Beyond the numbers: Turning data into decisions that drive actions

Anne leaned back in her chair, eyes scanning the quarterly dashboard. Sales were flat in two regions, yet inventory was piling up in one and disappearing too fast in another. Completely different from expectations and data from the last decade. “Why now?” she muttered. Before she even finished the thought, a text message from her LLM agent flickered her phone’s screen: “Promotions launched two weeks ago performed unevenly due to the southern region experiencing a heatwave earlier this year, reducing demand for the promoted outerwear, while the northern region saw a more windy April than usual, driving up sales beyond forecast. Combined with regional style preferences that weren’t accounted for, the result was a supply-demand mismatch. Suggested action: realign marketing offers and shift stock between stores 12 and 16.”

Anne smiled. That would have taken days of meetings last year…

But wait, we are not here to read science fiction, aren't we? 

Anyways, we’re long past the era when data was scarce. Today, most businesses are swimming in it — customer behavior logs, transaction records, CRM exports, web analytics, IoT sensor streams… Following best practices, companies are going to great lengths to ensure that data is regularly backed up and access to it is secured. It will not be lost and, what is most important, it will not be stolen by unauthorized people. Hey, wait, but that is not what is most important, isn’t it? 

Numbers and visions…

Of course not, the most important thing is to analyze the data. So, what smart companies do is invest in analytics. 

Avinash Kaushik, the author of the leading research & analytics blog Occam's Razor, often brings up an interesting anecdote. Once he decided to turn off the analytics reporting that was regularly being sent to decision makers in the company he worked for at the time. No one noticed for three weeks. Obviously, at least during that time, decisions were made and business operated as usual without decision-makers looking at the analytical data at all. So, turning the analytics back on was important, but not as important as putting the data to use. Or why bother to analyse it at all? 

On the other end of the spectrum, Alistair Croll and Benjamin Yoskovitz, in their book “Lean Analytics: Use Data to Build a Better Startup Faster”, after interviewing many startup founders, discovered that many of them, even though they take the data-driven approach, realise that they don’t want to rely on the numbers alone. They’re uncomfortable with their companies being optimized in a way that feels soulless, with numbers but without a vision, and recognize the importance of stepping back to consider the “bigger picture”. For example, retailers today have more access to customer data than ever before. Every scan, click, and return tells a story. But as “Lean Analytics” reminds us, metrics are only useful if they drive behavior. Retailers must start by asking: What problem are we trying to solve, and what does success look like? Without that north star, even the cleanest data leads nowhere.

Can we draw a paradoxical conclusion that both those who ignore the analytical data and those who overanalyze it reason similarly? In a way, yes. Both schools of thought come to similar conclusions, and that is that the “data-informed” approach makes much more sense than the solely “data-driven” approach, especially when it comes to making decisions about future steps. The difference is in the level of the “informed” part. Some decisions are, as in Kaushik’s anecdote, obviously made without the data, which remain in the vague FYI realm, and that is a pity indeed. And, wait a minute… the data analysis itself was not what’s most important — it’s how to make use of the data so that you and your team make better decisions and act on those decisions more efficiently!

… and actions! With the little help of LLM agents

Every once in a while, a technology emerges that has the potential to change how companies do business. Many believe that AI is one such technology, but still struggle to make use of it that goes beyond summarizing texts and answering questions from a knowledge base. Gartner, a leading global research and advisory firm, forecasts that by 2026, roughly one-third of new applications will incorporate AI—up from less than 5% today. But what does that mean for our use case? Can AI help companies analyse the data, but also to decide faster and act upon those decisions more efficiently and without losing the vision and “soul”? 

The short answer is yes. Infusing core company processes with AI applications can bring about a foundational change. For example, if a retailer’s strategic goal is to reduce stockouts by 20%, that objective should drive what data is gathered—from shelf-level inventory turnover to supplier lead times and regional demand spikes. Similarly, if the goal is to increase customer lifetime value (CLV), then collecting data on purchase frequency, product preferences, and engagement across channels becomes a priority. AI systems trained on this purpose-driven data can then deliver high-impact insights: forecasting demand with precision, recommending tailored promotions, or optimizing inventory allocation across stores. 

When AI operates in the service of clearly defined goals, data moves from being a passive asset to an active enabler of strategic progress and informed decisions. And, what is most important, that help is not just in creating nicer reports, it is proactive collaboration on more levels - management for sure, but also operations and sales. LLM agents can now actively participate in day-to-day operations, alerting warehouse managers when inventory levels dip below forecasted thresholds, or automatically reordering stock based on real-time demand and supplier availability. Sales associates can receive instant product recommendations or customer insights during interactions, enabling more personalized service on the floor. In logistics, AI can dynamically adjust delivery routes in response to traffic or weather changes, minimizing delays and costs. At the management level, LLM agents can surface emerging trends, flag anomalies in performance, or even simulate outcomes of strategic decisions using predictive modeling. 

The real power lies in agents that don't just inform, but perform real actions such as updating product listings, sales contracts, triggering marketing campaigns… This isn’t just analytics support; it’s an intelligent layer woven into the entire retail fabric, helping people make better decisions and act on those decisions more efficiently. 

And that’s reality, not science fiction anymore. Get in touch with us to learn how LLM agents can help your business!