While Generative AI continues getting adopted along the value chain of companies across verticals, use cases and challenges become more and more specific in terms of customer, industry and business function requirements. While end customers are getting used to working with GenAI tools in everyday life, expectations towards digital interactions with companies and general service quality increases significantly. Studies suggest that across age groups, just a fraction of customers (<15%) are satisfied with the current digital offering of their insurance provider.
When considering these two developments of increasing specificity of GenAI use cases and, on the other hand, rapidly increasing end customer expectations towards digital offerings and interactions, this begs the question of how GenAI can be utilised to enhance and establish a future proof (digital) customer journey. One of the first challenges that usually comes up when considering to implement and utilise GenAI for generating incremental value is where to start. What makes this decision particularly hard in the context of AI is the fast pace of innovation and broad availability of different services and technologies.
In order to tackle this first obstacle, the below overview provides guidance by mapping GenAI core technologies and expected business impact against the typical customer journey in insurance.
The resulting heat map provides three key takeaways:
- From ‘Awareness’ no ‘Advocacy’, there is no step along the customer journey where GenAI is expected to have no positive impact, which emphasises the importance of developing strategies to effectively utilise the technology
- Today, the core technology promising the biggest impact across the entire customer journey is conversational agents.There is a wide variety of possible applications ranging from external agents helping clients to find the most suitable product to internal agents used by customer service to easily search and analyse complex policies.
- Later stages of the customer journey, ‘Conversion’ and ‘Retention’ in particular, are the most promising starting point to identify high value use cases. A key reason for this is that these stages offer room for internal use cases that lower the bar in terms of regulatory requirements.
Once the focus is narrowed on a specific domain, there are several questions that have proven to be a valuable next step in order to develop a structured approach towards implementing GenAI use cases in general but particularly in a regulated industry such as insurance:
1. Clear use case definition: While a precise definition of the objectives and scope of a use case is critical for most data driven projects, it is especially important when implementing GenAI technology as even within a defined area of the business (or stage of the customer journey), usage can still be broad. When taking conversational agents as an example, usually the quality of answers increases when limiting the use case to specific types of questions or scenarios. Some questions worth to consider when defining use cases are:
- What is the overall objective of the use case, how does it deliver value to the business?
- How is success going to be measured, what are relevant KPIs?
- Who are the end users (can they be categorised to narrow the use case and increase quality)?
There are different workshops available representing best practices for use case definition, including Keepler’s Navigator workshop or AWS’s D2E program.
2. Technology Choice: As the availability of new GenAI services and technologies evolves on a daily basis, keeping an overview and making long term decisions is a significant challenge. In this context however, making no decision is worse than making a couple of wrong but reversible ones, which is why maximising flexibility and following a ‘best of breed’ approach proves to be best practice. Zooming in, the below questions help making choices regarding technology:
- How should the use case be integrated into existing infrastructure?
- How will the use case be consumed?
- Are there future iterations / extensions that can be anticipated already?
- What are the pros and cons of ‘make or buy?
- How to establish a future proof foundation beyond a specific use case (from technological and organisational perspective) for GenAI adoption?
3. Data Security and GDPR compliance: Concerns around Data Security and GDPR can be a showstopper for most GenAI use cases. To avoid this, both topics need to be considered early on. Two key actions usually help to establish a secure and compliant foundation for building the use case:
- Involve relevant functions (GDPR officer, IT security officer,…) early on, ideally when developing the use case idea
- For most GDPR and security concerns there are suitable answers and solutions in place. For example, there are template architectures and other best practices available to process PII data in a compliant way in the cloud (see e.g. Meet European Data Security and Privacy Compliance with Big Data Analytics in Public Cloud)
4. Building an organisational framework for GenAI use case implementation: Successfully adopting GenAI to implement use cases at scale is as much a technological endeavour as it is an organisational one. Successful implementation does not start with delivering a technical solution, rather it starts with preparing an organisational foundation for establishing a ‘use case factory’ that continuously delivers GenAI data products to increase business value. This requires key business functions and stakeholders to be involved from the beginning, ideally in a defined and process driven way. The below thoughts can help to trigger helpful mechanisms:
- Which business- and organisational functions (workers council, IT-Security,…) need to be involved when implementing GenAI technology?
- What questions need to be answered to “greenlight” a use case?
- How to ensure organisational adoption of the final solution (internal communication)?
- How is successful adoption of a use case defined?
5. Data Quality aspects: Answering the question of whether current data quality is sufficient to generate the expected business value is inevitable when evaluating GenAI use cases. On an individual use case level, experimentation can be an option to move ahead quickly. If following a more comprehensive approach, utilising existing services and best practices for measuring data quality and organisational maturity on a more holistic level can help (see Data Quality Assessment).
6. Typical costs, team setup and required skills: In order to greenlight use cases and ensure broad support of the organisation, answering effort and cost related questions upfront is key. Requesting estimates from different services companies can help to get first indications. In practice however, GenAI use cases tend to be highly specific which is why working in an explorative and iterative way with (external) experts helps to develop reliable effort and cost estimations, eventually enabling a more precise evaluation and planning.
It remains to be seen which stages of the customer journey are eventually going to benefit the most from GenAI technology in the long term. With advancing core applications and evolving organisational maturity, it is likely that the impact of GenAI in early and late stages of the customer journey is going to increase as well.
Florian Sieren
Manager Business Development. "In my current role at Keepler, I'm leading the business development and sales activities in the DACH region. We are helping companies to turn data into business value by developing public cloud based data products"
0 Comments