1. Management summary

1. Having better and consistent access to the market, client, sector, project and team knowledge across the network of an insurance agent
2. demonstrating the clients that we are knowledgeable experienced and mondial and therefore that we have a competitive edge (supporting the customer focus and that we are a knowledge based insurance agent)
3. creating performance support tools to ensure tasks are performed efficiently and consistently to reflect best practice, creating potential to improve effectiveness and efficiency for an insurance agent in the OPTIMISTâ work process.
4. Pilot project nature of the system on behalf of an insurance agent

1. Terms of reference
0. Statement of scope
0. immediate Cause

1. Objectives

1. Project importance.

1. Principles

1. Project scope

 

0. Areas of changes

1. Dependencies

1. Euro

Not applicable, however, as knowledge and insights are generated and captured in the system on the Euro and the related Insurance products then these will enhance the capability and efficiency of the Insurance agent to work on Euro related systems.

 

2. Millennium

1. Implementation

 

0. Project phases and activities

1. Planning

 

Here will be mentioned the major milestones:
Project initiation: July 1998
Finalize Pilot design August 1998
Built redefine and review Beta version October 1998
Release built final definition November 1998
Release built production version December 1998
Release roll out January 1999

2. Quality Assurance

Quality standards and requirements will be established and published at the first project meeting.
Quality Management, including change control for the IT Build, will be subject to the practices of the IT-Build partner but will be agreed by the project manager and have been incorporated into the project plan and tracking cycle.

All other formal deliverables (these will be confirmed by the project manager during the initial detailed project planning phase) must be reviewed using a "structured walk-through" approach and any review comments acted upon before "sign off' by either the appropriate stream lead or the project manager.

The aim of the approach proposed is to provide adequate peer review of key design and roll-out the system quickly.

Exceptions - Exceptions to be above (e.g. making changes to either the design or pilot system) must be agreed on an individual basis with the insurance agent’s project manager and the OPTIMISTâ project manager.

 

3. Project organization

1. Project structure and Roles

1. Backgrounds and global directions for use.
1. Income category

For the definition whether someone belongs to a certain profile, it is important to know in what income category someone belongs. Within the boundaries of OPTIMISTâ this category is not important, in OPTIMISTâ it can be recorded for the determination of new profile definitions.

2. Housing

For the definition whether someone belongs to a certain profile, it is important to know in what kind of house he lives. Within the boundaries of OPTIMISTâ the housing is not important, in OPTIMISTâ it can be recorded for the determination of new profile definitions.

3. Paying conduct

For the definition whether someone belongs to a certain profile, it is important to know how he is used to pay. Within the boundaries of OPTIMISTâ his paying conduct is not important, in OPTIMISTâ it can be recorded for the determination of new profile definitions

4. Phone numbers,

A Client can be reachable on more than one phone number, this can be recorded in OPTIMISTâ , for each number can be fixed what kind of number it is, it seems meaningful to know whether it is a fax number or not.

5. Kind of policy

For specification of the products that the expert is creating it is necessarily to indicate what kind of policy it concerns, in this part of OPTIMISTâ it is possible to import all kinds of policies that are waiting for a special insurance product.

6. profiles

The expert previously defines profiles. A profile is no more or less than a expectation pattern: the expectation is that a person that satisfies to certain profile, is more than average interested in a certain insurance product. Take for instance the profile "life annuity" it is determined that persons that have an income over 100.000 per year and are over 50 years old can be considered for this profile. This means that a lot of persons that are marked with the "Life annuity profile", according to the expectations of the expert, are interested for a Life annuity policy. This means that when a mailing action for a similar product is planned, peoples with that kind of profile will get mailed, whereby the expert expects that the response will be great. Anyhow the percentage of response will be greater than when the mail is shipped to random and not selected addresses.
Profiles are previously determined by the expert and possible readjusted with help of Data Mining techniques, as long as it handles of simple products for a large public, it makes no sense to create profiles with any kind of technique. At the determination of profiles for very specialist products, mentioned for a small public, it could be meaningful to go at the population of addresses with Data Mining techniques, to find out whether there are unknown relations (listening to the database). The danger rests in the fact that too much effort will be spend at nonsense relations, the expert has to be very alert at this danger, at the other hand it could be very valuable because it is very well possible that there will be found very meaningful relations. Once when is defined what conditions have to be satisfied by a client, a profile like that can be allocated to him. To select all clients with a certain profile you have to start a simple query per profile (this query could be advised by Alice). Each client that is diagnosed to fit the query, get a characterisation mark in his data recorded. This action can be run several times with different queries for different profiles, with the result that a client could have several profiles allocated.
The fact that a client has a certain profile, only means that there is a trigger for the insurance agent to approach this client for certain insurance products, for the idea: this approach will not necessarily be supported by Information Technology resources. The fact that a profile has not been allocated to a client does not mean that this client not is entitled to buy such an insurance product, the initiative has to come from this client. At the other hand, it is not for a 100% certain that a client with such a profile will be accepted by the insurance company, the request has to pass through the normal acceptation process, just like each other client.
In this part of OPTIMISTâ the profile is no more or less than a name, given to a certain profile and a global description of it. If you want to know what the specifications are for this profile, you have to apply for the expert whom might have recorded the selection criteria.

