ABSTRACT: Explore the essential characteristics to choose the right conversational query tool for your needs and environment.

Sponsored by QuaerisAI 

While colorful charts dazzle the boardroom, many business managers still lack the hard, fast facts they need to make tactical decisions or plan strategies. Business intelligence (BI) teams just don’t have the cycles to field all their ad-hoc questions. 

Conversational queries can help. This emerging category of business intelligence (BI) uses generative AI to provide real-time search for and retrieval of tabular data through a natural language interface. It complements traditional BI capabilities that measure, predict, and visualize market trends or key performance indicators (KPIs). Conversational queries enable business managers with minimal data expertise to rapidly assemble facts as they make tactical decisions, forecast performance, and brainstorm ideas. They respond to human questions such as the following:  

• What was the revenue and revenue growth last week for stores in Denver, by product?  

• What are the average inventory levels this month to date for our UK warehouses? For which products has inventory fallen? 

• Which customers bought the most books on Cyber Monday and what categories had the most change? 

Conversational query tools—i.e., BI tools that include these features—can empower business managers to answer their own questions rather than waiting on the assistance of busy data analysts. Reflecting this, a recent survey by BARC and Eckerson Group showed that 77% of business managers believe GenAI will improve the use of BI in their company to a moderate, high, or very high degree. Such self-service capabilities have the potential to create new analytics value, increase productivity, and reduce costs. 

Must-have characteristics 

To deliver business value, a conversational query tool must be multipurpose, intuitive, open, governed, and fast. While these adjectives apply to many new technologies, the details vary in meaningful ways. Let's explore each must-have characteristic to help you select the right conversational query tool for your use cases and environment.  

Multipurpose. To solve hard problems, business managers need more than chatbot responses. They need contextual information and reliable sourcing. A conversational query tool must pull up trusted facts from operational tables for finance, sales, or other functions. Then it must enrich those outputs by retrieving relevant documents based on keyword searches. Multipurpose capabilities like these enable decision makers to review numbers quickly and explore the underlying factors. 

Intuitive. A conversational query tool also must engage and respond to humans in ways they naturally understand. This requires learning domain-specific terminology, related to a given function or vertical, and fine-tuning the NLP or GenAI models based on user feedback. It also requires an appealing user interface that couples the chatbot with charts or other visual illustrations of outputs. Above all, an intuitive conversational query tool should suggest new interpretations and questions, guiding the business user through an iterative analytical process. 

Open. Analytics often involves inputs that span a variety of data stores, formats, and APIs. Users might need to explore and analyze sales records within a customer relationship management (CRM) database, shipment reports within a supply-chain management (SRM) database, or product documents within a content management system. An open conversational query tool provides unfettered access to tables, text, and documents across this ecosystem. 

Governed. As with any analytics project, conversational queries pose risks to privacy, compliance, and of course accuracy. Your tool should mitigate such risks by enabling administrators to oversee and control user actions, data retrieval, and data quality. It can do this with governance features such as role-based access controls, masking, observability, or audit logs for compliance reporting. Also ask how each tool can minimize erroneous outputs such as hallucinations. 

Fast. The interactive, iterative nature of conversational queries demands real-time retrieval of data and delivery of outputs. These tools should meet service level requirements for performance as measured by response time and data throughput. Also assess whether your conversational query tool supports sufficient numbers of concurrent users and data sources. 

To deliver business value, a conversational query tool must be multipurpose, intuitive, open, governed, and fast 

Getting started 

Perhaps the most compelling possibility created by generative AI is the ability to converse with software in the native language of business rather than the programming language of SQL. Conversational query tools capitalize on this possibility and reduce the need for business managers to wait for refined BI reports. You can enable your team to get started by choosing a tool that is multipurpose, intuitive, open, governed, and fast. To learn more about conversational queries, watch me tackle this topic with Rishi Bhatnagar, CEO of QuaerisAI, in a LinkedIn Live webinar that BARC hosted on July 10.