Database Development - Quality Assurance.

There is some confusion in the business world about the difference between data quality assurance(QA) and spot checking data. The confusion usually appears after electing for spot checking when QA should have been done. QA and spot checking serve a similar but not identical purpose. Spot checking is usually done by business analysts and involves reviewing a few figures looking for obvious data issues. This is usually what gets me involved in the developer side of data issues.  What I do is write sets of SQL Server user defined functions, user defined table functions and stored procedures with matching parameters to check every record and every data element for accuracy.  I find it useful to use a processing design that allows variable batch sizes so I usually code the master procedure as a while loop that with each iteration does a set based SQL operation .  A variable value can be used for the top(n) construct so the number of records process per iteration can be controlled. As far as the actual data comparison I would first want to confirm historical data had not changed by comparing existing data with a read only historical record of the data. I would then review the new data for glaring errors and have subject area experts (end users) review it. Once the data was accepted by the business analyst it would be appended to the read only historical record of the data. The read only historical record of the data should allow adjustment records for on going balancing. As a developer I put subject area experts knowledge into business logic but I am not an actual subject area expert on every industry. I write SQL statements to perform statistical analysis on groupings of alphanumeric text so it is more important that I understand how to aggregate data correctly then be a subject area expert.

Be aware that you may pay someone to do QA for you but if your interpretation of the definition for QA is not specifically stated in your contract then it's likely your data has only been spot checked by a business analyst.