The complexity involved in evaluating different flight options leads to an overvaluation of small cost differences. Travel agencies face limitations in the fares offered by airlines, unable to influence customers to opt for those that are potentially more profitable, even when they are more suitable for the user and have reduced cost variation.
The model developed uses airline information, historical transactional data and the application of LLM for contextualization.
Amazon SageMaker was used to create the customized model, while Amazon Bedrock was used by setting a prompt to restrict the recommendations, ensuring a more contextualized and humanized approach.
Data integration was carried out using AWS Glue with storage in Amazon S3. The resulting recommendations were stored in Amazon S3 and consumed via AWS Lambda Functions for the Amazon API Gateway.
The complete infrastructure was created via AWS CloudFormation, providing access to Amazon SageMaker Studio for the data scientists involved. It is a flexible structure that can be adapted to different projects.
Recommendation to adopt the most advantageous tariff
providing an improved customer experience and optimizing profits, even in situations where this tariff is not the lowest.
Improved conversion rate:
we are looking for an increase in conversion rates as a result of implementing this recommendation.
Improved user satisfaction:
we anticipate significant improvements in user satisfaction by offering personalized and appropriate recommendations.
Increase in average fare:
we project this increase to indicate the acceptance of sales with higher costs, but considered more convenient by users.
Reduction in search abandonment:
we expect a decrease in this rate, indicating greater efficiency in users' ability to find what they are looking for.
Read other stories
PagueMenos
Brazil’s second largest pharmacy chain uses AI & ML to personalize product ranking and experiences
Compre & Alugue Agora
Brazil’s first and only social media for Real Estate reduces chatbot costs by 80% and improves responses with AI