Chiasma Data is a registered Microsoft Partner and we use mainly, but not exclusively Microsoft's products and development tools. There are a number of reasons for this but chiefly:
- A very high percentage of companies, and our customers, already have the skills and licences necessary to run the products which reduces deployment costs.
- The products are highly scaleable and will run databases ranging from thousands of records to many millions with little change to the underlying systems.
- The products integrate well with the Microsoft Office environment, particularly when considering reporting and analysis services.
- Microsoft's professional products (SQL Server database, .net programming and the associated development tools) are highly featured and exceptionally stable.
- The costs are well contained with fully hosted and managed data warehouses available from around £1000 per month.
Which of these approaches is right depends on you.
For The Technically Curious
The marketing data warehouse consists of a number of core component:
- The heart of the system is a Microsoft SQL Server database (currently SQL Server 2008 Release 2) which sits within a processing harness written in
Microsoft's VB.net. The harness is modular in structure so that it can be augmented and configured to match an individual client's requirements - both functional and technical.
- Data being brought into the database (whether from a submitted file, web service or query to another database) undergoes an Extract Transform Load (ETL) process that interogates both
the structure of the supplied data and, at a record level, the compliance of each data element to the constraints placed on it in the configuration. A simple example of this
would be only allowing products to be entered that are on already known, or only allowing transactional values within a certain range.
- Any submitted record that fails validation can be configured to either reject the submitted file, the record or simply suppress the disallowed value. This allows us to maximise the
data capture while maintaining data quality.
- If the data supplied contains people's contact details then a process of deduplication is applied. This is rule based, configurable and de-duplicates people based on a combination
of the elements of their name, postal address, email address and telephone numbers. We avoid "fuzzy" matching using soundex or similar techniques as false positives (matching people who
are actually different) creates legal problems when dealing with data privacy consent and transactional information.
- To exploit the resulting information we need to connect with email broadcasters, mailhouses and websites: this is executed through a series of modules built into the harness that, for example,
push information to FTP sites, activates provider APIs or formulates web-service requests.
- Reporting and analytics are executed through Microsoft's Reporting Services and more complex analytics can be delivered through either OLAP cubes or a variety of predictive models built
using Data Mining Services. These can all be provided either as online services or integrated with the Microsoft Office suite of products.