Data, this small thing that pervades all of human society, might just be like having the power of the sun in the palm of someone’s hand.
However, like most powerful things, humans have fibbed data to the point of causing losses in the thousands and millions every year- this number is higher for organizations that deal with more data. And that produces more data in its quest to create more value for itself, which is, in effect, all companies now.
Data is currency.
But because of this wild chase (a goose chase for some, perhaps?), data is abundant, and meaning-making has been lost. It is this meaning-making, or stories, that exacerbates the problem.
Data without context is just symbols on a wall that no normal person can decipher. Of course, you need a linguist- these hieroglyphs cannot be broken by people who rely on dashboards to give them meaning.
No, these are experts who have developed intuition. And you have one on your team- they are always the one who asks: hey, have you guys noticed this thing? Or the ones who just know why the GA4 dashboard looks the way it does.
They are storytellers; natural data analysts who have a knack for breaking down silos, digital and physical. And they would be needed to bring context to data.
Here’s how you find that person, and know what they and you are looking for.
The meaning of data is decided before it reaches the dashboard.
A metric looks objective because it arrives as a number. But someone still decided what would count.
Consider a lead. Does a person become one after downloading an ebook, requesting a demo, matching the ideal customer profile, or showing a combination of intent and fit? Do two people from the same company represent two leads or one account? Are existing customers included? What happens to students, competitors, and fake form fills?
By the time “leads increased by 30%” appears in a report, all of these decisions have been compressed into one line.
The same happens with customers, active users, pipeline, churn, engagement, and marketing-sourced revenue. These are not objects the organization simply discovers in a database. They are definitions built around how the business operates.
This is why a silo is not only a place where data is stored.
The CRM reflects the sales process. Product analytics reflects the events engineering chose to track. Advertising platforms measure the interactions they can see. Finance recognizes revenue according to a different set of rules. Each system can be accurate and still describe a separate version of the business.
A warehouse can connect the records. It cannot settle the argument about what they mean.
Gartner estimates that poor data quality costs an organization USD 12.9 million a year on average. Importantly, Gartner does not treat quality as accuracy alone. Data must also be usable and applicable to the purpose for which it is being used.
A click may be accurately recorded. It can still be poor evidence of demand. A login may be accurate. It can still be poor evidence that a customer received value.
The data is not necessarily wrong. The conclusion is larger than the data can support.
The cost appears somewhere else in the business.
Organizations rarely see a line item called “loss caused by contextless data.”
They see salespeople spending time on people who were never likely to buy. They see marketing scale a channel because the attribution model rewarded the final click. They see product teams build for visible activity while customer success deals with accounts that never adopted the product. They see forecasts miss, inventory sit, customers leave, and teams repeat work.
Every function can improve its metric while the organization becomes less efficient.
That is the deeper problem with silos. They separate the measurement from the consequence. Marketing owns lead volume, sales absorbs poor qualification, finance sees the acquisition cost months later, and no one owns the complete chain.
Sometimes the loss becomes visible. IBM points to Unity Technologies, which disclosed in 2022 that bad data ingested into its advertising systems had damaged the performance of its machine-learning models and contributed to approximately USD 110 million in lost revenue.
The failure did not begin when the model produced a bad result. It began when data entered a system without enough understanding of how it would shape the result.
This is why data ownership cannot end with collecting, cleaning, or storing information. Someone has to remain responsible when that information becomes a business decision.
Storytelling is part of the analysis.
Data storytelling is often reduced to presentation. Add a cleaner chart, simplify the slide, and give the executive a memorable conclusion.
But the real work happens before the presentation.
A storyteller has to explain what produced the number, which comparison makes it meaningful, what changed around it, which explanation is supported, which explanation is only possible, and what the organization should do next.
This does not make the analysis less rigorous. It exposes the reasoning inside it.
If revenue increased because prices rose, the story is about pricing power and possible retention risk. If it increased because more customers bought, the story is about demand and acquisition. If one large contract created the increase, the story is about concentration. The number remains 8%, but the business reality changes.
Florence Nightingale understood this distinction long before organizations had dashboards.
During and after the Crimean War, she worked with inconsistent military records to understand why British soldiers were dying. She did not merely count the deaths. She separated deaths caused by battle from those caused by preventable disease, compared military mortality with civilian mortality, and showed what happened after sanitary conditions improved.
Her diagrams gave the evidence a sequence: what was happening, why it was happening, and what reform could change.
Scientific American’s account of Nightingale’s work explains that the diagrams reached the press and Parliament and helped advance sanitary reform. Their power did not come from making the data beautiful. Nightingale understood the conditions behind the records and stayed with the argument until the institution acted.
That is data storytelling.
It is not finding a persuasive narrative after the analysis. It is carrying the context of the evidence into a decision.
AI can make incoherent data sound certain.
AI makes this responsibility more urgent because it removes friction from interpretation.
Ask an AI system which marketing channel generated the most revenue and it can produce an answer in seconds. But it still needs to know whether the organization believes in first-touch, last-touch, multi-touch, or incrementality. It needs to know whether revenue means bookings, recognized revenue, or customer lifetime value. It needs to know what the systems could not observe.
Without that context, AI does not resolve the disagreement. It selects from it, blends it, and presents the result in coherent language.
The coherence is dangerous because it can make an unresolved business question look settled.
RAND’s research into failed AI projects found that the causes frequently began outside the model. Stakeholders misunderstood or miscommunicated the problem, organizations lacked the necessary data, teams optimized the wrong metric, or leaders pursued the technology without a durable business reason.
And the NIST AI Risk Management Framework warns that data can become detached from the context in which it was created or become stale in relation to where the system is used.
AI-ready data is therefore not simply clean data.
It is data with a known origin, a definition, an intended use, a limit, and a person who can explain why it should influence the decision.
Otherwise, AI helps the organization act on its confusion faster.
Businesses need storytellers because the customer does not live in one system.
A buyer may see an advertisement, read an article, hear about the company from a peer, attend an event, return through search, speak with sales, and purchase months later.
Every platform records the part it can see. The advertising platform claims the click. Web analytics claims the session. Marketing automation claims the form. The CRM records the opportunity. Finance records the revenue.
None of them records the customer’s complete reasoning.
Marketing cannot solve this by choosing one platform as the truth. It needs someone who can connect channel data with sales conversations, customer research, market conditions, product usage, and commercial outcomes.
This person may discover that a campaign with weak attribution changed how target accounts described the problem. Or that a campaign with excellent lead volume attracted buyers who could never use the product. The dashboard shows performance. The storyteller explains whether the performance mattered.
McKinsey calls a related role the analytics translator: someone who connects technical teams with domain experts and helps analytical work create operational impact.
But translation is not enough if responsibility ends after the meeting.
The storyteller should help define the metric, understand its source, challenge its limits, bring the relevant business context, recommend an action, and return later to see whether the action produced the expected outcome. If it did not, the story has to change.
That feedback is what keeps storytelling from becoming opinion.
It also turns ownership into something practical. The storyteller does not have to manage every database. They own the integrity of the conclusion the organization draws from it.
The business needs fewer unexplained numbers.
Organizations do not lose money because they have no data. They lose money because different teams can use the same data to support incompatible decisions, and no one is responsible for resolving the difference.
AI will not resolve it on its own. Marketing cannot avoid it by adding another attribution tool. And a centralized warehouse does not create a shared understanding of the customer.
The organization needs people who understand the systems, the business process, and the consequences of being wrong. People who can explain what the data proves, what it does not prove, and what should happen next.
A storyteller gives data its business context.
Ownership ensures the organization learns after it acts.