7. Mailings.

After have passed through the activities of determination and allocation of profiles, with a simple selection assignment the addresses of clients that satisfies to certain profiles can be extracted from the database. All these clients will be sent the dedicated mailing for this product and this will be recorded in the file of the client, which means that per client is known what mailings have been sent to him.

8. Manipulator

All possible manipulators of OPTIMISTâ have to make known to the system, in this screen all significant data can be imported.

9. Sub agent

In the last part of this system, all data of clients and their offer’s will be sent to for this purpose selected sub-agents. These sub-agents are classified per postcode. In this part of OPTIMISTâ it is possible to change the data of these sub-agents. Next to the obvious data it is possible to specify the postcode area that is covered by this agent.

10. Products.

Just like the profiles here are just mentioned a name and a global description. The product has been composed by the expert, with external help there is made a prospectus and there is defined a special procedure for the tender of this insurance product.

11. Companies

All special Insurance products have to be offered by an insurance company. All data of these companies is recorded in OPTIMISTâ it is just to support the manipulator/operator at his work.

12. Client data.

At the moment the screen with client data is started, the prompt is already positioned in the area where the name of the client can be imported. By entering the first character of the name the scrollbar immediately shows all the names that start with that character in an alphabetical order. By entering the second character, the scrollbar shows all names that start with these two characters, this goes on with all next characters till the wanted name is on the screen. If there are more equal names, together with the phoning client it is possible to search at address so you have found the correct data belonging to this client.
While the correct client is selected it is possible to alter the discriminating data of this client. By dual click on the icon with a little book, at a glance you see the most significant data of the client: What insurance products this client already has effected, what offers are sent to this client, what profile this client meets and what mailings were sent to this client. Now this client has taken the trouble to phone it stands a good chance that he is interested in the received mailing. By a dual click on the rule belonging to the mailing, the operator gets the support concentrated on the mailing that is in picture. This can be a simple switch to a WORD blank form, a switch to an offer application or a link to a decision support system that is special made for this product.
It is obvious that the straightforward products not need much of special support, however when it depends a product that requires specific knowledge decision support can be of good help. Of course a special constructed decision support tool will be used in that cases where specific knowledge is required, now the expert is able to care about other things and another person with little knowledge of this product will help the client.
When all questions concerning the received mail are answered and there is made an offer, the operator can returning at the main screen again in a glance he can see what the most significant data is of this client. If appears that this client is allocated to more profiles, the operator will be able to activate some cross selling activities. If the client is interested in other products also from this point there is support for the operator.
Further it is up to the operator to complete and to update all significant data that is filed for this client; these can be used by future profile determinations.

 

2. What is missing on this prototype?

At the start it is not (yet) possible to pass all known data of the client to the offer applications, of course this has to do with the age of most applications. Most of them are made for just a MS-DOS system whom is not able to receive data from other systems in a simple standard way.
Next it is also not possible (yet) to pass data from offer applications to OPTIMISTâ , however new versions of this application appear more and more in a WINDOWS environment which that passing data should be within the possibilities.
I have to stipulate that at this moment there is no strict protocol for the passing of data, this makes communication between the different applications and OPTIMISTâ a time consuming and expensive thing.

KnowManâ Analyser utilises the Self-Optimising Universal Learner (S.O.U.L.), as an advanced statistical tool for high order non-linear statistical analysis and modelling. KnowManâ Analyser has extensive data-mining features and is well suited for modelling on the basis of complex and non-transparent data.

KnowManâ Analyser is in use at several client sites.

1. To generate models.

KnowManâ Analyser can automatically generate "intelligent programs", commonly known as "real-world models", which can be used for either the prediction of a value (e.g. a price) or for the classification of cases (diagnosis). Such models are always based on descriptions of the context and content of cases; data which you can either type in yourself or simply download from a given database.

2. To make higher order statistical analysis.