Why we still use spreadsheets for everything (even when they keep breaking)

Despite all the SAAS products, all the AI hype, the humble spreadsheet still dominates. Take excel for example, it’s estimated to have up to 1.2 billion regular users!

But why, what is it about the humble spreadsheet that makes them so popular, how have they survived so long?

And how indeed, have they survived so long when we know they are fragile, error-prone, and difficult to govern?

There’s a long history of documented failures including secret services tapping the wrong phone numbers, governments implementing austerity, broken financial models, incorrect reporting dashboards, misplaced formulas that quietly distort decisions for months or years. You’ve probably experienced some version of this yourself, even on a small scale.

“The biggest problem is that anyone can create Excel spreadsheets – badly,” US scholar, James Kwak

And yet, despite all of that, spreadsheets remain the default tool for almost any desk based work. Forecasting. Planning. Analysis. Operations. Reporting. Even lightweight databases and internal systems often start life as a spreadsheet and never fully leave it.

We're addicted to them!

This is not an accident, and it is not simply inertia in the casual sense of “people don’t bother to change.” There’s a substantial body of research in information systems, human-computer interaction, and organisational behaviour that explains why this happens. The short version is that spreadsheets survive not because they are good, but because they are structurally aligned with how work actually gets done under constraints of time and coordination.

To understand why we still use spreadsheets, you need to stop thinking of them as a piece of software and more like a bridge between ‘figuring something out’ and ‘getting something done’.

In the absence of a specific piece of software for any given task, particularly a new task, then a spreadsheet is your notepad and working papers, you can just start working.

A workaround for the absence of systems

One of the most important findings in end-user computing research is that we don’t just use spreadsheets as calculators or data grids. They function as a form of end-user development platform.

Raymond Panko, one of the most cited researchers on spreadsheet errors, has argued that spreadsheets are effectively “de facto applications” built by non-programmers. In organisations, they frequently replace formal software systems. They become planning tools, decision engines, financial models, and reporting pipelines without ever being designed as such.

This matters because it reframes what you are actually doing when you open Excel or Google Sheets.

You're not “using software” in the conventional sense. You’re building a system in the only environment that doesn’t need permission, coordination, or formal design.

That distinction is crucial. In most organisations, to get a proper system in place you’ll need:

• a specification
• a budget
• agreement with other people
• engineering time
• implementation time
• training time
• ongoing maintenance

But by some clever trick, spreadsheets skip all of that. We’re all permitted to build systems and apps in spreadsheets without any of the formal bureaucratic barriers that businesses put in front of us.

This is why they dominate: we can avoid the bureaucracy, no need to enter the corporate maze of discussion, negotiation, authorization, and prevarication. Spreadsheets let us ‘just get on with our jobs’.

And once you understand that it becomes much harder to frame spreadsheets as a “bad choice.”

They can’t compete on quality. But they absolutely nail it on immediacy.

Zero friction system building

To understand why spreadsheets persist, you have to understand where they nail it.

It’s not correctness. It is not scalability. It is not robustness.

So what is it?

It is friction, or more specifically, lack of.

Every alternative system introduces unwanted friction. Even simple software tools need you to define some sort of structure or setuop before you can do anything meaningful with them. Web apps need setting up. Databases need schemas. Business information tools need data models. Internal applications need design decisions that must be agreed in advance.

Spreadsheets invert this requirement entirely.

Just fire it up and get going, you literally start with nothing.

And a blank sheet is not a system. It becomes a system only gradually, as you add structure through use. But even a well-built spreadsheet is rarely regarded asa complete ‘system’, it’s just a spreadsheet, it can be thrown away as easy as it can be set up.

This creates a fundamentally different relationship to work. Instead of “design then execute,” spreadsheets let us “execute while designing.”

That distinction is one of the reasons they are so persistent in our day-to-day work.Work inside organisations is rarely stable enough to justify upfront design. Directors change priorities, your manager changes their mind, people have different ideas, assumptions shift, priorities move.

Do spreadsheets function as a kind of placeholder for systems that have not yet justified formalisation?

Probably, yes.

But in most cases that formalisation never arrives.

Spreadsheets for problem solving

There is also a cognitive explanation for spreadsheet dominance that is often under estimated.

Human-computer interaction research has consistently shown that spreadsheets support exploratory thinking better than a structured system. We don’t use spreadsheets as static databases. We use with them iteratively, changing values, watching outcomes, adjusting assumptions, we refine their structure in real time.

This is more closely aligned with how we humans actually solve problems – trial and error.

When you’re faced with uncertainty, you don’t start solving problems with a fully formed model. You start with a partial understanding. You test hypotheses. You adjust incrementally. You refine based on feedback. An ever-decreasing circle toward a solution.

It pains me to say it, but spreadsheets support you in this process pretty well, because they provide immediate feedback, without commitment. You change a value and the entire model updates instantly. Want to graph the results – click click click goes your mouse. You have this tightness between action and consequence – and if you don’t like it you can undo, redo, overtype, edit, format, or maybe attempt to build a macro.

This is powerful stuff from a cognitive perspective because it reduces the time between thinking, action, and results. Instead of writing a specification, then implementing it - you’re thinking and working, in a live environment.

This is why spreadsheets feel intuitive even when we know the underlying structure is fragile. You’re not interacting with a carefully designed system; you’re iterating your way into understanding.

