Becoming compatible with analytics is perhaps one of the most important evolutionary steps an organization can take. The rewards of data-driven decision making can be a powerful boost to results. For insurance companies, this can include using predictive underwriting models to increase profitability through more granular pricing, resulting in a six to eight point reduction in loss ratios. On the claims side, predictive models have helped insurers better segment and sort workers’ compensation claims and high severity bodily injury, resulting in a four to 10 point reduction in claims expenses.
An important part of the analytics journey is overcoming the many challenges an organization faces as it experiences the end-to-end development and deployment of predictive models. Model development (e.g., data evaluation, data acquisition, data cleansing), scoring engine development (e.g., scoring engine and database design, development, testing, deployment) and implementation business (eg, strategy training, change management, business measurement tools) are some of the common questions organizations should consider.
Some of the concerns the authors observed in developing and implementing advanced analytics include:
1. Executive property
Without top management buy-in and a clear business strategy for integrating predictive models, advanced analytics efforts can get stuck in model development. To be effective, analytical efforts must involve key leaders who can help drive acceptance and change throughout the organization. Senior leaders should emphasize that there is a clear correlation between the actions to be taken through the implementation of the model and the expected business benefits to be realized. Without accountability for a targeted return on investment, organizations risk spending a lot of time “doing” rather than “getting things done.”
2. IT involvement
Failure to involve IT early on in the analytics journey can lead to significant issues along the way if technology gaps and limitations are not understood up front. Modelers can find a way to access internal and external data, but without the help and involvement of IT, it’s nearly impossible to bring models to life in the day-to-day functioning of the organization.
3. Production data available versus cleaned modeling data
Accessing historical data for model development is very different from accessing real-time data in production, and a strong model is only good in its ability to be practically implemented within the business. technological infrastructure. Real-life limitations may restrict the data available for historical modeling. Sometimes a proxy variable can be used for modeling until the data is available. Analysis initiatives often risk being thwarted by the belief that the data for modeling must be perfectly clean and organized. The development of predictive models is not an accounting exercise, but rather a statistical process where there are many techniques to remove “data dirt”.
4. Project Management Office (PMO)
Lack of clear ownership of the end-to-end journey is a common stumbling block for organizations that have struggled (and failed) to implement their predictive models. Without the right project management structure in place, a clear cadence of project milestones, and ownership of deliverables by pre-identified business owners, the project could be doomed to failure before it begins. Most importantly, the PMO must be able to connect with all interested parties and take an agile approach.
5. End user participation and buy-in
Lack of end-user involvement in the planning, design and final deployment of predictive models can hamper efforts. For underwriting or claims models, it is essential to involve underwriters, marketing, actuaries, adjusters, nurse case managers, and Special Investigation Unit (SIU) resources from the outset. start of the process. End users also have a better understanding of the business process and may be able to better identify potential gaps or obstacles to successfully integrate models into day-to-day operations. If end users feel they have an interest in deploying the predictive model, then the business is more likely to realize the potential financial benefits. If done correctly, some of the early skeptics may eventually become supporters of the analysis.
6. Change management
Organizations often fail to understand how predictive models change current business and technology operations – policies, procedures, standards, management metrics, compliance guidelines, etc. Without the proper design, development and deployment of training materials to address relevant audiences in the field and in the home office, the analytical journey can come to an abrupt end. It is important to educate end users and other relevant stakeholders on how the model will be used on a daily basis and how their lives can change. A communication plan should be developed to answer frequently asked questions (FAQs), address common concerns, and help end users appreciate the organization’s strategic vision. Change management does not start and end with training; it starts from day one and lasts well beyond the deployment of models.
7. Explainability vs. “the perfect lift”
It is important to balance building an accurate statistical model with the ability to explain the model and how it produces results. What good is using a nonlinear model or a complicated machine learning method if the end user has no way of translating score drivers and reason codes into actionable business results? Experience shows that a less complex statistical model development method yields similar results to more complex approaches, and a small sacrifice in predictive power can lead to a marked improvement in the explainability of technical model recommendations for models. end users.
The size of the insurance company matters
Every business is different, and designing a successful approach to implementing predictive models in the business process can vary widely. What works for a large national insurance company may not work for a regional mutual and vice versa. Some differences include:
1. Communication and training
When it comes to change management, large organizations can struggle to execute communication and training plans in an organization that is often siled, functionally diverse, and geographically dispersed. A major challenge is to provide tailor-made communication and training to meet the business processes of multiple stakeholders with different expectations and office cultures. Large businesses may have to balance corporate communication protocols while small businesses may not have the same challenges.
Small businesses may underestimate the effort involved in managing change and may downplay the importance of multiple communications, gain buy-in from stakeholders, and organize effective training. There is also the possibility of being over-stretched to provide the right level of dedication and focus, as they are involved in other aspects of the project in addition to their day-to-day responsibilities.
2. Competing initiatives and system challenges
Large organizations may also face additional challenges when working on new and existing systems. It is extremely important to involve technical and business experts (SMEs) in the model creation and implementation phases to get a complete view of the current and target state of each affected business process. Large organizations may also find themselves juggling multiple initiatives at once (for example, deploying a new claims management or policy administration system as well as predictive models), which can be a challenge for allocation. Resource. Another consideration is how predictive analytics and other initiatives will influence each other. When related projects are underway, a future state of multi-phase analytics, starting with a semi-integrated predictive analytics solution, is a common approach to bypass initiatives that may be affected by predictive analytics.
3. Resource constraints
Large companies may also face difficulties in obtaining the right resources, and in particular the decision makers, involved in the process. During the model building phase, it is usually easy to identify who needs to be involved (eg, analysis team, actuaries, data SMEs). Making decisions on how to implement can be more complicated. These decisions should involve business unit managers and, in some cases, senior management – busy people who may not have the dedicated time necessary for discussions and implementation decisions.
While it may be easier for small businesses to involve business unit managers and even executives in a predictive analytics project, they can suffer from having too many people involved because project managers may not want to leave anyone out of the decisions.
4. End-user buy-in and use of the model
Getting the most out of predictive analytics is a challenge typically encountered by all businesses, but it can be more complicated in small businesses due to the sensitivity of expectations regarding the use of the model. While some companies tend to view a predictive model as an additional tool to help make better decisions, the degree to which employees are expected to follow the model can vary widely.
In small businesses, employees may fear that their roles will be replaced by predictive analytics. Models are generally not designed to replace people, but rather as useful tools for making more informed, objective, and measurement-based decisions. This balanced approach and intention must be communicated continuously to all project participants as a basic philosophy essential for organizational buy-in and effective change management.
As Larry Winget said in It’s called work for a reason, âKnowledge is not power; the application of knowledge is power. The successful creation and deployment of predictive analytics goes far beyond developing a highly predictive model and robust scoring engine technology. It is a holistic business that requires a focused effort on practical implementation within business operations as well as organizational, people and process considerations for what is most effective for end users and the business culture. enterprise in the broad sense.