Monday, March 14, 2016

Adventures in Speech Analytics- Part II

In part one of “Adventures in Speech Analytics” we discussed the basics of how a Speech Analytics (SA) program operates.  Here in Part II we will focus on one major aspect of Speech Analytics- Queries.
Queries, Queries, Queries:
A query is a pre-built search that has been created within the SA tool to help an end user best refine their ad hoc searches.  SA programs will have queries built that are of value to the organization.  The methods in which a query can be built vary per the software but the fundamentals are usually the same- a combination of phrases that have been saved with a few conditional rules attached to it (i.e. to hit in the first 30 seconds of a call or to only hit on the agent or customer line).  Query building is almost like making a great batch of marinara sauce- the process requires you put a little of this and a little of that until it comes out perfect.  Query development consists of the following:
  • Establishing a Business Need- Query building takes time and resources so it is very important that an organization determines what type of call drivers they would like to direct their focus towards. For instance, telecommunications companies will usually have queries built around lines of business they offer along with queries that may focus on other drivers like billing or technical support. Oftentimes, real-time business impacting events may warrant the need for a query’s creation such as program negotiations between a telecommunications company and TV networks.
  • Building the Logic- This process can vary depending on the SA software being used. Your query builder is going to perform a few ad hoc searches in the tool to get a feel for the language being used in the calls as well as finding out what is known as a confidence rating. A confidence rating is what allows the builder to see how accurate his/her search is. Once the builder has an acceptable confidence rating they save that search and build a query against that search using Logic. Logic is special language that the SA tool uses to determine when it needs to hit for a particular keyword or phrase.
  • Validation- Once the query has been built; the query builder will submit their query for validation. Similar to building the Logic, a new confidence rating will be issued to determine the accuracy of the query. The validation process is crucial to query development because a query is only as useful as how accurate it is. The requirements for what is considered acceptable for queries can vary per program. If the query passes it moves on to the production phase; however if the query fails then it’s back to the drawing board.
  • Production- When the query passes validation it is then ready for the production phase. This means that your brand new query is ready to go live in your tool for your end users to use and report against. The method of how a query is introduced into the live environment varies on the SA tool but for the most part there is a weekly maintenance period where the session is copied and updated so that new and old calls can ingest against the new query.
The query building process takes a bit of patience but the rewards are certainly worth the time. Queries are among the most trusted allies when it comes to assisting an organization in identifying trends in their business.  Queries allow you to deep dive into the heart of specific business impacting topics or events by using solid data to back-up the findings.  In my next article I will further elaborate on another major aspect of Speech Analytics that I briefly touched on in both this and the previous article- Ad Hoc searches!

Operations Manager, Speech Analytics 

With more than 4 years of Quality Assurance experience in a call center environment, Diana's objective is to simultaneously promote and educate the world of Speech Analytics with a human touch; one which further emphasizes the importance of First Call Resolution, and overall customer experience.

Follow Diana on LinkedIn.

No comments:

Post a Comment