Use Cases

Any questions? Feel completely free to contact us

Improving Customer Retention with Churn Analytics

Customer Churn Analytics : a short Explanation

Customer churn - or attrition - measures the number of clients who discontinue a service (cellphone plan, bank account, SaaS application...) or stop buying products (retail, e-commerce...) in a given time period. Churn rate is an important business metric as it reflects customer response to service, pricing, competition... As such, measuring churn, understanding the underlying reasons and being able to anticipate and manage risks associated to customer churn are key areas for continuous increase in business value.

Reducing Churn Rates through Predictive Analytics

Essentially, predictive churn modelling will achieve three goals : understand the key factors of client attrition, identify the clients most at risk of leaving, and provide targeted insights on which retention actions should be implemented.

In the data-driven world, predictive churn management is a trending topic full of new possibilities :
  • More Data : new data sources can be used on top of the usual transactional CRM data (profile and purchase history) to improve the performance of the predictive models - service usage, client feedback and customer service requests, social media interactions, web tracking data...can all be revelatory of customer churn.
  • More Analytics : advanced data science techniques are often needed to make sense of these heterogenous data sources and build smarter models that will take into account seasonality, dynamic user segments, and evolve with the company's catalog, business processes and client base.
  • More Solution Design : another opportunity (or challenge ?) resides in the design of the actual churn management solution that will be developped on top of the model - are we only mining data to understand the reasons for churn or are we building a fully integrated business application that scores potential churners and recommends specific retention actions ?
  • Data Science Studio for Churn Analytics : from raw data to business impact

    Churn patterns hide deeply complex behaviors. With DSS’s ability to handle large varieties of data sources and formats, you will be able to add new sources and signals, hence more features to your churn models, ultimately improving them for significantly optimized results.

    Getting and cleaning Data : Data Science Studio's advanced data wrangling features will facilitate the combination of structured (CRM, web analytics...) and unstructured (emails, calls...) data sources to get a 360° view of your clients.

    Predictive Modelling : whether you are building an automated decision system or trying to understand the underlying mechanisms of churn, DSS will let you use and compare a broad range of algorithms, from simple linear models to complex ensemble methods, and build the best solution for your needs.

  • From our blog : Building an end to end churn prediction model in DSS
  • Deploying a Data Product : with DSS's model deployment and data flow automation features, you will be able to go quickly from a model to a business application. From creating dashboards to feeding databases or marketing tools with scores and predictions or deploying a model on a API to be queried in real time by sales teams, all applications developped in DSS can be production ready.

    Featured Use Case : Strengthening Loyalty on Mobile Apps

    CoyoteCoyote worked with Dataiku to conduct behavioral analyses on their clients' usage of the Coyote mobile application. Data Science Studio was instrumental in combining and cleaning anonymized user data and real-time device data and create features to be used in the modeling: app usage, mileage, time spent on road, number of alerts, …
    As a result, Coyote was able to segment users and score potential churners to optimize sales and marketing campaigns, increasing conversion rate of outbound calls by 11% and significantly improving the efficiency of data management activities.

    Dataiku Churn Analytics example: Coyote

    Download the full Success Story

    Churn Analytics : other examples

    Churn for Telecom Providers :

    Challenges : very high acquisition costs paired with a decreasing ability to lock in customers
    Data : everything from basic CRM to device usage and interaction data (call details, terminal usage - app and data, customer service, location, social media interactions...)
    Typical Use Case : anticipating subscription cancellations and proposing specific commercial actions to foster loyalty  

    Churn for E-commerce Players :

    Challenges : low conversion rates, increased need for differentiation through recommendation / personnalisation
    Data : mostly CRM, catalog and clickstream data (behavior of website visitors)
    Typical Use Case : increasing loyalty and client lifetime value by activating personnalized campaigns to "dormant" clients - pushing the right product at the right time through the right channel  

    Churn for Banking & Insurance Companies :

    Challenges : very high acquisition costs and increasingly competitive market with fully data-driven new entrants
    Data : everything from basic CRM to granular credit card and insurance details data
    Typical Use Case : predicting life events from behavioral data to anticipate structural changes in the client's consumption profile that may signal churn or upsell / cross-sell opportunities