Last week in the article, “Applying Good to Great Principles to AI Adoption”, we related Jim Collins’ Good to Great chapter on technology accelerators to modern AI. To illustrate how the principles apply to AI, let’s consider the example of a hospitality company aiming to improve its sales insights through AI.
Collins’ recommendation is to focus on what you can be “the best in the world at.” By identifying the customer acquisition cost KPI as the first use case, the company was aligned and was able to focus on turning an already effective process into an exceptional one.
Setting Target Thresholds for AI Precision
Once the organization identified a precise use case, optimizing customer acquisition costs, they were able to set goals and expectations on the outcome of AI project. The company set a clear target threshold, using AI to improve upon existing results from manual or semi-automated processes. For example, if their traditional methods achieved a 70% accuracy rate in tracking customer acquisition metrics, the goal for AI was to surpass this baseline—say, achieving 80% accuracy. By setting a tangible threshold, the company could directly measure the value of AI, ensuring it wasn’t implemented for its own sake but to truly improve a core business process.
This aligns with the second piece of Collins’ Hedgehog Concept, ensuring the use of new technology supported “what drives their economic engine.” By setting clear performance targets from the outset, the company was able to track ROI on the customer acquisition project in a concrete way. Demonstrating measurable returns on this initial project not only validated the investment in AI but also built confidence for future data and AI initiatives within the organization, creating momentum for long-term growth and innovation.
Supporting Sales Queries with Conversational AI
To make sales data more accessible, the company used a conversational AI tool that allowed executives to ask specific questions, like “What were our sales yesterday?” or “How did our sales compare last week?” This tool could recognize context and maintain a “stream of consciousness,” understanding follow-up questions like “What about the following week?” without requiring restated context. This conversational interface opened up sales data to less technical users, making it possible for leaders to access insights without learning how to navigate a dashboard or SQL query.
This approach demonstrates Collins’ emphasis on making it about, “what they are deeply passionate about.” By providing business users with an intuitive, conversational way to access data, the AI tool empowered leaders who were eager to make informed, data-backed decisions but previously found traditional analytics tools inaccessible or cumbersome. With this new interface, users could engage more deeply with the company’s metrics, strengthening their confidence and enthusiasm for data-driven strategies. This passion for informed decision-making helped drive adoption and set the stage for a culture that values insights grounded in reliable data.
Ensuring Data Accuracy with Large Language Models (LLMs)
Although LLMs offer powerful conversational capabilities, they sometimes lack the precision required for business analytics. To address this, the company balanced LLM-driven conversational ease with structured data checks to maintain accuracy for critical metrics. For example, while the LLM could handle general questions in a natural tone, specific questions about sales figures were routed through a structured query system to ensure accuracy. This dual approach allowed the company to benefit from the ease of LLMs while safeguarding the precision necessary for data-driven decisions.
Image Generated with AI (DALL-E via Microsoft Bing on 2024-12-20)
This balance reflects Collins’ warning message on “confronting the brutal facts.” The organization recognized that, while conversational AI had potential, LLMs alone might not meet their exacting standards for accuracy in sales analytics. By addressing this challenge head-on, they could implement a more reliable solution that met both accessibility and precision needs by focusing on the underlying data architecture and modeling necessary to an effective AI solution.
Conclusion
Through targeted, thoughtful AI deployment, this company not only improved its sales process but laid the groundwork for a scalable, sustainable AI infrastructure. This strategic approach embodies Good to Great principles, leveraging AI as an accelerator of existing strengths rather than as a distracting novelty.
Datalere consultants specialize in data architecture, AI integration, and strategic guidance to support you at every stage of your AI journey. Let us help you create a smarter, data-driven future—contact us today to get started!
