How to Gain Value with Human-Based Verbatim Analysis

Today, companies invest significant amounts in customer surveys to learn how to serve their customers better.  Automated surveys are valuable. However, allowing customers to leave comments (verbatim), expressing exactly what they think and feel, empowers companies to hear the voice of the customer (VoC).

To determine the best verbatim analysis solution, companies need to understand the process for collecting, organizing and mining actionable data.

We interviewed two experts at SkyCreek, a customer communications service provider, to gain insight into the methods, tools, systems and processes involved in verbatim analysis.  SkyCreek’s Jack Havas, CEO and Jack Fleming, CCO, have more than 60 years of combined experience focused on customer interactions and communications.  They are passionate about helping their customers develop survey programs that provide the most accurate and near real-time feedback.

 

FLEMING:  Every customer is offered an opportunity to leave a comment in their own words.  Once the verbatim are captured and stored in a file, we have a system that distributes the content to transcribers.  For each survey program, our transcribers are given two customized lists of sentiments, one for positive feedback & one for negative feedback.

As transcribers review the verbatim, they click on the sentiments that apply.  Our system then performs the sentiment analysis to categorize whether the verbatim is positive, negative or mixed as well as assigning alert status and exceptional performance status.

In addition, some clients opt for complete transcriptions of verbatim customer comments.  This is where the human involvement is particularly paramount. The ability to correctly understand all that is being said is critical.

HAVAS:  Our team of transcribers processes thousands of verbatim each month.  They have many years of experience and are well trained to identify communication nuances, such as word choice, voice tone, and inflection.

 

FLEMING:  There is no doubt that speech recognition has come a long way, even in just the last three to five years. However, in our experience, for our customers to really understand the quality of service provided to their customers, there can’t be any confusion or misunderstandings.

When leaving a verbatim, there are many factors such as tone, speech dialects, background noise, and interference where technology solutions are not able to provide 100% accuracy.  The best way is for a human to listen and transcribe what they hear.

HAVAS:  If you want accuracy, there is no doubt that human-based transcription services are significantly more accurate (99%+) than machine-based (60-80%).  If you are attempting to analyze millions of verbatim, then machine-based is probably an okay solution to give an overall picture.  But if you are using the verbatim analysis to manage dissatisfied customers, in order to get them back on track, you need to know the details that human-based transcription provides and have confidence it is highly accurate.

 

FLEMING:  Absolutely!  Our programs analyze and flag important verbatim as soon as the transcriber submits the sentiments.  We have a table-driven process, based on customer-defined criteria, that trigger alerts.

Alerts automatically generate messages to the appropriate recipients so that action can be taken immediately.  Our customers are able to see the results and verbatim in the eTouchPoint interface as soon as the alert is triggered.

 

FLEMING:  We provide services to very big clients, often with tens of thousands of verbatim each month.  We give our clients the ability to manage that data by removing the upfront processing that an organization would have to undertake.

Through our program, we do all the work – we review and transcribe the verbatim, identify sentiments, determine positive/negative/mixed and further evaluate in order to identify and bring to the forefront the verbatim that trigger customer dissatisfaction alerts.  We drop all of this feedback in their laps, so they are able to take near real-time action.

 

FLEMING:  It is not uncommon in a survey program for companies to focus only on negative or dissatisfied customers.  Often they take the negative feedback and use it to deal with problems and the employees responsible by means of punishment or additional training.

We encourage our customers to take the information and to also use it as a positive reinforcement tool.  Our process allows customers to define and identify verbatim where an employee has gone above and beyond, thus flagging an exceptional alert.

One of our largest customers has created an employee incentive program based on customer verbatim that trigger a positive alert for exceptional service.  Verbatim flagged as an exceptional result in a letter of commendation being written and sent to the employee.  This positive reinforcement has a big influence on employee performance.

HAVAS:  Our exceptional alert process allows customers to look beyond using the tool as a hammer to minimize customer dissatisfaction and to also focus on using the information in a positive way to reinforce exceptional behavior.