Capturing Value from Data in Insurance: Identify, Explore and Evaluate Proportions of Value with Data-Driven Use Cases

Introduction

Coming back with an interesting topic around implementing your data strategy. Most likely a lot of businesses defined their data strategy including vision, objectives, value tree, trade-off decisions, direction for next years, business roadmap, and operating model. Next step is to find specific portions of value contributing to the objectives, prioritize and realize them.

Data & Analytics services are important for service businesses and therefore information-driven businesses as they can improve decision outcomes for all types of decisions (macro, micro, real-time, cyclical, strategic, tactical and operational). Still realizing value from data captured along the customer journey and related company-wide processing remains a major challenge for insurance firms.

Today we are taking a quick look into the insurance business and lay out one possible approach towards implementing your strategy to capture value from data processing. The data strategy itself (processed before or progressing in parallel), operating model and (technical) foundations of the data platform to be setup in parallel are not in scope here.

The five-step approach below summarizes a way towards capturing value from data based on business-driven use cases and provides the anchor for our efforts. It should help to work on concrete results requested from business stakeholders and link these to strategic dimensions.

(1) Revisiting the strategic objectives around growth, risk and efficiency, and related key decisions taken during the strategy phase to clarify the go-to-target and metrics. The objectives in the three dimensions are building the foundation to link the cases to value.

(2/3) Potential data-driven use cases are identified and explored, afterwards summarized and evaluated against key dimensions of e.g. value and complexity/effort which support prioritization for execution planning. More in the following.

(4) The use cases and related enabling activities are planned towards a data & analytics roadmap covering short-term specifics and long-term directional plan incl. timeline, resource indication, responsibilities and expected deliverables.

(5) After respective approval prioritized use cases are executed, meaning iterated and scaled.

In parallel there is a need to run enabling activities to build and operate the data platform (technology), motivate and scale talent (people), establish standard routines (process), and clean-up, structure and process information handled (information).

The approach should not be taken as a strict sequential route, more a helpful structure and pathway; defined and confirmed business use cases might directly go to MVP implementation serving as a speed boat.

Going forward we focus on step 2 and step 3 of the approach to focus on identifying and evaluating data-driven use cases.

Identify and explore potential data use cases

Information and underlying data processing are at the heart of insurance businesses providing risk protection in form of a guarantee (contract) of compensation for specified loss or damage in return for payment of a specified premium. We set aside the big chunk investment management part and side hustles of insurance companies. These could be analyzed separately according to their value chain.

A value chain perspective supports to analyze key business capabilities or high-level processes orchestrated to deliver your services (via a value stream), aligning discussions around where data is processed within the firm and identify and explore valuable data-driven use cases. It provides a high-level overview and structure from customer touch points via marketing and agency sales activities, along risk analysis and actual processing, and supporting functions.

Selected data-driven use cases along the Insurance value chain are outlined below, using different types of analytics (descriptive, predictive and prescriptive) based on analytical data, operational data and meta/master data processing.

How could we use the value chain overview in our planning process? See three steps below:

Collect ideas, scenarios and use cases from various external and internal sources, compare them to existing solutions and desired target, and indicate first priorities – serving as a great foundation for alignments with various stakeholders across the insurance firm. Beside a rather broad analysis of existing data sources in areas of common interest might help to indicate data quality and usefulness of information for the value cases.

Drive workshops to understand current state and value efforts, risks and collect ideas. Depending on the product offerings – the value chain might be specific for different areas of the business like health, car insurance etc. based on strategic discussions with stakeholders. Key is to be well prepared for the workshops, meaning setting agenda, selecting key circle to discuss, and work towards clear outcomes.

Process (tech) demos and PoCs support to understand and explore the use cases in a "real" environment compared to conceptual perspectives before. If the value is clearly seen (and confirmed) teams could already start with an MVP to iterate towards a valuable data product (fast track). Supporting to try, fail and learn early is the mindset here.

