Getting the best out of data

16-05-2018

Getting the best out of data

Meghan Anzelc, chief analytics officer, AXIS Capital

Many re/insurers understand the value and potential of data but leveraging its full potential requires a systematic approach designed to achieve very specific outcomes, says Meghan Anzelc, chief analytics officer, AXIS Capital.

How do you transform data into a powerful, profitable tool for your specialty insurance business? Everyone is aware of its potential, but converting the possibilities trapped within data, internal or external, into tangible business benefits is not straightforward.

Whether the desired outcome is to empower better decision-making, or improve business processes, or both, a systematic and creative approach will be more successful than one which deploys the newest technology simply because it is the current buzz, or attempts to accumulate vast external data sources without first knowing exactly what purpose they will serve.

The goal of any data-related project should be to solve a specific problem and deliver a quantifiably useful benefit. Therefore, the first stage in harnessing the power of data is to identify real-world areas where it will yield value.

The job begins with the formation of a combined data and analytics team to identify and create analytical tools which will benefit the business where it counts most: the bottom line. The initial step is for the team to engage with business leaders and their team members to understand their business strategies, challenges, and points of frustration.

It is also important to ask questions and collect information on how data is being collected and used and how the workflows are designed. The approach should be systematic, but at the same time it must be creative. Methodology ensures nothing mandatory is missed; creativity does the same for possibilities.

 

Making life easier

The job, in short, is to examine processes and data usages to identify ways they could be improved and that also solve existing challenges or points of frustration. It can be as simple as asking people how they use information and how information informs and improves their work, and then thinking about how that could be made easier.

An important component of this process is to involve the ultimate end users of any data analytics solution that is developed. This helps alleviate anxieties over data usage and achieve the buy-in that is essential to ensuring that any data analytics project yields the maximum benefits. In an ideal world, a strong partnership will exist between the end users—often actuarial, claims and underwriting teams—and the data analytics team, ensuring tight alignment of objectives and use cases and a continuous feedback loop throughout the project development and after implementation.

A new data analytics tool or programme is only as useful as the benefit its users derive from it. It’s important, therefore, that your data analytics team is in constant communication with the end users to gather their input, ideas and feedback.

The result is a list of data analytics projects, each of which could be executed to deliver an anticipated benefit. Some will be discarded immediately because they are impossible—the data may not exist, for example. Others may require a change of business process that simply isn’t feasible.

Projects which remain should then be assessed on a cost-benefit basis. Some will be quick wins; others may require the medium-term dedication of time and resources. Either way, the data analytics team should choose to pursue only those projects which present the best business case.

The earliest projects should target getting the most from existing internal data. This could involve, for example, analysing existing data and reports to help bring forward actionable insights from the data. Rather than just reporting the numbers, which often entails large volumes of charts and tables, partnering with the user group and IT, to better design user interfaces to draw attention to key points or worrying trends, can add value to the business.

A second internal data strategy is to release data trapped in electronic formats such as unstructured text fields and PDF documents. This can hold many valuable insights, but is often left out of the equation because the people who could benefit from the knowledge either have no idea it exists or cannot readily gain access. This could include, for example, the story of a claim: what happened and where.

When stripped out, structured and brought into a coherent data analytics system, such information can, for example, help management to understand the key drivers of claims and their severity, or support reserving and underwriting. Creating such a tool is complicated, but the business benefit is potentially sufficient to make it worthwhile.

 

Avoiding pitfalls

Throughout a project, other relevant internal departments should also be consulted, ranging from legal and compliance to actuarial and finance, to ensure potential enhancements and pitfalls are revealed early.

Especially tight collaboration is needed between end users, IT and analytics: all should be involved in defining the problem, determining how to solve it and deciding if it is actionable. Several questions must be considered:

  • What would implementation look like, and how would it dovetail with existing processes?
  • When the solution adds tasks to a workflow, will it be necessary to remove others?
  • What will the solution demand in terms of ongoing maintenance?
  • What level of change management is needed?

Finding answers and pinning down ideal outcomes may also require the involvement of legal, finance, data privacy and security, compliance, actuarial, reserving, operations and the other business functions that may be impacted by data analytics solutions.

Execution requires committed resources. These include a data analytics team resource acting as a project lead. Some end-user-group resources must be allocated for development and feedback, from conception and inception to implementation. IT must be involved very early to manage implementation paths, outline limitations and potential interdependencies between systems, locate algorithm implementation, and deal with other factors, including decisions about software selection and use.

 

Finding the benefits

Before going down the path of investing in a new data analytics tool or programme, it’s useful to illustrate the advantages that may be derived from utilising a systematic, strategic approach to discover where data analytics solutions could yield business benefit. This may be as simple as asking end users: “What would make your job easier?”

Take, for example, a typical property underwriting team. It’s very likely that the team regularly consults geographical imaging when considering submissions, as this helps them understand the topology and other environmental factors that may impact the risk.

They may look up each relevant location using Google Maps, and will sometimes use additional resources supplied by external vendors, again looking up locations manually. They often do the same for existing renewal risks. Each investigation is a separate, time-consuming and highly manual process.

It’s quite likely that this property underwriting team would like to be able to source all geographical information and data—public, commercial and internal—from a single interface with just one enquiry, for existing clients’ risks and for new submissions. They would want a mapping tool that would package the output from all data sources in a user-friendly way to reduce the manual effort.

Relative to other potential projects identified in our data analytics audit, the mapping tool would need to deliver high levels of time-saving compared to its cost. It would have potential underwriting benefits, too, since its end users would be more likely to consult more geographical data during decision-making.

The underwriting team, therefore, is far more likely to carry the idea forward and be its internal champion, rather than raising roadblocks and red flags from the outset. They would feel that the mapping tool has a clear and specific benefit to their work, rather than something that has been thrust upon them by management.

With the tool in use, the team’s underwriters would spend less time searching, and would make strategically better portfolio decisions, because they could see everything in a single place.

This example, which would represent a relatively small win for most specialty insurers and reinsurers, would be the result of a creative, systematic approach to the identification and implementation of data analytics projects which will yield genuine business benefits.

When multiple projects are completed across the firm, the cumulative impact is tangible and significant, enabling the company to remain relevant and competitive in the ever-evolving specialty re/insurance market.

AXIS Capital, Big Data, Meghan Anzelc, Business Development, Global

Intelligent Insurer