"The anti-script doesn't mean that you should wing it on every call... what anti-script means is, think about a physical paper script and an agent who is reading it off word for word... you're taking the most powerful part of the human out of the human."

- Marc Bernstein, Balto Founder & CEO

A quick search for “contact center scripts” yields thousands of results aimed at teaching managers the best practices for enforcing script adherence. What’s not so evident from these results is how to go beyond enforcement and use scripts as an enablement tool. 

Though scripts are a cornerstone of contact centers, they have not advanced at the same pace as other industry technologies. This presents pain points for managers and agents alike as companies have to pay for enforcement tools, perform ongoing training, and coach agents on a tool that may not lend itself to optimal results in terms of agent satisfaction and retention. 

Let’s go back to the basics: What is the goal of a script? Is it a prescriptive talk track that places success within predefined parameters of compliance? Or is it a foundation, a jumping-off point for agent improvisation? 

We talked to over 560 agents, and here’s what we found:

  • Agents don’t go off script because they forget what to say, they go off script because they want to — and that’s a good thing. 
  • The more involved agents are in writing their scripts, the more they’ll follow them, and the happier they’ll be overall.
  • Agents at larger contact centers follow their scripts the least.
  • Script adherence is heavily influenced by an agent’s trust in their manager and beliefs about their coworkers’ work ethics. 
  • The agent experience is universal. Factors like age, gender, tenure, industry, and more had no statistically significant effect on script adherence. 

Get a Copy of the Report in Your Inbox

Read the results on your own time or share them with your team. No strings attached.

Subscribe to future Conversation Excellence Lab reports.


We surveyed 567 agents across the United States. We defined agents in this context as people who make or receive calls with an established script as part of their job. Jump to the Methodology section below for more information about how our research was conducted.

Figure 1: Geographic Distribution of Survey Participants

Figure 1: Geographic Distribution of Survey Participants

Our respondents were 52.2% female and 47.3% male. The most represented age group was 25-34 (34.2%), followed by 35-44 (25.2%), 18-24 (17.1%), 45-54 (15.9%), and 55+ (7.6%). The average respondent age was 36.

Figure 2: Age Distribution of Survey Participants

The most represented industries in our survey were Retail (17.5%), Financial and Professional Services (16.2%), and Healthcare (12.2%). The least represented were Real Estate (2.7%), Collections (3.4%), and Government and Non-Profit (3.7%)

Figure 3: Industry Distribution of Survey Participants

Almost half of our participants (48.2%) were in Customer Service, followed by Sales, Answering/Transferring Calls, Consulting, Customer Account Cancellation/Retention, and Collections.

In terms of education, 37.2% of respondents had a high school diploma, 30% had a 2-year college degree, 25.4% had a 4-year college degree, 6.9% had an advanced degree, and less than one percent completed their education at the middle school level. 

We codified education on a scale of 1 (Middle School) to 5 (Advanced Degree). Both the average and median level of educational attainment among our respondents was 3, which represented a 2-year college degree.

Company sizes varied across participants, as well. 67% of agents worked in contact centers with 5-300 agents, 16.6% had more than 300 agents, and 16.4% had less than 5. The average contact center size was approximately 300 agents. 

Figure 4: Contact Center Size among Survey Participants

The average respondent had worked at their current job for 3.8 years, and in the contact center industry as a whole for 4.6 years.

Other demographic variables:

  • 43.2% of our respondents worked fully in-person, 21.7% worked fully remote, and 35.1% worked a combination of the two. 
  • 82.7% of respondents worked full-time, while 17.3% worked part-time. 
  • 77.4% of respondents were paid on an hourly basis and 22.6% were paid a yearly salary. 
  • 15.7% of respondents received no additional compensation, while the remainder received either a performance bonus, sales commission, form of gain/profit sharing, overtime pay, or combination of the above. The most common form of additional pay was overtime pay, followed by a performance bonus. 
  • On a scale of 1-5, respondent satisfaction with pay was an average of 3.6.

