From reading the room to changing it: why real digital intelligence is about better decisions, not more data

By Jarlath Mulhern, Senior Data Scientist 

Over the past eighteen months, what we can measure has transformed beyond recognition. Large language models, once the preserve of research and technologists are now embedded in everyday life with no industry left untouched, the insights and communications world included. Techniques for the analysis of news, social media, policy documents and global messaging that once might have taken weeks can now run in an afternoon.

Data platforms can surface conversations quickly, group themes together, and track how they evolve in real time. What we can identify, we can now also measure at a scale and speed that would have been unthinkable even a few years ago.

While this is a huge leap forward, a deeper sense of understanding hasn’t naturally followed. We can now measure any and everything definable within language itself. The harder question is whether what we’re measuring what matters and what to do about it.

Describing conversations isn’t the same as understanding people

Social media monitoring can map what’s being said, how often and by whom. While this is useful, it’s not the same as understanding why something is being said.

When United Airlines forcibly removed a passenger from an overbooked flight in 2017 the internet collectively responded with outrage, sarcasm and memes. Despite the backlash, sentiment tools still registered largely positive scores, mistaking waves of ironic tweets for genuine praise.

Data was moving faster than anyone could interpret it, and by the time the picture was clear, crisis comms were already behind.

Since then, modern tools have gotten better at reading irony. But better isn’t the same as reliable, and companies still risk falling into the same trap. The United Airlines data wasn’t wrong; it was just unfinished.

Turning data into intelligence means deciding what’s worth measuring, adding structure to the analysis, and interpreting what the evidence actually means for a specific organisation, audience or decision. Most automated tools stop well before that point. 

Visibility isn’t the same thing as impact

A data platform can tell you how many impressions a story generated but it can’t tell you whether that story changed minds or behaviour.

Influence isn’t evenly distributed. Some voices and audiences carry more weight than others. Conversation doesn’t move through volume alone; it moves when the right people say the right things in the right way.

Understanding who those people are, how they’re connected and what they respond to requires understanding who actually shapes the decisions that matter, not just monitoring feeds.

Not all narratives are created equal

Themes and patterns are often presented as insight in their own right, but a pattern is only useful when you know what it means, who it matters to, and whether it has any bearing on the decision in front of you.

Some narratives shape decisions, some create resistance and some are simply background noise. The work is knowing which is which, linking narratives to the stakeholders who carry them, the incentive structures behind them, and the outcomes they might plausibly affect.

AI tools don’t turn data into decisions (people do)

Tools have never been more powerful, but AI still can’t sit in a room with you, understand your priorities and map them onto a stakeholder ecosystem. It cannot reliably distinguish between what’s interesting and what’s commercially, politically or reputationally important. A dashboard can surface patterns, but it cannot weigh the significance of a narrative, understand the context behind a conversation, or judge what a spike in activity actually means for your organisation.

This requires human expertise, the kind that knows what questions to ask, what the evidence is actually saying and what to do about it. This is where automated analysis consistently falls short.

And it’s where Madano’s digital intelligence starts

At Madano, we start with the business problem and work outwards. This means:

• Defining the landscape in a way that reflects the decisions you need to make
• Mapping influence across the stakeholders who shape those decisions
• Understanding how audiences think, and where messages will land or be challenges
• Adding qualitative judgement back into analysis, so scale doesn’t come at the expense of meaning
• Evaluating performance in a way that feeds back into strategy, not just reporting

Going beyond reading the room to knowing how to change it takes data science and consultancy in equal measure. One without the other doesn’t get you very far.

Data is easy to find. Clarity in a complex world takes work.