.png)
O2, the largest telecommunications operator in the Czech Republic, interacts with tens of thousands of customers across multiple channels every day. With the No-Code NLP platform by SentiSquare, their customer service center gained full control over the quality and outcomes of customer care. In-depth analytics led to a 9-point increase in NPS and several percent increase in sales performance, all without AI expertise and using standard hardware.
O2 CZ is a leader in providing voice, internet, and data services to both households and businesses. With over 5 million calls and chats annually, plus hundreds of thousands of interactions across other channels, O2 faces an enormous volume of customer communication. This communication not only needs to be managed but also analyzed effectively to support business growth.
For managing quality, sales, and customer experience, having a detailed understanding of what’s really happening in those interactions is essential.
In collaboration with SentiSquare, O2 gained the ability to create its own AI models for communication analysis without the need for expert knowledge in AI/NLP or any programming. And the results?
We defined clear goals for our collaboration with O2. Their key requirement was a self-service solution capable of processing millions of customer communication records with high accuracy, in a short time, and without the need for special hardware or expert knowledge in AI.
Before our collaboration began, O2’s customer center was using a basic rule-based analysis, where the system searched for predefined keywords or phrases to determine the topic of the call. For example, if an operator said, “I’m offering you a new tariff,” the system correctly recorded that an offer was made, but any slight change in wording or word order could cause the rule to fail. This meant many rule combinations had to be created to cover various ways a person could express the same thing. This system was very problematic to develop and maintain.
The original approach had limitations and did not allow detailed evaluation of quality or sales opportunities in communication. Rule-based analysis cannot capture synonyms, emotions, or context. Call center managers therefore had only a limited overview of quality. A large portion of communication remained uncovered, and important interactions were not evaluated. The original solution did not offer enough flexibility to quickly adapt to team needs.
O2 was also tesing ChatGPT. In an environment with such a high volume of interactions, generative AI without targeting specific cases can easily get lost. Evaluation takes too much time and is financially costly. With the specific needs of individual teams, daily operations require a higher level of control and repeatability of outputs.
The volume of customer communication was so large that it could not be effectively covered manually, by rule-based analysis, or by generative AI. For in-depth analysis and faster response times, O2 chose our No-Code NLP platform.
In addition to rapidity, flexibility to evaluate any specific situation as needed is also crucial.
Thanks to small specialized language models (SLM), SentiSquare enables O2 to automatically evaluate specific situations that would otherwise take hundreds of hours to analyze manually. O2 currently processes analyses the following areas:
Our models are always custom-built. O2 configures them according to its specific needs without any coding. The locally deployed No-Code NLP platform enables:
The entire solution runs on standard hardware, with no need for GPU or cloud. O2 currently operates more than 30 custom SLMs within its own system, with full local control.
The SentiSquare platform now analyzes 100% of customer interactions, including offer quality, operator behavior, and promise fulfillment.
O2 currently creates and manages over 30 of its own small specialized models without any knowledge of NLP or development. Customer service managers have a detailed overview of operator performance, promise compliance, and sales effectiveness. Thanks to the platform’s flexibility, they can quickly adapt the analysis logic to meet new or changing needs.
Because O2 now analyzes 100% of customer interactions, the team has an immediate overview of whether agents follow internal processes, business rules, and legal requirements. SentiSquare enables the setup of custom classification models that automatically identify specific situations, such as unfulfilled promises, missed offers, missing identity verification, or any other procedural element.
Instead of manually reviewing records or relying on random sampling, CX or Compliance managers at O2 now have full traceability of all deviations within seconds. This increases security, transparency, and management quality while significantly reducing the risk of fines and reputational damage.
Most companies still rely on NPS surveys, which are only completed by a small fraction of customers and often with a delay. With SentiSquare, O2 gains a more accurate and objective insight into real customer satisfaction directly from the content of conversations, emotions, topics, objections, and the overall tone of each interaction.
By analyzing every single interaction without exception, O2 can monitor customer satisfaction in real time and without bias. Managers have access to detailed, actionable data that reflects what the customer actually experienced, not just what they were willing to write in a survey. This approach led to a tangible 9-point improvement in NPS.
O2’s primary goal in deploying the No-Code NLP platform was to gain full control over all customer interactions across channels, teams, and communication stages. But analyzing 100% of calls brought an unexpected benefit: an increase in sales performance. By having a detailed overview of what’s actually said in conversations, how agents handle objections, and whether they present offers effectively, O2 was able to improve exactly what drives business results. Sales quality has since become a top priority and the area where the most analytical focus is now directed.
The O2 case study shows that cutting-edge AI doesn’t have to be reserved for AI experts. With SentiSquare, O2 built a fully scalable, self-managed analytics framework over millions of customer interactions with immediate impact on CX and commercial performance.
Want 100% control over your customer communication and a measurable performance boost? Leave us your contact details and find out how SentiSquare can work for your organization.