Why Agents Go Off Script

Respondents were asked to rank the following options in order of how often they cause them to go off script: 

  • I perform better when I improvise based on the customer’s needs
  • My script feels unnatural to read out loud
  • My script does not help me achieve successful calls
  • My script is not relevant to the content of my calls
  • I forget my script while on a call
  • I get bored of using my script
  • I’m not required to use my script on calls

Agents ranked the desire to improvise as the number one reason they go off script, followed by unnaturalness, unhelpfulness, and relevancy.

Figure 5: Why do agents go off script?
The lower the score, the higher the item was ranked by agents.

Figure 6: Tukey's Range Test of reasons for going off script
We used this test to discern whether there were differences in responses between metrics. Anywhere that the range crosses the dashed line was not a significant difference between the responses for the two metrics.

Interestingly, our 2021 agent survey found that the primary reason that agents make mistakes on calls is because they forget their material, get bored, or get nervous. This is consistent with the results of this survey — the differentiating factor here is intent:

  • When agents make mistakes unintentionally, they attribute these to the aforementioned factors.
  • When they stray from their script on purpose, they are most likely to attribute this to the desire to improvise or deficits in the script itself. 

What does this mean for contact centers? Contact center managers should first and foremost ensure that their scripts are natural-sounding, helpful, and relevant to the conversations at hand. 

Then, instead of worrying about agents forgetting script material or getting bored and distracted during calls, managers should consider what improvisation, and the desire to improvise, means. 

Improvising requires an internalized understanding of the situation at hand. Think about learning an instrument: You have to learn all the notes and chords before you can improvise. 

The goal, then, is not to eliminate improvisation. If an agent improvises on a call, it often means they know what they’re talking about. They’ve been in their role for a while and they know how a successful call goes, and which information they need to communicate for compliance reasons. When agents go off script, it’s because they’ve discovered how to turn their learnings into music. 

Agent Involvement in Script Writing

So, how do we capitalize on these learnings and share the music with other agents? A reliable strategy for making the most of agent best practices is to incorporate them into the script writing process. 

We asked the participants in our study to rate, on a scale of 1-5, how involved they were in writing their scripts. Our analysis found that the more agents are involved in writing their scripts, the more they’ll follow them. 

When you help put something together — a piece of furniture you ordered online, a meal you created from scratch, or a talk track you use hundreds of times a day — you feel a closer connection to it.

This is called the IKEA effect, and it is closely related to the endowment effect which states that we have a preference for things that are ours. When agents take part in writing their script, they feel that they have built something themselves. They put a part of themselves into the material, increasing their psychological closeness and attachment to their talk tracks. 

If agents are given ownership and adequately involved in script writing processes, they will have fewer complaints about their script, and adhere to its key concepts more in turn. 

We can also learn a lot from cognitive dissonance, a phenomenon that occurs when your brain has to resolve two opposing thoughts or beliefs. The very act of spending time on improving their script causes an agent to be less likely to dislike it. Why? Because if they disliked it, why would they have spent time improving it? 

At the end of our survey we asked participants what, if anything, they would change about their script. Approximately 64% of agents had something to change, but 36% did not. 

Agent involvement in script writing was a major indicator of whether or not agents wanted to change anything about their scripts on average. Those who reported the lowest involvement in script writing were the most likely to want to change everything about it, and those who wanted to change nothing at all had the highest rate of involvement as a group. 

The second-highest rate of involvement as a group belonged to those who stated that they wanted more ownership over their company’s script. This suggests that involvement begets more involvement: Once agents are given the opportunity to provide feedback on their script, they want to keep doing so. 

Figure 7: What agents want to change about their scripts

This shows initiative, buy-in, and an ownership mentality, all of which lend themselves to better performance and retention outcomes. See our 2022 Attrition Report for more information about how to prevent agent turnover. 

To increase both job and script satisfaction, managers should take advantage of the various psychological benefits of bringing agents into the script writing process; from the IKEA effect to the endowment effect, to cognitive bias and an ownership mentality — the takeaway is simple: agents are happier when they feel involved. 

