AI reviews call scripts objectively, improving consistency and quality.
1. Issue: Subjectivity in call script evaluation
The company provides sales support services and focuses on inside sales operations where sales operators call potential customers to set up appointments. The operators develop their individual call scripts, which are then reviewed by both themselves and their team leader.
However, the reliance on the reviewer’s experience and intuition resulted in a lack of clear evaluation criteria and objectivity. This made it challenging to utilize the reviews to enhance the script based on quantifiable data.
2. Solution: Introduce objective reviews with generative AI
To address this, the process owner implemented a system where a generative AI analyzes the script’s structure and phrasing after submission to determine if revisions are needed. The AI then clearly outlines any necessary changes.
The AI provides feedback, and the script writer reviews and resubmits the content if needed. The team leader then conducts a final check, ensuring a more objective and accurate review process.
AI ensures consistent document quality through standardized reviews, eliminating author-based variation.
An ID will be automatically assigned to the call script.
x3-1. Set subject
The subject will be set automatically.
2. Check
The sales team leader will conduct the review.
1x. Revision
If the reader rejects the submission, the writer will revise the content and resubmit it.
x3-2. Add [Withdrawn] to the subject
The process will be terminated, and the subject will be automatically updated to “Withdrawal” if you cancel during the creation or modification of the call script.
An ID will be automatically assigned to the call script.
x2-1.Generate prompt
Prompts are automatically generated based on your input.
x2-2. AI review
The AI reviews the call script based on the generated prompts.
x3-1. Set subject
The subject will be set automatically.
2. Check
The sales team leader will conduct the review.
1x. Revision
If the reader rejects the submission, the writer will revise the content and resubmit it.
x3-2. Add [Withdrawn] to the subject
The process will be terminated, and the subject will be automatically updated to “Withdrawal” if you cancel during the creation or modification of the call script.
Processing screen preview (click to enlarge)
Before / After Comparison (you can move the slider)
Rotational task allocation balances operator workload and improves response speed.
1. Issue: A system that allows task avoidance
The contact center, staffed with 20 operators, handles email inquiries as part of its BPO outsourcing services.
Inquiries are handled by operators who take on the case themselves. When an operator is available, they select an inquiry from the waiting list and take it on.
Hourly paid operators are evaluated based on customer feedback. Positive evaluations are given for excellent service, while complaints result in negative evaluations.
Since the evaluation system doesn’t consider the number of cases handled, employees lack the motivation to take on cases proactively. This results in procrastination and a decline in overall productivity.
2. Solution: Rotate and allocate evenly
To address this, the process owner implemented a rotation allocation method. This new approach ensures inquiries are distributed equally, replacing the previous system where responders selected work voluntarily.
Each job is assigned a process serial number. The serial number is divided by 20, and the remainder plus 1 is used to assign the job to one of 20 operators. Each number between 1 and 20 corresponds to a specific operator. For example, Mr. E is assigned jobs numbered 5, Mr. N is assigned jobs numbered 14, and so on.
Workload is balanced among operators, improving response speed and reducing inquiry delays.
1. Issue: Inability to analyze incomplete requests
The website of ◯△× Company was built with WordPress.com, and is currently operated and managed by the Marketing Department. The Marketing Department utilizes a Workflow App to manage and fulfill article posting requests from other departments.
Recently, there have been instances where published articles based on requests contained outdated or incorrect product information, such as pricing and specifications. In the past, the marketing department would review the content before publishing. If any errors were found, an authorized team member would delete the problematic request on the Workflow app.
The increasing number of deficiencies in requests makes it challenging to identify trends, such as which departments or content types are causing the most issues. Additionally, the complete removal of deleted requests from the Workflow app prevents searching and analyzing past cases. This lack of data-driven insights hinders the implementation of improvement measures and increases the risk of recurring deficiencies.
2. Solution: Add keywords to the subject of abandoned requests
The Process Owner has configured two options to resolve this: discarding the draft or sending it to WordPress after proofreading. If the draft is discarded, the workflow proceeds as follows:
The subject of the request will automatically be marked as “[Abandoned]”
The request will be considered completed, but the history will be recorded
For example, if your subject is “Blog Adding New Features for 2025”, it will be displayed as “[Abandoned] Blog Adding New Features for 2025”, making it easier to narrow down your search.
Marketing department staff can submit their draft.
Back up your requested manuscript
Manuscripts requested by other departments are automatically backed up.
Proofreading
A marketing team member will proofread the manuscript if any required publication items are missing.
Set subject
The subject line will automatically be updated to “[Abandoned]”.
Image to WordPress
Upload your media files to WordPress.com.
Submit manuscript to WordPress
Create a draft post on WordPress.com.
Published with minor corrections
The marketing department will make minor edits to the draft and publish it.
Release Notification
The publication of the article will be shared within the department.
Before/After (Slide to move)
3. Effects
Identifying request errors
By adding “[Abandoned]” to the subject line, you can quickly identify past request errors.
Improved searchability of request history
The history of deleted requests is retained, allowing you to easily review past requests by searching for the keyword “[Abandoned]”.
Accelerating data analysis
This simplifies the process of visualizing error trends and analyzing problematic departments and request contents.
4. Other business applications
Review and approval flow management
Product information correction requests and advertisement placement approval flows can help to identify, record, and prevent the recurrence of mistaken requests.
Managing customer history
This will clearly identify errors made by customer support (CS) and prevent them from happening again.
Optimization of ordering and procurement management
By keeping a history of ordering errors, you can review and strengthen your purchasing rules.
Streamlined project management
Identify unnecessary tasks and incorrect specification changes to reduce wasteful work.
