Your Data is Your Most Defensible Moat

Let me set a scene for you. A frustrated CEO storms into a cluttered and dimly lit space, “Hey, we’re suddenly losing customers, what is happening?” The camera pans to two analysts, hunched over littered desktops. One replies, “I can set up a dashboard, but it’s going to take me a while.” Crickets.

Suddenly, an overhead light comes on and the music swells, angels are harmonizing, the other analyst has typed into a chat “explain churn increase by segment.”

As the chat begins to scroll, relief washes over the analyst, “the largest driver is onboarding delays” he says. The CEO comes in close, at this point all three are now looking directly into the computer, “what can we do about it?”

“Let’s find out” the analyst says while typing “recommend next best action” into the chat.

We pan out, the basement scene is gone, the desks are uncluttered. The team of three are still staring at the laptop. Suddenly a notebook begins to levitate, there’s a sound effect reminiscent of magic, a tape dispenser has left the table, so has a mug.

There’s a quick cut and now we’re in the boardroom, a slide deck sits in the background titled “the churn spike was not a mystery — it was a handoff failure.” The analysts describe what they’re going to automate next and that a new skill has been made for future self-service.

Our narrator proclaims “With the agentic enterprise, everyone becomes a strategic force, shaping what the business does next.”

Nodding and clapping ensue, the problem is solved, we fade out.

The scene I just described is the fairy tale dream that every data professional and CEO is being sold right now. A dream on display in the 2026 keynote speech at Snowflake’s customer conference. Where their CEO also proclaims “your data is your most defensible moat. It’s what makes your company truly your company.”

Looking at this glossy tech demo, I have one question I want to ask the team. How exactly did the CEO find out that they were losing customers? How was he alerted?

The how behind his discovery, what could help orient us, is crucially missing from the scene. Did he see a number somewhere, did he get an angry call from an important client, did a family member try to sign up and leave confused? How did he gain awareness of the situation?

And then there’s this question. Why doesn’t the CEO already know this answer? Why don’t the analysts? (Selfishly I also wonder why the analyst is suddenly relieved when the LLM returned a succinct answer.)

If we are to believe what we are witnessing, at this fictitious organization, they have no current tools to observe the business process of onboarding customers. Not only do they not have current tools, apparently the CEO is the first line of defense in alerting everyone else, he’s the human at the company who notices it first.

The lesson here is one about situational awareness in the business. The absence of which is on full display in this fantasy. It is also the exact thing that question and answer analytics have never been able to resolve.

The exercise of asking a question and receiving an answer in business isn’t new. It is the level of maturity we’ve reached with data in the organization, and that maturity level is shockingly low.

Consider what is at the heart of self-service analytics: non-technical folk will be able to query organizational data and get a numerical result. We can confidently say that the fashion and style have changed. What was originally unsexy data dumps became submitting tickets to the data team. In 2026, the leading style has now become asking an LLM.

But looking past the fast fashion, the core output is the same. Receiving the answer faster doesn’t solve the problem. It doesn’t prevent this from happening again. It is reactive, and in this iteration, it quietly allows many businesses to no longer cosplay at being data-driven and instead fully embrace reaction mode.

The stuff that aids in resolving the problem is presented in that same fictitious boardroom — identifying early churn signals, triggering interventions, and establishing proactive outreach.

Identifying early churn signals is the task of gaining situational awareness. And here we can absolutely rely on data. The data can help us both construct indicators and develop an analytical method of turning those indicators into a relevant signal. It is the dashboard recommendation that resolves to silence.

Triggering interventions is similar in concept, but different in execution. Here we’re moving beyond situational awareness and instead instigating a business process change or likely net-new process. Instead of customers slipping through the cracks, we can deploy a ladder of action they can take to climb back up. Maybe each rung of that ladder is deployed automatically.

The last solution is rooted in human behavior change. Because apparently at this organization, outreach has always been reactive. We can tell a story that perhaps nobody thought it was their responsibility, but the real story is that there was no way to know who needed said outreach. What was missing all along: situational awareness paired with the ability to dispatch team members to prevent customers from slipping in the first place.

What we’ve seen is a demonstration of exactly how out of touch the technology has become from the business itself. The arrogance with which technology vendors believe they have earned a place to dictate strategy to the business simply because they have a copy of all the data.

The fantasy is so magical that you can blink and miss it. It even includes the CEO and board members as accomplices. But stop for one minute and think to yourself, are the two analysts from the basement really going to be invited up into the boardroom to confidently announce a “handoff failure?” Is the board really going to nod and say great work? The board whose members include Mark, Head of Sales, the guy responsible for the handoff failure and also your Friday golf buddy?

I think levitating office supplies is more likely.

In this particular instance, the technology vendors, and the data leaders they’ve incepted, are selling you a sea of ephemeral charts, metrics, and generated insights. A bland next best action recommendation spit out into a gorgeous slide deck. Shiny new artifacts that won’t move the needle on why you’re losing customers. A fault not to be credited to AI, but to be credited to those who are currently poised to control it.

Not only that, they’re also implicitly asking you for your permission to lock up the data more tightly. They want your buy-in and approval to spend more time writing down what each column of data means, meticulously mapping the relationship between data points, and making sure documents get written so that AI agents can understand your business.

Their marketing rhetoric is also whipping you into an unnecessary state of urgency. Claiming that if you don’t do this work now, your competitors will win. Work shown to yield a 21% accuracy rate when answering data questions.

Make no mistake, there is a lot of work to be done. And AI does make that easier, just not in the fantasy version you’re being sold.

But to do that work we’re going to have to start asking hard questions. We’re going to have to sit down and reimagine business. Not fix, not modernize, not migrate, reimagine. Redesign the process around the full loop: detect the signal, interpret what it means, make a decision, intervene, and learn from what happens.

We’re going to have to put our own egos aside. We’re going to have to recognize how blinding power and the hierarchy can be. Politics can no longer be waved off as the way things are. It is part of the business, and belongs in the solution.

Everyone is going to have to open themselves up to that uncomfortable phrase “I don’t know.” An uncomfortable phrase that can be made more comfortable with a simple redesign: “I don’t know yet, but I can find out.”

A phrase that needs to be backed up by skills that we have sorely underinvested in. Skills that include the ability to spot patterns, to ponder what those patterns mean, to learn from mistakes, to imagine future scenarios, and to develop tools that improve our awareness.

Today, those who are best equipped with these skills do in fact exist in the business, and oftentimes within analytics. Not merely the people who nerd out over tidy queries or copying data to a single repository, but the people who are relentlessly curious. The two that couldn’t provide a quick answer, but could provide two grounded ways of how to get started: the one offering to build a dashboard and the one wanting to understand churn by segment.

The two team members who are being presented as basement dwellers. Check they haven’t been managed out. Or left. Their work is your actual moat.