Job Satisfaction and Script Adherence

While script adherence was positively correlated with script writing involvement, it was negatively correlated with job satisfaction. We asked respondents to rate their satisfaction with the day-to-day activities of their job on a scale of 1-5, and found that those with lower levels of script adherence were happier as a whole. 

A linear regression model outlines the effect of different variables on script adherence, with various results. As seen below, satisfaction with one’s day-to-day activities was the only variable with a negative effect on script adherence. 

Figure 8: Table of coefficients in relation to script adherence
Any coefficient with a p-value under 0.05 is significant.

This is not to say that script writing involvement leads to lower job satisfaction via increased script adherence, but rather that script writing involvement is a mediating variable that positively affects both of these variables. 

We found that, in general, agent job satisfaction decreased over time, but that this effect was mediated by involvement in script writing. Agents who were more involved in their scripts had higher satisfaction overall. 

The fact is, agents like to go off script. They like to take ownership of calls, improvise, and use their creative thinking. Without these, job satisfaction goes down.

Contact Center Size and Script Adherence

When analyzing differences in demographic conditions, contact center size was the only demographic factor that had a statistically significant impact on script adherence. 

Figure 9: Table of demographic coefficients in relation to script adherence
Any coefficient with a p-value under 0.05 is significant.

Regardless of age, gender, industry, education, tenure, job, and whether an employee is virtual or in-person, script adherence remains the same. Even in compliance-heavy industries like collections, call center script adherence levels were not statistically different. This suggests a unified experience that only varies slightly based on how many other people are making or receiving calls at an employee’s place of work. 

We measured contact center size in four buckets: 5-20, 21-300, 301-1,000, and 1,000+. The larger the contact center, the less agents follow their scripts. If a company’s scripts aren’t working for agents, the problem will only be exacerbated with scale.

Why could this be? There are vast differences in structure, recruitment, training, coaching, efficiency, and technology between a 5-person contact center and a 5,000-person contact center, creating plenty of opportunity for differences in outcome to arise. 

The larger the company, the larger the scope of responsibility per manager. When a manager has more agents to coach and QA, they cannot be as hands-on with monitoring each and every call. Problems with scripts are only magnified as a company scales. 

In 2021, we surveyed over 500 managers and discovered that they spend more time re-coaching agents at larger contact centers, and that coaching gets less effective with scale. This means that managers are spending more time preparing for less effective coaching sessions. 

In our 2021 agent survey, we surveyed over 1,000 agents and found that agents at larger contact centers forget the right thing to say on calls 8% more than agents at smaller contact centers, suggesting that there is less script internalization at play. 

The findings from both of these surveys are consistent with the idea that resources at larger contact centers are either less effectively channeled towards script adherence, or that adherence is less of an operational priority. 

Manager and Coworker Beliefs

In addition to job satisfaction and contact center size, agent script adherence is highly affected by variables that call back to agent relationships with and perceptions of their managers and coworkers. 

To discern these effects, we asked respondents to answer the following questions on a scale of 1-5:

  • How much you agree with the following statement: I trust management to set me up for success with a strong call script.
  • How much of your call script does your manager believe you follow?
  • How much of your call script do you think your coworkers follow?
  • How important is it to your manager that you follow your call script?

Figure 10: Table of coefficients in relation to script adherence
Any coefficient with a p-value under 0.05 is significant.

The table above shows a statistically significant, positive relationship between all of the highlighted rows above and script adherence. This means that:

  • The more an agent trusts their manager’s script writing, the more they’ll follow their script.
  • The more a manager expects an agent to follow their script, the more they will.
  • The more an agent thinks their coworkers are following their scripts, the more they will as well. 

Trust, modeling, communication of expectations. All of these takeaways come back to a question of culture. If you want to optimize for script adherence, ensure that expectations of adherence are effectively passed down, your agents understand the thought processes behind their scripts, and they have visibility into the workflows of other agents who can model your desired behaviors. 

