Aviva: Powering Insurance With Data, Machine Learning, & AI

In the past, Aviva had no integrated analytics platform. The data stores were siloed, analysis was run on on-premise commodity hardware, and it was underpowered — not to mention expensive. Very few people could use any data at any scale, and different parts of the organization used different technologies, adding to their analytics challenges.

Today, Aviva has a standardized, centralized, and modernized infrastructure for data and analytics. The most important part of this is the Dataiku workbench, providing consistent connectivity to all the source datasets, a uniform platform to operate analytics functions, appropriate governance and controls, and standardized deployment of models and workflows to production.

“The most beneficial thing about Dataiku is having everything in one place, so you don’t have to go from one program to another to another and have them work all at the same time. Dataiku takes away that hassle.” Ayca Kandur Data Scientist at Aviva

In the past, deployment on their antiquated platform could take multiple weeks versus an overnight deployment in Dataiku. The change means a nearly 98% reduction in lead time for deployments. Dataiku has also brought dramatic improvements in working practices by removing silos and allowing data analysts, data engineers, data scientists, and IT teams to work together in the same tool set. This led to a 75% improvement in time to market from idea to production.

Data, Machine Learning, & AI Use Cases at Aviva

In the past five years, the IT platforms team at Aviva has gone from major transformation projects to improving the infrastructure, delivery, and availability of Dataiku. This allowed Aviva to roll out Dataiku to an ever-growing audience of users, from around 10 data scientists to roughly 250 data scientists and 2,000 data and AI consumers. 

This, in turn, enabled widespread consolidation of tooling used in Aviva that produced sizable savings in licensing costs and hosting infrastructure costs, plus gains in the efficiency of support that comes with a reduction in tools used across the organization. 

Here are just some of the data, machine learning (ML), and AI use cases in play at Aviva, built with Dataiku.

Improved Process Analytics & Auditor Efficiency

The 100-employee-strong Aviva Audit Team was looking to provide a smooth and efficient customer experience without sacrificing the accuracy of the audit process. With the help of Dataiku’s plug-and-play process mining solution, the team now has a more real-time, play-by-play relationship between data analytics and the business. They have a full view of the processes and can focus on areas that might be broken or might not be running as expected.

The value added by process mining has been significant for both analysts and auditors. On the analyst side, the greatest benefit has been in terms of time saved — cutting their time spent by about 50%; whereas for the auditors it has been in terms of the quality of the output, allowing them to quickly focus on what matters. 

And that’s not even counting the accelerated initial set up: it took two days for data analysts to get up and running with the plug-and-play process mining solution on Aviva data — compared to the nine to 12 months it would have required to set up their own solution.

動画を視る

Improved Process Analytics & Auditor Efficiency

The 100-employee-strong Aviva Audit Team was looking to provide a smooth and efficient customer experience without sacrificing the accuracy of the audit process. With the help of Dataiku’s plug-and-play process mining solution, the team now has a more real-time, play-by-play relationship between data analytics and the business. They have a full view of the processes and can focus on areas that might be broken or might not be running as expected.

The value added by process mining has been significant for both analysts and auditors. On the analyst side, the greatest benefit has been in terms of time saved — cutting their time spent by about 50%; whereas for the auditors it has been in terms of the quality of the output, allowing them to quickly focus on what matters. 

And that’s not even counting the accelerated initial set up: it took two days for data analysts to get up and running with the plug-and-play process mining solution on Aviva data — compared to the nine to 12 months it would have required to set up their own solution.

“Process mining allowed us, on the analytical side, to increase efficiencies and deliver the work in half the time and, on the audit side, to increase the quality of the output, because they are able to zoom in on the potential discrepancies or deficiencies in the customer journey.” Francisco Pavao Martins Senior Manager of Data Analytics at Aviva

Meet ADA, Aviva’s Algorithmic Decision Agent

Aviva also leverages Dataiku to drive its world-class customer data science team. ADA is Aviva’s algorithmic decision agent, a customer-first AI that is powering omni-channel, hyper-personalized marketing. ADA is a supervised ML model ensemble that uses Aviva’s customer data and XGBoosting methods to provide predictions of customers’ next best actions.

Not only does this project showcase the power of data to provide predictions of the future behavior of customers, it also demonstrates the full life cycle of a fintech project, from inception to design and delivery of an MVP. The team worked on extensive A/B testing, working with stakeholders across the business to ensure ADA becomes widely used in all aspects of the business, rather than simply being another data science curiosity that gathers dust on the shelf.

“When we started building ADA … it was taking us quite a long time with the existing legacy systems. But through using Dataiku and the API functionality, we reduced the amount of time from beginning to end to build a model and push out the model into the marketing channel.” Ayca Kandur Data Scientist at Aviva

Sentiment Analysis & Topic Categorization for Customer Feedback

For a customer-first organization like Aviva, deeply understanding customer pain points to make informed decisions and enhance the online experiences is critical. However, traditional methods of manual analysis are usually time-consuming, error-prone, and lack scalability. This can lead to missed opportunities in understanding feedback themes, identifying emerging trends, and thus addressing pain points for improving customer satisfaction.

Aviva used Dataiku to power their ML-based solution for customer feedback, which features:

• Seamless ingestion of survey data from an Amazon S3 bucket.
• Preprocessing of feedback data to consolidate multiple survey sources into a unified and coherent view.
• Advanced natural language processing (NLP) capabilities to extract meaningful insights from feedback data.
• Development and implementation of a sentiment analysis model.
• Creation of a themes categorization model that categorizes themes or topics.
• Integration of output from above models resulting in powerful analytics and well-rounded view of customer feedback.
• Intuitive and visually appealing dashboard to provide actionable insights to stakeholders.

Dataiku reduced the need for manual effort to analyze customer feedback data, reducing the time to generate weekly reports by at least 50%. The associated cost savings of this solution alone is approximately £10,000 per month.