An enterprise can have useful reporting and still make it hard for employees to find the right view. In the account Centurysoft published about its work with Unilever, the challenge was partly a Tableau knowledge gap: employees needed an easier way to locate reports and assess performance across the organization.
The project connected a conversational assistant to a defined set of interactive Tableau reports. Rather than asking every user to know the reporting structure in advance, the experience allowed people to request a view in natural language.
The challenge: making a reporting estate usable
Unilever’s stated requirements covered three related needs. Employees needed support when they did not know Tableau well. They needed to find and assess performance across locations. And the reporting experience had to fit the organization’s structure, so that a question could lead to an appropriate view rather than a long search through reports.
This is a common tension in enterprise analytics. A reporting team may organize dashboards logically, yet the person with a business question does not necessarily know the report name, the available filters or where the relevant measure lives. The interface to the information becomes part of the problem.
The work: a conversational route to Tableau reports
Centurysoft’s published account describes a SILVIA-based AI assistant deployed to help users locate Tableau reports through natural-language processing and contextual understanding. It also describes a chatbot with speech-to-text and text-to-speech functions, giving employees a spoken as well as typed way to request reporting.
The team defined interactive Tableau reports, with five drill-down levels, covering different aspects of sales and orders. The assistant was integrated into an organizational portal, so employees could ask for a report without starting from a dashboard directory. A request such as “Show sales report for Boston” illustrates the intended interaction: the person states the business question, and the interface routes them toward the relevant view.
The combination is important. A conversational layer alone cannot solve a poorly defined reporting model. The reports and their drill-down structure provide a specific information set; the assistant changes how users reach it. Both pieces need to make sense together.
The Benefits:
• Unilever’s teams received an AI-enabled chatbot within their organization portal, for reporting and general queries
• Employees can call up the six defined Tableau reports using natural-language requests such as “Show sales report for Boston”
• The assistant can interpret complex data, capture trends and identify patterns
• It understands both speech and text, and reacts to the sentiment conveyed
What the published account supports
The documented outcome is a working access path: an assistant in the portal, natural-language and voice interaction, and a defined set of Tableau reports for sales and order analysis. The available account does not establish how many employees used the service or how much time it saved.
The original write-up describes broader possibilities for recognizing patterns and supporting analysis. It does not publish adoption numbers, time savings or a before-and-after measure. Those figures would require separate evidence before appearing in a current case study.
What this work illustrates
Business intelligence becomes more useful when people can ask a familiar question and reach an appropriate report without learning the structure of every dashboard. This project paired an accessible interface with a finite, organized reporting set. For leaders considering a similar approach, the first question is which recurring questions employees struggle to answer today, and whether the underlying reports can answer them consistently.