Consider these statistics in tandem with the other findings in this report about agent desire for autonomy and involvement in script writing. How can you best check the need for script adherence with the freedom to improvise? When do agents need to stick closely to a talk track, and when is it less important that they do so? 

Striking a careful balance between adherence and freedom is imperative to ensure your business needs are being met while also empowering your agents and increasing job satisfaction. Reference our report on agent attrition for other strategies to keep agents happy. 


There is one nuance to the topic of script adherence that has not yet been discussed: agents may be adhering to the intent of a script, even if they are not adhering to a script verbatim. 

Not all lack of adherence is created equally. If agents go off script because they haven’t taken the time to learn their talk track, that should be addressed with training and coaching. But if agents are going off script because they perform better when they personalize their wording, that is a good thing, and something to encourage when possible. 

In this way, a great script is not necessarily a script at all, but rather a tool that encourages both the internalization of key material and morale-boosting improvisation. It makes room for agent feedback and turns prescriptive talk tracks into opportunities for agent collaboration and empowerment. 

Balto, for example, uses dynamic scripting and intent-based adherence measurements to retain key elements of adherence while also encouraging agents to make each call their own. These and other software options can be useful to help you strike this balance. 

Take the time to understand why your agents are going off script, and what form their variances take. Assess the trust levels they have in their managers and scripts, and understand the power of coworker modeling for script adherence. 

Above all, recognize that the agent experience is universal. Apart from contact center size, no other demographic variable had a significant impact on script adherence. No industry or use case is immune to these conditions. 

At the end of the day, agents want agency. 

Subscribe to the Conversation Excellence Lab newsletter to receive new reports directly in your inbox. 


This research was conducted in collaboration with third-party research firm Centiment in May 2022. Survey participants were asked to self-report on 16 demographic questions and 12 questions about their script adherence and script beliefs. The survey included an attention check at the mid-way point to ensure respondents fully understood what they were being asked. 

The sample population spanned a range of agent experience levels, industries, use cases, age groups, and other demographic parameters. 

The methods used for statistical analysis included linear regression models, ANOVA, and Tukey’s HSD. 

Subscribe to future Conversation Excellence Lab reports.


Balto. (2021, January 11). 2021 Contact Center Agent Survey Report​ - Balto Ai. RSS. Retrieved June 28, 2022, from https://www.balto.ai/research/2021-agent-report/

Balto. (2021, October 22). We Surveyed 500 Managers About Call Center Coaching. Conclusion: It’s Not Working - Balto Ai. RSS. Retrieved June 28, 2022, from https://www.balto.ai/research/call-center-coaching-manager-survey/

Balto. (2022, March 29). Contact Center Attrition: What Agents Want in 2022 - Balto Ai. RSS. Retrieved June 28, 2022, from https://www.balto.ai/research/contact-center-attrition-2022/

Balto. (2022, May 2). 2 Out of 3 Contact Center Agents Want to Change Their Script - Balto Ai. RSS. Retrieved June 28, 2022, from https://www.balto.ai/research/why-contact-center-agents-want-to-change-their-script/

Norton, M.I. (2012, July). The IKEA Effect: When Labor Leads to Love. Journal of Consumer Psychology. Retrieved June 28, 2022, from https://www.hbs.edu/faculty/Pages/item.aspx?num=41121

Pickford, H.C., Joy, G., & Roll, K. (2016, October 20). Psychological Ownership: Effects and Applications. Economics of Mutuality. Retrieved June 28, 2022, from https://eom.org/content-hub-blog/psychological-ownership#conclusion

Psychology Today. (n.d.). Cognitive Dissonance. Psychology Today. Retrieved June 28, 2022, from https://www.psychologytoday.com/us/basics/cognitive-dissonance

The Decision Lab. (n.d.). Why do we value items more if they belong to us? The Decision Lab. Retrieved June 28, 2022, from https://thedecisionlab.com/biases/endowment-effect

Cite this report

Balto. (2022, June 28). The Case For the Anti-Script: A Multifactor Analysis of Script Adherence - Balto Ai. RSS. Retrieved from https://www.balto.ai/research/script-adherence