By Hiran de Silva
Every few months, a familiar headline resurfaces: “95% of spreadsheets contain errors.” It’s become a mantra in boardrooms, consulting pitches, and vendor white papers. The implicit conclusion is always the same: spreadsheets are unreliable, dangerous, and perhaps should be replaced altogether.
But let’s pause and test the logic.
The Motorist Analogy
Imagine a study that claimed 90% of motorists are dangerous. They don’t know the road rules, can’t handle their cars, and cause pile-ups. If that were true, would the rational conclusion be to ban cars altogether and make everyone take the bus?
Or would the sensible response be to ask why so many motorists are incompetent in the first place? Perhaps they were never trained to drive on public roads, never tested, never licensed. Maybe they’re still driving as though they were in a private car park, where rules don’t matter.
That second explanation sounds much closer to reality. And when you apply it back to spreadsheets, the parallel is exact.
Enterprise Spreadsheets Are Not Private Toys
In business, we are not talking about somebody’s personal shopping list. We’re talking about spreadsheets that underpin corporate processes, drive decisions, and sometimes move billions of dollars. These are not private playthings. They are public-highway assets: they must conform to standards, be robust, and operate within an environment of shared accountability.
But here’s the rub: most spreadsheet creators have never been taught the difference between a personal spreadsheet and an enterprise spreadsheet. They’ve never been trained, never tested, never “licensed” to operate at enterprise scale.
And why? Because virtually all of the training available—whether online courses, YouTube tutorials, or social media tips—is aimed at the single user, building a spreadsheet for personal use. Not one of them equips people to design for enterprise architecture.
So when a study declares that 95% of enterprise spreadsheets contain errors, the conclusion is technically correct but fundamentally misinterpreted. Of course errors abound—because the people building them have never been introduced to the road rules of enterprise design.
The Dishonest Spin
Here’s where things become troubling. The “95% error” statistic is often used as a weapon—an argument that spreadsheets themselves are inherently flawed and should be replaced by alternative tools.
That’s like saying: “Most motorists are dangerous, therefore cars should be abolished.” It’s not just illogical; it’s misleading. The real problem isn’t the car—it’s the lack of competent drivers. The real problem isn’t Excel—it’s the lack of competent enterprise training.
The Real Issue: Architecture, Not Power Features
Another red herring is the focus on Excel’s so-called “power tools”: Power Query, Dynamic Arrays, LAMBDA. These are brilliant—but they’re still designed with the single user in mind. They do not in themselves solve the architectural challenge of spreadsheets in enterprise systems.
Enterprise spreadsheets require something different: robust architecture. Standards. A client-server mindset. Hub-and-spoke models where data is separated from reporting. These are proven, scalable methods—well within the capability of Excel when designed properly.
What the 95% Really Tells Us
So, what should we conclude from the “95% of spreadsheets contain errors” mantra? Not that Excel is broken. Not that spreadsheets should be banned.
Rather:
- Most enterprise spreadsheets are built by people without training in enterprise architecture.
- Most training available focuses on single-user use cases, not enterprise standards.
- The solution isn’t to abolish spreadsheets, but to professionalize their enterprise use.
Just as motorists need licenses to drive safely on public roads, enterprise spreadsheet creators need education, standards, and testing to deliver reliable models for business processes.
Until we acknowledge that, we’ll keep misusing the 95% statistic—treating it as a weapon against Excel instead of a wake-up call about training, competence, and architecture.
👉 Takeaway: The next time you hear someone quote the 95% error statistic as proof that “Excel doesn’t work,” ask them this: Are you blaming the car—or the driver?
Add comment