ICMI is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.


Speech Analytics: Here Comes the Sun


Before the recent tech growth of the last decade and the rise of cloud-based SaaS solutions for contact centers, most functions of every department were highly manual processes. The quality management department was no different. In its infancy, team leaders or supervisors conducted call center quality management by listening to live calls while filling out a checklist to identify whether defined aspects of the call were handled appropriately. The primary flaw with this method was that the leaders couldn’t play and review calls in a coaching session. Flash-forward to the early 2000’s and some quality managers/analysts began monitoring recorded calls instead of listening live. These individual quality agents painstakingly listened to hand-picked calls (sometimes multiple times each), pausing and hand-scoring several agent calls per hour. The main purpose was to get feedback to agents and supervisors to continually improve the performance of each agent. The general issue here was that the quality team spent a tremendous amount of time searching for calls worthy of evaluating. Typically, a quality analyst would aim to evaluate a specific type of call (i.e. a successfully closed sale, a missed sales opportunity, a service call, etc.) and the process of trying to find specific call types were more about luck than ability.

The Beatles

A Hard Day’s Night

If you were really lucky, evaluation scores were tracked in a spreadsheet, but more often than not coaches recorded scores on printed forms, transposed them into something else and then printed on a Dot Matrix printer. Measuring and tracking trends for agent improvement, compliance items, and first call resolution often meant lots of arithmetic (on an actual calculator) and time spent compiling info from multiple locations. For the few that had data stored in a spreadsheet, it took less time to gather and analyze info, but often still took effort to manually create the formulas and functions needed to answer ad hoc queries. In addition, keep in mind that Excel wasn’t as robust as it is now.


As technology progressed, along came custom software that enabled QA agents to score calls in a centralized system. This greatly reduced the amount of effort required to find trends and issues within the database of stored QA scores. You could easily find the average score during a selected time period for a pre-selected group of agents. It was no longer difficult to see the improvement or regression of scores on a single line item over time. But it was still difficult to identify which calls to score without having to listen to portions of a call, one at a time.

Eight Days a Week

Despite the improvements in software during that time period, there was still a major restraint on the effectiveness of many quality management programs. The dataset provided limited the information and insight that software could produce. Providing that software with data meant teams of people pouring through hundreds of calls around the clock. Not only was labor costly, but even with QA agents scoring calls 24/7, your dataset was never complete. The small fraction of scored customer interactions limited the results software could provide, and a normal ratio of evaluations of agents was roughly 5 evaluations per month per agent. This meant less than 1% of calls were actually scored, effectively making the sample size far too small to be statistically valid.

Come Together

Fortunately, technology has never stopped progressing. Recent advances make it possible for computers to accurately transcribe and analyze human speech at scale. These advancements have made the product we now know as speech analytics possible. This software enables a quality management program to analyze the content of one-hundred percent of the customer interactions that a contact center may have. A single interface with the same technology can complete the analysis of phone, email, chat, and even social media interactions. Combined with the more powerful and modern analytical capabilities of big data, the sky is now limit for gaining insight.

With a Little Help from My Friends

For smaller contact centers, software of this caliber seemed out of reach for a long time due to substantial price tags. But thanks to competition in the market and the reduced price of data storage, this is no longer the case. With the rise of cloud computing came the creation of Software-as-a-Service companies, third-parties who can provide contact center software for a nominal subscription fee, or in some cases, charge based on usage.  Now the latest and greatest technical capabilities are available for even the smallest contact center operations. There is literally a speech analytics solution for every budget if you find the right technology partner(s).

Ticket to Ride

With the growing ubiquity of speech analytics software in the contact center industry, you can be sure that if you’re not using it, you have competitors who are. More importantly, if you’re not using it you’re not gathering statistically valid sample sizes of data from interactions with your customers. This means other companies who are can gain a competitive advantage against you. Similarly, if you’re not listening to and quantifying customer feedback, you’re likely only letting complaints drive process improvement. But why let things become sour for your customers before you react? The quality management train is at your station and the ticket is in your hands, jump on board and reap the benefits!