Business Intelligence in Insurance

//Business Intelligence in Insurance

Business Intelligence in Insurance

Written by Hanan Taiber, CTO, and Uri Taiber, CEO, InsFocus Systems

About Business Intellignce in insurance

So what sets insurance business intelligence apart from any other vertical? I believe the answer lies in the complexity of the data modeling, on one hand, and the relatively small data sizes, on the other.

Insurance data tends to be very complex – a policy can be split to insured objects, covers and risks. Every product type has different risk parameters. Claims are events that carry on payments, outstanding estimates – future payments, and IBNR – incurred but not enough reported damages. Moreover, complex organizational structures, product structures and reinsurance data make the data even more complex.

However, data sizes are less of an issue. A client typically has a few policies on which he makes one or two transactions a year. A telco client, on the other hand, may create one transaction per minute!

Sp, in order to build a good reporting & analysis solution for an insurance company, one must use highly trained personnel in insurance terms.. When building the data warehouse and choosing the technology, I would emphasize:

  • Complex calculation capabilities – not just your normal slice and dice, but advanced calculations such as earned and unearned premiums, outstanding claims, reinsurance cessions, IBNR and IBNER.
  • Optimized data model – don’t just start with a normal relational data model, but be sure the data model includes all insurance calculations beforehand.
  • Data quality – admin system are typically old and complex. Data tends to look ugly. Pay careful attention to data cleansing and handling of data quality.
  • Ad-hoc capabilities – insurance personnel like running ad-hoc queries even from the days they got paper reports. Make sure the solution allows complex ad-hoc queries, and not just fixed reports and slicing and dicing.
  • Drill-down capabilities – statistics are cool, but a single claim event can alter the whole statistics. Make sure a user can drill-down from any level and to any level in a report.
  • Advanced filtering – in insurance, one often finds the need to ask “show me X but ignore all claims above Y”, or “show me X for all clients that pass condition Y”. Make sure the system supports such ad-hoc pre-filtering.

Why is InsFocus BI better suited for insurance companies’ analysis and reporting needs?

When we set out to build InsFocus a few years ago, we found there is a fundamental shortcoming in the way reporting and analysis is done in insurance companies. The main problem was that there was a clear separation between the tool and the model. Thus, a company would go into a process of choosing a tool that looked fit for the purpose (usually Cognos or BO). As a second phase, the company will hire professionals or use in-force persons to build the data model and the ETL processes.

The outcome, in many companies’ view, is disappointing. The tool, that had looked very nice on “demo” data (which is usually simple retail data), suddenly looks rigid and not flexible enough. The model, that is developed by BI professionals, misses the sophistication of insurance. Then, the company goes into endless cycles of data model improvements and enhancements.

So why do we believe that InsFocus BI is better? For a few reasons:

  1. Built-in data model – InsFocus BI ships with a comprehensive insurance data model, covering most insurance aspects available. The model includes earned and unearned premiums, deferred acquisition costs, outstanding claims, IBNR calculations, and a lot more. So basically, we know the end result before even starting. This alone saves tons of executive time, BI developer time and saves the company from making a lot of mistakes.
  2. A tool that is built for insurance – InsFocus BI is not just a shallow data model, but a sophisticated tool that is built specifically for insurance. So, unique insurance techniques, such as changing calculations according to the presented time base (e.g. underwriting, accounting), showing triangulation views, calculating run-off using claim filters, finding the sums insured at a specific snapshot date, and many many more, are already built into the tool.
  3. Granular reporting and advanced filtering – a unique characteristic of insurance reporting is the figures distortion that can be caused by a single event – one big claim can distort an agent’s full record if looking only at top-level averages. To support insurance analysis, InsFocus BI provides two important capabilities:
    1. Granular reporting and advanced filtering – a unique characteristic of insurance reporting is the figures distortion that can be caused by a single event – one big claim can distort an agent’s full record if looking only at top-level averages. To support insurance analysis, InsFocus BI provides two important capabilities:
    2. Pre-filters – InsFocus BI allows easy creation of pre-query filter on the measurements themselves, enabling analysis such as “show me profitability ignoring claims above X” or “show me a list of the top biggest claims in the last 1 year”.
  4. Ad-hoc reporting – generic solutions have traditionally split users into two superficial types. One type is the so-called “simple user”, who presumably is not computer literate and only wants to see ready-made reports. The second type is the “power user”, who wants to do slicing-and-dicing and maybe alter some reports. In InsFocus, we decided we won’t do not agree with this distinction, and we enabled ad-hoc querying for all users.
    This is especially useful as insurance information workers need ad-hoc querying, and a lot. New reporting needs arise almost daily – and going to the IT department for every new report is a huge waste of time and resources. Using InsFocus BI, we enabled all information workers to perform their own analysis using a very simple to use interface.
2018-09-14T14:20:04+00:00