Testing The Optimal Order

Can good communications be maintained when testing is offsite?

The customer is an international Information Management Services company, specialising in the automotive industry. Atrial were asked to help in the testing of an application being developed.

The application was devised to aid stock controllers at car dealerships to build a parts order based on statistical analysis of usage trends. The application processed sales data for the defined trading period using statistical analysis and produced a recommended order level based on seasonal trends, peak sales, outstanding orders and current stock levels of each branch. The user could then alter the level manually the levels if required.

The Solution

Atrial provided two consultants to work from Atrial's offices on the production of test documentation and the execution of the tests.

The client provided a set of design documents including UML based Use Case Specifications, which were used to design a Test Specification document. The document was approved by the client and used as the basis for testing.

The testing was carried out at Atrial's offices in Letchworth. Multiple test environments were created allowing the tests to be run on a selection of PC hardware and operating systems. Remote access to the clients faults recording system was arranged and the usual fault, fix and re-test iteration was performed.

New releases of the application were FTP'ed from the client site on a regular basis, along with documentation and release notes. Each new release was "sanity checked" prior to commencement of testing and a report sent to the client.

The testing had three main focus areas:

GUI	User level testing to verify functionality, ease of use, data integrity, consistent presentation and layout.
Data input	Verify sales data was accurately read into the application and all errors were accurately reported by the system.
Recommended Order	Verify the statistical analysis was accurate in all cases and this was reflected in the recommended order values.

In order to ensure the statistical analysis could deal with the defined business rules and unusual sales trends a custom sales database was developed to exercise all the possible scenarios. Using this, known input values could be used to generate outputs, which could be compared to the expected output values.

Added Value

Atrial Systems always prefer to work with the client, becoming immersed in the technology and business sector. This policy often leads to adding unforeseen value to the project.

A beta version of the application had been developed by the time Atrial became involved in the testing. Comparison between the application and the original documentation was used in a second documentation iteration, ensuring continued accuracy.

Atrial consultants became expert users of the system during the test phase of the project and in recognition of this the client requested that Atrial write the help files for the application.