How AI replaces a supervisor in an IT academy for children by analyzing customer conversations

Last year, we introduced artificial intelligence to replace a service quality control specialist. The tool automatically analyzes phone calls, finds errors, and advises on improving service. And now, every business can train artificial intelligence for its own needs on internal scripts, instructions, and documentation. Find out how Ringostat’s AI Supervisor is already working to help the GoITeens team reduce the cost of staffing assessors.

Why the Academy needed artificial intelligence

GoITeens is an online academy that has been teaching IT skills to students aged 7 to 17 for eight years. More than 50,000 children have been trained during this time. 

The academy offers 14 courses, and prospective students must first communicate with GoITeens representatives to make an informed choice. This usually happens by phone, so the company carefully monitors call processing. After all, the quality of the consultation determines whether a student wants to study and how well they choose the right field of education.

Previously, GoITeens used audio recordings of calls as a control tool. The manager had to listen to them and rate them according to dozens of criteria. But there were still several difficulties.

1. Monitoring conversations takes a lot of time. Dialogues with customers sometimes last 8–11 minutes. Listening to at least 10 of these conversations every day while making notes on the manager’s work will take about two hours. In addition, you need to analyze test lessons, which take even longer.
2. The need for a staff of assessors. Due to this workload, five specialists had to monitor calls, which meant additional salary costs.
3. Calls were selectively monitored. Even with assessors, it is impossible to analyze all calls. Therefore, it had to be done randomly, and some crucial points could be overlooked.
4. Listening to well-handled calls was useless. It was unclear which calls really needed attention, so the team spent most of its time on calls that were unproblematic.

Because of all this, GoITeens was interested in a solution significantly reducing monitoring time. They found it in Ringostat, an AI telephony, call tracking, and call analytics platform. The Academy already used Ringostat’s virtual PBX, callback widget, and call tracking. Later, they discovered that Ringostat had added artificial intelligence for call monitoring.

What was the complexity of the project?

1. The company does not have one common conversation script. The call center, the trial lessons department, and the sales department talk to customers. Each of them has a separate script that contains mandatory steps. Accordingly, the work of the departments should be evaluated separately.
2. Trial lessons should also be analyzed. The conversation can develop differently, and the evaluation criteria should change. For example, the employee should ask questions if the student has already taken some courses. If not, then others.
3. You need to consider the client’s and employee’s moods. Children’s moods have a stronger impact on their desire to learn than adults. Therefore, the sales rep should ask the necessary questions to ensure the potential student is satisfied. You must also track whether the employee was polite and friendly, even if the child was not always attentive.

Having learned about these features of the project, Ringostat developers offered Ringostat AI Supervisor. As part of this service, artificial intelligence is customized for a specific company based on its rules and documentation. Therefore, the AI was trained on the data provided by the academy, taking into account:

• scripts for each department; 
• the stages that the conversation should be divided into for each department, and the manager’s mandatory remarks during them;
• keywords that should be used in the dialogue;
• a list of the main points of the conversation that should be used to evaluate the manager’s performance.

All of this allowed GoITeens to get a customized solution that fully analyzes conversations, taking into account their context.

How AI works to analyze conversations in GoITeens

Deciphering a conversation and identifying its stages

Artificial intelligence converts the audio of each call into text. The academy also uploads videos of trial lessons to it. Transcribing a conversation saves time because GoITeens don’t have to listen to the entire dialogue. It is enough to quickly re-read the text or find the desired fragment using the search.

AI can be set to automatically translate conversations into English. Or you can use transcription in the original language, as artificial intelligence supports more than 50 languages.

In addition, AI understands when one stage of a conversation ends, and another begins. That’s why you can also read the dialogue transcription separately by stage. This is useful if you need to analyze only the closing or product presentation.

Capture a summary of the conversation and key points

To understand the main point of the dialogue, it is enough to analyze whether the manager has taken steps that are fundamentally important for the company. AI can distinguish between them in the dialogue and record it in a separate report block. For example:

• whether there were any objections, what they were, and how the employee handled them;
• whether different payment options were presented;
• whether the client was satisfied with the information provided — this is assessed by the presence of clarifying questions and mood.

Let’s dwell on the last point in more detail. Artificial intelligence can distinguish between the general mood of the conversation and the mood of the employee and the client separately. This is extremely important for understanding the impression of communication on the student. If a student is upset or lacks information, it can be quickly corrected and prevented from dropping out. If a manager always has satisfied clients, this is a reason to share their approach with colleagues.

Evaluation of the manager’s work according to the checklist 

Each dialogue is evaluated according to the appropriate checklist for the department provided by GoITeens. Ringostat distinguishes which department is talking to the student and sets the appropriate evaluation criteria. For example, if the dialogue is with a call center, it is a mistake not to invite a student to a trial lesson. But it won’t be a mistake for the sales department, because they are already discussing course payment and other final issues.

If the employee has said the required line, AI marks it with a tick. If not, it marks it with a cross. This way, the company can easily check whether the employee followed the script by asking all the questions to help the student decide on a course.

Some examples of evaluation criteria:

• at the opening stage: whether the manager said hello, got to know the prospective student, found out or informed about the reason for the call;
• not at the stage of identifying needs: whether he or she has learned about the child’s interests and expected results from the training;
• whether he/she listed the names of relevant courses during the communication  — so that the student has a sufficient choice;
• whether he offered a choice of time for a trial lesson, etc.

Recap of the conversation and best next steps

Each conversation should end with clear agreements — even if the potential student has not yet decided to enrol in the course. That’s why artificial intelligence separately records how the dialogue ended and advises on what to do next:

• result — for example, a student has enrolled in a course;
• whether the deal was closed;
• next steps — let’s say, contacting the customer when they choose a schedule and a tariff;
• deadline.

This helps avoid situations when deals are stalled because employees can easily check what to do next. А sales rep can always recall agreements with a particular client. 

Record important information about the student

Previously, after a call, employees had to manually record information about a potential student, which took extra time. Now, AI collects important data about the child from his or her answers or the replicas of the parents involved in the calls. For example:

• age;
• whether and what kind of previous educational experience students have;
• interests, etc.

This, first, saves the sales rep’s time, and second, allows them to check how relevant the courses offered by the employee are.

Conclusion of AI implementation for conversation analysis

“Previously, we didn’t know which calls needed attention, so we listened to them selectively. At the same time, 80% of the calls that we listened were handled well. So we were only wasting time monitoring them. Five assessors monitored this, filtering out the calls that really required management attention.

With AI, we “narrow the circle” and see which dialogues are problematic and need improvement. As a result, we have reduced the number of assessors and can save on salary costs. Analyzing about 300 targeted calls using AI takes about 10 hours of work. We save at least 5% of our working time for analysis and 5% for searching for such calls”.  

Daryna Kryshen Head of the GoITeens call center