A User-Centric Approach to Insurance Analytics Solution Design

Jean-Stephane Faubert
Insurance Experts' Forum, April 21, 2014

(This is the first blog in a two-part series.)

In this series we describe an important principle that, when followed, ensures the successful outcome of any insurance analytics project. In it, we propose that an analytics project should be developed from the point of view of, and in close partnership with, its target business audience. We describe a four-step process for designing analytics solutions centered on end-user experience, and explain how the business community should be involved in every step.

Getting started

Before going deeper into the subject let us examine the four basic components of any analytics solution:

1. The operational systems that contain the source data. These systems are not part of the analytics solution itself, but they provide all the information needed for decision-making.

2. The integration layer that holds a consolidated view of the operational data. The integration layer houses a unified and historical view of data — often referred to as “a single version of operational truth”. It is often is implemented within a single database, called a data warehouse.

3. The presentation layer. This layer hosts all the information needed by a business, in a form suitable to address its information-consumption patterns. The presentation layer is structured differently than the previously mentioned integration layer. The purpose here is to support the business’ analytical and reporting processes. It usually is implemented as a set of data marts. These are the most important constructs of the whole design. Their structure determines the end-user experience.  

4. The access layer. This gives the business access to the presentation layer though different tools while enforcing security measures. End-users gain access to information in the underlying presentation layer through the access layer. In most cases, this layer is completely transparent to users, offering direct access to presentation layer content.

The most important layers

As already stated, the two most important elements in analytics are the integration and presentation layers. Of these two, the presentation layer is the most-critical because user experience depends on it. How the information is presented has a direct impact on how easily and efficiently business users can answer their analytical questions.

The main problem with many analytics projects is that the most emphasis often is placed on the integration layer. Such projects are driven as data integration projects. IT experts look at what is available in the source systems, and then bring it into a data warehouse or directly into data marts. Finally, they put an analytical tool of their own choosing on top of it. All this is designed without the collaboration of the business, and yet end-users are expected to be satisfied with the end results.

The main flaw with the data-integration approach is the assumption that the form in which the information is stored does not really matter, as long as it is there. It’s also important to note that end-users have little or no involvement in this kind of data-integration-driven project. And this is absolutely reflected in the final product: the semantic gap between what is offered and what the business needs is wide. An approach emphasizing data integration while ignoring the needs of end users makes for a suboptimal solution where refactoring already is required after the first implementation.

The right approach

We recommend that analytics projects should be designed, developed and elaborated from a data presentation point of view in close partnership with the business. This implies that the subject-matter experts should be involved in the project from the start and remain engaged throughout its duration. They actually should “own” the presentation layer data model, as it should speak their language and efficiently answer their analytical questions.

The presentation layer data model is the heart and soul of an analytics solution. Its design is not to be taken lightly, nor should it be designed as an afterthought to the other layers. Because it is what end-users see and work with on a daily basis, it is far more important than the data integration layer data model.

We propose a four-step approach for the design of the presentation layer data model, in close partnership with the business. This approach can be summarized as follows:

  1. Define the analytical domain
  2. Collect business questions
  3. Identify analytical patterns
  4. Design the analytical data model

In the next part, we will explore these four steps.

Jean-Stephane Faubert is a senior solutions architect at InEdge.

Readers are encouraged to respond to Jean-Stephane using the “Add Your Comments” box below.

This blog was exclusively written for Insurance Networking News. It may not be reposted or reused without permission from Insurance Networking News.

The opinions of bloggers on do not necessarily reflect those of Insurance Networking News.

Comments (0)

Be the first to comment on this post using the section below.

Add Your Comments...

Already Registered?

If you have already registered to Insurance Networking News, please use the form below to login. When completed you will immeditely be directed to post a comment.

Forgot your password?

Not Registered?

You must be registered to post a comment. Click here to register.

Blog Archive

The Internet of Things: Helping Insurers Make Better-Informed Decisions about Risk

The IoT is a major game changer for the insurance industry, and will likely affect every part of the insurance value chain. After all, insurance is data-driven, and that’s exactly what the IoT can deliver—relevant, actionable, real-time data that can provide an accurate picture of what is being—or may be—insured.

Coordinate Coverages to Manage Social Media Exposures

The bottom line is that no one policy will cover all the exposures in the social media realm.

Software-Defined Everything

What does it take to virtualize all the key components in your data center?

Competing with the Coasts for Tech Talent

Are heartland-based insurers at a recruiting disadvantage for tech skills?

On Thanking the Regulator … Really

The Financial Conduct Authority is demanding higher standards of consumer protection from insurers, which could lead to greater customer engagement and understanding.

Boosting Performance with Integrated Underwriting Tools

A unified, comprehensive platform can help underwriters perform their jobs more efficiently — and profitably.