Case Study 1 ( The Rezidor Hotel Group )Case Study 1: Predictive Modelling & Segmentation for Rezidor
The Challenge
The Rezidor Hotel Group is one of the fastest growing hotel companies in the world, operating five brands including Radisson SAS Hotels & Resorts, Park Inn, Country Inn, Regent and Missoni.
Rezidor engaged Cogent Analytics to undertake exploratory analysis of their loyalty programme (Goldpoints Plus) database with a view to understanding who their customers are and to deliver data analysis that could be used to drive retention and incremental revenue.
Segmentation
Cogent Analytics created a multi-dimensional 'Pen Portrait' segmentation model that drew data from 4 different sources; historical transaction data, profile data, market research data and 3rd party socio-demographic overlays.
The four data sources were pooled and modelled down to a customer level and then segmented across a number of factors (frequency, value, travel patterns, socio-demographics and competitor behaviour amongst others) before being distilled into eight distinct (and manageable) consumer segments.
These segments were then worked up into visuals using interesting imagery and key supporting figures to bring the characteristics of each segment to life.
Predictive Modelling
The Elite tier card-holders in the programme are extremely valuable customers in the contribution they make. Cogent Analytics was given the brief to develop a model that predicted the potential contribution a member could make as early on as possible.
We carried out exploratory analysis using historical data, discovering that two key variables did indeed suggest high value at a very early stage in the first membership year.
A model was then created using SAS software that scores each new member according to potential value and then tested, again using historical data.
The test showed that the prediction is accurate in 80% of cases and the model was made operational.
All new members are now scored according to value potential, and segmented communication streams are applied to each customer depending on predicted score.