Coming back to the data-driven use cases - in the following three examples, each addressing mainly one business value dimension, are summarized – these are to be tailored and detailed out for specific context (more technical service related documentation covered in sources):

1 Campaign Optimization (growth): Using a model to predict actions maximizing the purchase rate of leads targeted by a campaign. The ML model uses historical campaign data to predict customer responses and recommend when and how to contact leads. Recommendations could include best channel (e.g. mail, SMS, call), day of week and time slot to contact leads.

2 Actuarial risk analysis and financial model (risk): Standards and regulations like IFRS17 require actuaries to use compute-intensive techniques when modeling assets and liabilities. Much of the analysis will make use of stochastically generated scenario data from separated inputs. Beyond regulatory needs, actuaries process a fair amount of financial modeling and computation.

3 Automated document processing (efficiency): Automated document processing supports to seamlessly acquire, detect, enrich and feed information from different types of documents. Natural language processing (NLP) models and custom models enrich the data.

Summarize and evaluate use cases

How could we use the proposed use cases and PoCs? Three more steps below:

Consolidate the use cases to achieve a common understanding of each one by capturing them in an agreed standard format covering key information (e.g. value contribution, description and key steps, producer/consumer, key information processed, data capabilities required, performance metrics).

Key point here is a balance to be one the one side realistic to understand the data existing from a quality perspective and if it is helping to answer the questions coming with the use case – if the data is not existing currently being realistic about efforts for generating it or sourcing it from external partners. On the other side will aspirational thinking and selected ambitious cases support to retain the vision, drive the efforts and capture real value, not just small improvements to daily business activities.

Evaluate the use cases towards their priority for the Insurance firm – based on key dimensions covering estimated value vs effort to implement.

Value: estimated quantitative (and qualitative) impact of the use case on revenue, risk and cost base

Complexity/Effort: estimated resources and time frame required to execute the use case, related to e.g. technical, human resources

Adding colors of cases could highlight the value dimension of each use case around growth, efficiency and risk. The four quadrants indicate characteristics (daily business optimizations, big bet, low hanging fruit, tough cookies) supporting to decide which use case to prioritize and execute.

Decide for first prioritized use cases moving to execution phases, some might be worth analyzing in more detail and covered in a later wave. Regular pragmatic decision board session support to fill the pipe for implementation.

Upcoming steps and summary

Next steps would take you towards putting these on a ambitious, yet realistic roadmap (covering quick wins, short, and long-term flow towards the objectives) and setup an implementation in close iteration with key customers / internal stakeholders, on a larger scale industrializing use case management with a factory approach to address the common problem to bridge from testing / PoCs to real production. In addition the use cases should be considered evolving towards data products with clear governance and documentation on e.g. data access / API, included data sets, transformations and database sources creating trust with customers relying on the information.

Summing up the key is to link strategic direction to the implementation of value increments via data-driven use cases addressing the value dimensions along the Insurance value chain - keeping in mind to move fast from conceptual to a try-it out phase. A foundational data platform and architecture needs to be setup and operated to support the analytical, operational and meta/master data processing.

In future posting we could take a look at both sides data trends & strategy and use case execution management, or the foundational structure of a data platform. Looking forward to your comments and ideas.

Note: This article reflects the private opinion of the author. Unpaid advertisement.

Sources

Accenture (2022) Insurance Blog - Future of Insurance

Accenture (2018) How to use AI throughout the insurance value chain, starting with sales and distribution

Gartner (2022) What is Data & Analytics? Everything you need to know

Gartner (2022) 12 Data and Analytics trends to keep on your radar

HBR (2022) Why becoming a data-driven organization is so hard

HBR (2014) Insurance companies untapped digital opportunity

Medium (2021) 5 best ways to prioritize your product backlog

Microsoft (2022) Campaign optimization, Azure Services documentation

Microsoft (2022) Actuarial risk analysis and financial modelling, Azure Services documentation

Microsoft (2022) Automated document processing, Azure Services documentation