[Actual State of AI Utilization of Call Recording Data] About 80% of Companies Recording Customer Calls Have Introduced "AI Analysis," but "Utilization Gap" Becomes Clear: Why Doesn't Introduction Lead to Operational Efficiency?

Thinca Co., Ltd. (Headquarters: Chiyoda-ku, Tokyo; Representative Director, President & CEO: Takahiro Ejiri; hereinafter "Thinca"), which develops and sells the AI communication integration platform "Kaikura," conducted a survey on "The State of Call Recording Data Management and AI Utilization in Companies" targeting 1,019 managers and supervisors at companies that record calls. The results revealed that while over 80% of companies recording customer calls are using AI, persistent issues such as "it takes time to grasp the content" on the front lines highlight a "utilization gap" where the presence of AI does not directly translate to operational efficiency.

Background of the Survey

Against the backdrop of strengthening compliance and measures against customer harassment, "recording all calls" has become standard in many companies. However, there are not a few cases where the vast amounts of accumulated voice data remain unused, becoming a "black box."

Recently, efforts to utilize AI for transcribing and summarizing recorded data for business purposes have accelerated, but issues such as "not being able to fully utilize AI functions" and "it doesn't lead to operational efficiency" are also observed on the ground.

Therefore, we conducted a survey on the actual state of data management and AI utilization in companies that hold call recording data.

Survey Overview: Survey on "The State of Call Recording Data Management and AI Utilization in Companies"

[Survey Period] Thursday, March 26, 2026 – Tuesday, March 31, 2026

[Survey Method] Internet survey via PRIZMA (https://www.prizma-link.com/press)
[Number of Respondents] 1,019 people
[Survey Target] Monitors who responded as employees or managers/supervisors in call-handling departments at companies with 50 or more employees that record calls

[Survey Source] Thinca Co., Ltd. (https://www.thinca.co.jp/)
[Monitor Provider] Sacrisa

About 80% of Companies Recording Calls Use AI for Analysis!

First, when asked, "Does your company use AI for processing or analyzing customer call recording data?", a combined total of about 80% responded that they are using AI, with 37.9% saying they 'actively use it' and 45.4% saying they 'use it partially.'

While the introduction of AI technology is progressing in many workplaces, about half of these are still at the 'partial use' stage. It is presumed that while the tools themselves have been introduced, there are cases where company-wide operation or full-scale business efficiency has not yet been achieved.

So, how is it actually being used?

Those who answered 'actively use it' or 'use it partially' in the previous question were asked, "How do you use AI for processing or analyzing customer call recording data?" The most common response was 'Automatically summarizing long calls to grasp content (41.1%),' followed by 'Transcribing call content for visual confirmation (36.5%)' and 'Judging complaints/customer harassment from emotional fluctuations (35.1%).'

This suggests an intention to not only achieve 'efficient understanding' through summarization and visualization but also to objectively analyze customer emotions for more advanced response management.

Despite Using "AI Summarization," About 70% of Companies Take "Around 5 Minutes" to Find Target Voice Data

As AI utilization advances, how much has the time required to check necessary call content been streamlined per case?

Despite the progress in AI utilization, we asked, "On average, how much time does it take per case to find specific data from past recordings and confirm the necessary call content?"

The results revealed that it takes:

'1 minute to less than 5 minutes (31.4%)'

'5 minutes to less than 10 minutes (41.8%)'

'10 minutes to less than 30 minutes (12.0%)'

'Around 5 minutes' is the most common range, indicating that a certain amount of time is spent searching for the target audio. Even if the search time is only a few minutes, the accumulation of daily tasks can lead to a loss of working hours for the entire organization, and may also cause a decline in response quality, such as delays in responding to customers.

Aiming to Improve Response Quality, but Recording Data is "Hard to Find": The Utilization Gap Caused by Data Explosion

To explore the reasons why voice search takes time despite the introduction of AI, we first asked about the accumulation status of recorded data.

When asked, "To what extent does your company record customer calls?", about 80% responded, 'We record all calls (83.0%).'

It is thought that the continuous accumulation of vast amounts of voice data daily creates the challenge of 'being unable to find necessary call content or specific voice data.'

How is this vast amount of data stored?

Regarding management methods, 'Cloud services (SaaS) (49.9%)' was the most common, followed by 'In-house server/on-premises (45.7%)' and 'Dedicated call recording equipment (20.7%).'

However, simply storing audio files in the cloud does not guarantee searchability for necessary data. To effectively utilize it in practice, it is necessary to not just 'store' audio but also to prepare a system environment that allows quick access to target data.

For what purposes is this vast accumulated data being used?

When asked, "For what purposes do you use customer call recording data?", the most common response was 'Improving response quality/evaluating operators (50.3%),' followed by 'Analyzing and taking measures against complaints (45.0%)' and 'Education/training (e.g., new employee training) (34.5%).'

In addition to quality improvement and complaint management, about 30% of companies use it for 'education/training (e.g., new employee training),' indicating that recorded data is used not just as a record but also for human resource development. While such utilization progresses, the 'inability to quickly find necessary data' appears to be a major barrier (challenge) to business improvement.

What challenges are actually being felt on the front lines?

When asked, "What are the current hurdles (challenges) in utilizing call recording data more effectively for business?", the most common response was 'Difficulty finding the necessary audio (31.6%),' followed by 'Cannot visually confirm because it is not transcribed (28.6%)' and 'It takes time to grasp the content (28.2%).'

Issues related to searchability and content comprehension, such as 'difficult to find' and 'cannot visually confirm because it is not transcribed,' top the list, suggesting that the necessary voice data cannot be accessed quickly. No matter how advanced the analysis tools are, if the searchability and visibility of the underlying data are insufficient, it will be difficult to fully utilize the accumulated information in practice.

Based on these challenges, we asked about the need to further strengthen AI utilization.

Finally, when asked, "Do you feel the need to strengthen the use of AI for processing and analyzing customer call recording data in the future?", a combined total of about 90% responded that they feel the need, with 36.9% saying they 'strongly feel it' and 55.4% saying they 'feel it to some extent.'