In that sense, spreadsheets are not just tools for getting a job done. They are tools for reasoning, for understanding the job.

The paradox of spreadsheet errors

One of the most studied aspects of spreadsheet use is error rates. Research by Panko and others has repeatedly found that spreadsheet errors are “both common and non-trivial”, even in professional environments. In controlled studies and field observations, error rates can be surprisingly high in non-trivial spreadsheets.

However, the key insight is not that errors exist. It is that they rarely produce immediate system failure.

And despite that spreadsheet errors can lead to poor decisions and cost millions of dollars most spreadsheet errors are:

• local rather than systemic
• subtle rather than catastrophic
• discovered after use rather than during creation

This creates a unique dynamic where errors don’t invalidate the entire system. Instead, they produce incremental corrections.

We respond by fixing rather than replacing:

• building in error checks
• improving the format
• introducing manual review steps
• layering additional logic on top of existing structure

What we rarely do is abandon them completely.

This is important because system replacement usually starts with a triggering event: a failure severe enough to justify disruption. Spreadsheets rarely produce that kind of failure. Instead, they produce manageable imperfections.

So your spreadsheet ‘system’ persists.

Overconfidence and the illusion of correctness

There’s also a psychological component that reinforces spreadsheet persistence.

Studies on decision-making with spreadsheets suggest that users tend to over-trust outputs, even when they are aware of potential errors. This is not unique to spreadsheets; it is consistent with broader cognitive biases such as overconfidence and reliance on familiar tools.

In practice this means that unless something clearly breaks, spreadsheet out puts are treated as sufficiently reliable for decision-making.

This creates a subtle but important effect: spreadsheets don’t need to be perfect to be trusted. They just need to be good enough.

Asa result, we continue to operate in a state where known imperfections are tolerated indefinitely because they are not disruptive enough for us to walk into the bureaucratic hell of company structure and negotiation to find abetter system (more on this below).

It reminds me of using AI. We have no idea if ChatGP's responses are accurate, in fact we know AI often lies (lovingly referred to as ‘hallucinating’), but responses are presented with confidence, as fact, and we accept them with open arms because we are looking for that immediacy.

Why we tolerate error-prone spreadsheets

At this point, a natural question emerges: if spreadsheets are so error-prone, why don’t we replace them with more reliable systems?

The answer lies in organisational disincentives rather than technical capability, in other words bureaucracy!

Research in management information systems consistently shows that bringing in new systems and software depends less on the superiority of the new system, and more on coordination cost. In other words, the more people involved, the more we need to agree on how a system should work, the harder it becomes to adopt.

Spreadsheets avoid coordination cost entirely. Just get on with it. There’s no need to agree a design. No need for governance structures. No need to standardise design and use across teams. Each user sets their own logic!

This decentralisation is not a bug. It’s the very reason spreadsheets run wild across companies.

Replacing them with proper systems introduces significant work, it involves other people, and hardest for many – decisions! Someone needs to define how data should be structured. Someone needs to enforce consistency. Someone needs to maintain the system. These are not tiny tasks; these are work and commitments.

And commitments are expensive!

So we’re lost in a predictable pattern: spreadsheets persist not because they are better (they are cheap and cheerful) but because the cost of replacing them is too hard, we stick it to the bottom of the pile, choose the easy way out.

Why “better tools”consistently lose

It is tempting to assume that spreadsheets persist because there are no better tools out there. But this explanation doesn’t hold up under scrutiny. In most domains, far better alternatives already exist.

The issue is not availability. It is adoption dynamics.

Better tools tend to require:

• upfront structuring of data
• agreement on workflows
• formal ownership
• training and onboarding
• migration of existing systems

Each of these steps introduces friction.

Spreadsheets avoid all of this by allowing incremental, unstructured growth.

This creates a fundamental asymmetry:

Better tools reduce long-term risk but there’s a short-term cost.

Spreadsheets reduce short-term cost but increase long-term risk.

But we all know that most decisions are made under short-term constraints. So spreadsheets consistently win.

Balancing structure and chaos

If you step back from the tool itself, spreadsheets are best understood not as software, but as a compromise between structure and chaos.

They are:

• structured enough to support computation
• unstructured enough to avoid coordination
• flexible enough to absorb uncertainty
• simple enough to be universally accessible

The combination is rare!

Most systems optimise for one dimension at the expense of others. Spreadsheets sit in the middle space where organisational reality is messy and time-constrained.

The real conclusion

In trying to understand why spreadsheets dominate your work life, it’s very easy to say they are familiar and easy, but the deeper explanation of that ‘familiar and easy’ is that spreadsheets sit at the intersection of three forces:

First, they eliminate friction at the point of creation, they avoid bureaucracy, they let us get on with our work.

Second, they match how we humans naturally think under uncertainty, they support our iterative way of exploration rather than upfront design.

Third, we can work alone, they reduce the coordination requirement where aligning with others in the company is expensive and time consuming.

Together, these forces make spreadsheets less of a choice and more of a default.

We certainly don’t use spreadsheets because they are the best option for the job.

We use them because the alternatives need a level of structure, discipline, and coordination that most businesses don’t, won’t, or can’t provide.

And until that changes, the humble spreadsheet, with your KPI data, your HR records, your payroll data, your budget and forecast, your birthday wish list, your holiday plan, your house renovation…

is here to stay.

‍