A Scrip for Scripts

By Dennis Adsit, PhD

July/August 2009

A number of companies are offering scripting tools for agents, and the use of these solutions is expanding.  This is a positive first step on the road to improving agent output measures.  However, because of their inherent limitations, scripts often overpromise and under-deliver the desired improvements in process adherence, call resolution, customer satisfaction, and call and handle time.  This essay, then, is a scrip (as in prescription) for addressing the shortfalls in scripting tools to ensure they deliver the desired improvements in agent output measures.

First, let's be clear on why scripting a call is an essential first step.  As you may know, the approach used by Toyota to manufacture cars, known as the Toyota Production System, not only vaulted Toyota to the world's leading global automobile manufacturer, but it revolutionized the manufacturing world.  A Harvard Business Review article entitled "Decoding the DNA of the Toyota Production System" outlined the four key rules that were critical to the success of the Toyota's system.  The first rule was "all work shall be highly specified as to content, sequence, timing, and outcome."  The level of specificity in manufacturing can come down to exactly how many turns to tighten a screw.

Now, if someone were to say, "Specificity is not a hallmark of work in call centers," that would have to be viewed as a gigantic understatement.  Unfortunately, when you put agents, monitors, trainers, and coaches in a room together, they often cannot agree on how a particular call should be handled.  Therefore, is it any wonder that we see the variability that we do in agent process and output measures?  How can we improve agent outputs when there is no agreement on exactly how a credit limit increase call or a cell phone activation call are supposed to go?

Limitations of Scripts: Scripting what agents are supposed to do and say on a call is essential, but that is not all.  You have to make sure the agents actually use the scripts, and then you have to make sure those scripts are continuously improved.

It may seem counterintuitive, but for many reasons agents often don't follow the scripts they have been given.  First, agents are under a lot of pressure to perform quickly, and it can be hard to read from a script and do what they are supposed to do in their systems at the same time.  Also, after handling a given call type hundreds of times agents often try to memorize what to say so the call sounds more natural.  Finally, processes change so frequently, the scripts are often wrong or outdated and so the agents rightly ignore them.

The second problem with scripts is that companies don't dedicate the resources to continuously improve them.  Notice that I didn't say: "Resources are not dedicated to ‘keeping them up-to-date.'"  This is assumed.  After all, if you are not going to dedicate resources to ensure that scripts are up-to-date, why use them at all?

Someone needs to stand back and actually study what is working and what isn't so that changes can be made to improve calls.  For example, in offshore centers, scripts are often written by native English speakers, but are then deployed in centers where English is the second language.  If someone would just listen to calls and take suggestions from agents, difficult-to-say sentences could be made easier and still accomplish the desired objectives.

Addressing Scripts' Shortfalls: We have identified two main shortfalls with scripts:  agents don't use them, and no one continuously improves them.

Two things can be done to address the first problem.  First, show the agents the performance link between following the scripts and call metrics.  If you have built the scripts correctly, agents who follow them should have greater customer satisfaction, increased first call resolution, higher quality compliance, and lower handle time.  In addition, if you are rewarding your agents based on call metrics, showing them the statistical link between script adherence and the call metrics should drive greater use of the scripts.

Second, prerecord sections of the call that you want the agents to get exactly right and allow them to play the recordings.  This is "scripts on steroids," and it helps in multiple ways.  First, the caller gets a clear voice with less accent issues.  Second, you know exactly what the caller is hearing.  Third, by tracking keystrokes, you know whether the agents are playing the required recordings on the required calls. 

You might be wondering how prerecording audio can help with script-adherence.  The answer is simple: the agents want to use the prerecorded sections as much as possible because it gives them a chance to rest their voice.  Talking all day is extremely tiring.  If the agents can get a break, especially on long, boring repetitive sections, they are going to use it every time.

Start by breaking down incoming call volume into call types.  Next, assume the call is going to go in a straight line and define exactly what the agents should say to callers.  Finally, make it easy for the agent to execute the defined process by preprogramming system actions and prerecording audio files. 

The result is a live agent on the call listening to the caller and navigating the call using the paths that have been built.  If the call goes in a straight line, the agents may never have to break in with their live voices.  However, if the caller asks a question that is not part of the call flow, the agent can just take over the call.

With this solution, sustainable improvements are driven in every call handling metric.  The agents' satisfaction increases because their jobs have become less stressful.  Also, process adherence and compliance can approach the 99.999 percent level because, just as in manufacturing, the correct process has been built into the call, making it easy for agents to execute it.  Finally, costs have been reduced because by engineering the call, handle time has been systematically reduced. 

The final piece of the puzzle for making scripts more effective is continuous improvement.  Call center leaders need to establish a process engineering function to complement the existing training, QA, scheduling, and operations functions.  If performance on a given call type is not at the desired level, it is up to the process engineering team to close the gap.  This team would study a script, listen to calls, experiment with different alternatives, and examine the effects on output measures to ensure that performance on a given call type is continuously improving. 

There is no question that scripts are an essential first step to improving agent outputs, but they are not enough.  Without efforts to ensure the use of those scripts, and without the investment in a process engineering function to continuously improve the scripts, any uptick in outputs that are achieved will quickly stagnate, and the return on investment of tools of this type will be in jeopardy. 

Dr. Adsit has been achieving results with organizations for over twenty years.  He is currently the VP of business development for KomBea Corporation.  Prior to KomBea, he was senior vice president of call center operations and process excellence at Intuit, where he drove dramatic change and tens of millions of dollars in benefits.  From 1995 to 2000, Dennis was senior vice president and Six Sigma practice leader for Rath & Strong Management Consultants.  He can be reached at dennis.adsit@kombea.com.