How we democratised access to business metrics through conversational AI
The challenge
In telecommunications, data moves fast. Churn, ARPU, new sign-ups, number portability, NPS… everything changes week by week (and sometimes day by day). And when a business needs to respond quickly, relying on monthly reports or a technical team to answer basic questions can quickly become a bottleneck.
Our client, a leading telecommunications operator, was facing a very common situation: they had a huge amount of data and KPIs available, but accessing them was not as straightforward as it should have been.
The key question was clear:
How can any team access key KPIs in real time, without relying on complex dashboards or the BI team?
Before implementing the solution, several recurring challenges remained:
-
Dependence on technical teams: many requests ended up with the BI team or data analysts, even for relatively simple questions.
-
Delayed information: traditional reports were generated on a fixed schedule, making it harder to react to rapid changes in the business.
-
Technical barrier: not everyone felt comfortable interpreting dashboards, navigating BI tools, or working with SQL.
-
Fragmented data: KPIs were spread across multiple systems: CRM, billing, network, commercial platforms, and customer service.
-
Lack of context: even when the data was available, it often lacked interpretation: is this good? has it worsened? how are we performing against target?
In short: the data was there, but it was not easily accessible across the organisation.
The solution: a bot that answers KPI queries in natural language
To address this, we designed and implemented an intelligent conversational assistant, integrated directly into the company’s corporate communication platform.
The goal was simple: to enable anyone, regardless of their technical background, to access business metrics just by typing a question, as if they were asking a colleague.
The bot can handle queries such as:
-
“What is this month’s churn rate compared with last month?”
-
“Give me the ARPU for the residential segment in the northern region.”
-
“How many fibre sign-ups have we had this week?”
-
“Show me the NPS trend over the last 6 months.”
-
“How are we doing against our portability target?”
And best of all: it doesn’t just return a number. It provides context and comparisons so the data is genuinely useful.
Solution architecture
The solution was built on a modern architecture designed to combine generative AI, secure access to corporate data, and a scalable query model.
The main components were:
-
Integration layer: secure connectors to data sources (Data Warehouse, transactional systems, and BI tools).
-
Conversational engine: advanced models capable of understanding natural language and translating it into structured queries.
-
Semantic layer (business glossary): translates everyday business terms (“cancellations”, “active customers”, “northern region”) into technical metrics and real dimensions.
-
Response engine:builds clear answers, including trends, comparisons, and relevant context.
- Security layer: role-based access control to ensure each user can only access authorised KPIs.
This way, the assistant doesn’t just respond: it responds accurately, with consistent and governed data.
How does a query work?
The full workflow is designed to make the experience fast and natural:
-
The user asks a question in the corporate chat.
-
The AI interprets the intent, identifies the requested KPI and the required filters (time period, region, segment, channel, etc.).
-
Permissions are validated, ensuring the user is authorised to access the requested data based on their role.
-
The data is retrieved from corporate sources to provide up-to-date values.
-
The response is enriched, adding comparisons, trends, and historical evolution.
-
The bot responds in a clear, actionable format, ready to support decision-making.
Results and benefits
Implementing the assistant transformed the way the organisation consumed information and made decisions:
→ Democratised access to data: any employee can access KPIs without relying on technical roles.
→ 70% reduction in ad-hoc requests to the BI team for on-demand reporting.
→ Real-time access, available 24/7 without waiting for scheduled reports.
→ Faster decision-making, particularly across commercial and operational teams.
→ A real boost to a data-driven culture, increasing the use of data at all levels.
→ High internal adoption, thanks to the bot being embedded in the organisation’s everyday communication channel.
Conclusion
This project highlights the potential of conversational AI applied to Business Intelligence: when you remove technical barriers between users and data, information naturally reaches the people who need it most.
Beyond the technology itself, the success of the project was driven by four key factors:
-
User experience: we brought KPIs to where teams already work.
-
Business language: the bot speaks the organisation’s language, not technical dashboard jargon.
-
Real context: it doesn’t just provide numbers, it delivers useful interpretation (comparisons and trends).
-
Built-in security: controlled access and data governance embedded from the start.
Can you imagine anyone in your organisation being able to query business KPIs just by asking a question?
Let’s talk, and we’ll show you how to make it happen.