Quality Assurance?!

Quality Assurance is the 800lb gorilla in any conversation that requires mass production or mass replication of tasks in the data and information world.

It's a conversation that should never be one sided and should always be as realisitc in execution as it is in conception. By this I mean you should have a realistic understanding of what level of quality is needed. Some of our clients require a crisp and firm 97% or greater quality level while others are comfortable with a 85% quality level (based on the magnitude of data consumed)

Quality costs money but quality information is very valuable.

If your business utilizes a one-size-fits-allapproach to quality assurance testing, then you may be over- (or under-) investing in QA.

Here are a few rules that Shore Group Associates applies when designing QA processesand building QA teams for our partners:

Rule 1 - Initiate QA at the point of data capture (aka garbage in yields garbage out) -- applying simple data entry validation rules at the moment of capture, such as ensuring a phone number or date are composed of the appropriate count of numeric values, will reduce the likelihood of subsequently revisiting these fields.

Rule 2 - Not all data records are created equal -- identify which fields and records deliver the most value to customers, and define your accuracy and quality targets for each group of records

Rule 3 - Use technology first -- a host of automated collection and validation tools exist today; whether you build your own, engage a partner to configure their a proprietary toolset, or license an off-the-shelf solution, start first by defining and programming your business rules into a validation algorithm that kicks out exception reports (the best machine learning / AI tools will combine these reports with your subsequently remediated records in order to improve original collection tools (scrapers, parsers, matching algorithms)

Rule 4 - Segment records based upon Rules 2 & 3 -- this allows you to rout the automated collection results into one of three places...(a) directly into production databases, if regression tests have proved the accuracy levels meet your standards; (b) to QA Associates for review, if your confidence level is not quite high enough to release directly into product; or (c) to your analyst pool if primary research due to the complexity of the records. Applying these rules often reduces the number of records that are routed to QA Associates, thereby allowing you to reduce investment in these relatively more expensive SMEs.

SME's can get pricey. They are smart people who are well worth the investment when they are needed but if you can create a effective and efficient quality assurance process and reduce the heavy lifting it's a win win.

Besides - We all want the 800lb Gorilla out of the room.

Neal.Conlon@shoregrp.com