Project Summary

AXA, a global insurance leader with operations in over 50 countries, initiated “Data for the Future” in 2020, an enterprise-wide analytics and AI transformation. Facing threats from insurtech startups and changing customer expectations, AXA aimed to evolve from a reactive insurance provider to a proactive risk partner through data-driven decision making and AI-powered customer experiences.

Solutions Adopted

AXA implemented a multi-layered analytics and AI ecosystem:

  • Google Cloud Platform is the primary cloud provider for AI/ML workloads.
  • TensorFlow and PyTorch for deep learning model development.
  • DataRobot and H2O.ai for automated machine learning.
  • Dataiku for data science collaboration and model management.
  • BigQuery for cloud data warehousing.
  • Looker for enterprise analytics and dashboards.
  • Computer vision systems for claims assessment automation.
  • Natural Language Processing for Policy Document Analysis.
  • Reinforcement learning for dynamic pricing optimisation.
  • Explainable AI frameworks ensuring regulatory compliance.
  • Enterprise knowledge graph for entity relationship mapping.

Implementation Costs

  • Cloud infrastructure and platform: €48 million.
  • Data science and AI tools: €24 million.
  • Custom model development: €32 million.
  • Data engineering and integration: €30 million.
  • Talent acquisition and upskilling: €22 million.
  • External expertise and consulting: €18 million.
  • Change management and adoption: €14 million.
  • Governance and ethics frameworks: €7 million.
  • Total investment: Approximately €195 million.

Implementation Duration

  • Strategy and planning: 5 months (January-May 2020)
  • Platform and foundation building: 7 months (June-December 2020)
  • Centre of Excellence establishment: 4 months (October 2020-January 2021)
  • Use case implementation phases:
    • Claims optimisation: 9 months (November 2020-July 2021)
    • Pricing sophistication: 11 months (February-December 2021)
    • Fraud detection: 6 months (May-October 2021)
    • Customer churn prediction: 5 months (August-December 2021)
    • Risk assessment: 8 months (January-August 2022)
    • Customer experience personalisation: 10 months (March-December 2022)
  • Scaled enterprise adoption: Ongoing (2022-present)
  • Total initial implementation: 3 years (January 2020-December 2022)

Savings and Benefits

  • Claims processing costs reduced by 28% through automation, saving €145 million annually.
  • Fraud detection improved by 42%, preventing €85 million in fraudulent claims.
  • Customer churn reduced by 18%, retaining €210 million in annual premiums.
  • Underwriting accuracy improved by 31%, optimising risk assessment.
  • 22% reduction in claims settlement time (from 18 days to 14 days average).
  • Dynamic pricing optimisation increasing policy profitability by 8.5%.
  • Customer satisfaction scores improved by 26 points for digital interactions.
  • Predictive maintenance models for commercial property clients reducing claims by 24%.
  • AI-powered health coaching reducing health claims by 15% for participating customers.
  • Total annual value creation: Approximately €560 million.
  • ROI achieved within 22 months of initial investment.