Sometimes Automation Begins With a Very Boring Spreadsheet

Some of the best opportunities for automation aren't hiding in your technology strategy. They're sitting quietly in someone's spreadsheet.

Somewhere in your organization, there is probably a spreadsheet doing far more important work than anyone realizes.

Maybe Janet created it seven years ago. It started simply. She needed information the system didn't provide, so she exported some data and added a few columns.

Then Finance needed another field. Operations needed a different calculation. Someone added a tab for exceptions. A formula was created to reconcile two systems. Then someone needed the information every Tuesday, so Janet started sending it to them. Eventually another department began relying on it.

Today, the file has 14 tabs, three macros, several mysterious yellow cells nobody is allowed to touch, and a filename something like FINAL_v7_USE_THIS_ONE.xlsx.

Janet knows exactly how it works. Everyone else knows one thing:

Don't break Janet's spreadsheet.

It's easy to laugh. But that spreadsheet may be telling you something important about your organization.

The spreadsheet probably isn't the problem

Spreadsheets are extraordinarily useful. They're flexible, accessible, familiar, and when an employee needs to solve a problem quickly, a spreadsheet can be one of the fastest ways to do it.

So the lesson here isn't: Spreadsheets are bad. Replace them.

Quite the opposite. Sometimes a spreadsheet is evidence of ingenuity.

Someone saw a gap between what the organization needed and what its systems could provide - and built something to bridge it. That's worth understanding.

Because when a spreadsheet becomes essential to an ongoing business process, it may be showing you exactly where an opportunity exists.

The spreadsheet may be the workaround. The interesting question is what it's working around.

Follow the work

Suppose every Monday someone downloads a report, opens another system, downloads another report, copies information from one into the other, reformats several columns, looks up missing information, applies a formula, checks for exceptions, corrects records, adds information from an email, updates a spreadsheet, sends it to four people, and saves a copy in a shared folder.

Tuesday comes. Someone else takes that spreadsheet and uses it to update something else. Wednesday? Another department uses the same information to produce a report.

Nothing about this process necessarily looks dramatic. Nobody has declared an emergency. The work gets done. Every week.

That's exactly why it can survive for years.

Manual work can become invisible

Organizations notice broken processes. They don't always notice processes that work inefficiently.

If Janet fails to produce the spreadsheet, everyone notices. If Janet spends four hours producing it exactly as she has every week for six years? Nobody notices. That's simply her job.

But four hours a week is more than 200 hours a year. And perhaps Janet isn't the only one touching the process.

Maybe Finance spends another three hours reconciling it. Operations spends two hours correcting exceptions. A manager spends an hour reviewing it.

Now the organization isn't spending four hours. It's spending ten. More than 500 hours every year. On something everyone considers routine.

Repetitive work often hides in plain sight because success makes the effort behind it invisible.

Don't automate the spreadsheet yet

Once we identify repetitive manual work, the temptation is to immediately say: Automate it!

Not so fast.

Before automating a process, understand why it exists.

Why does Janet create the spreadsheet? Who uses it? What decisions depend on it? Where does the information originate? Why isn't that information available in the system? Which calculations are being performed? Which exceptions require judgment? Which steps exist because of a legitimate business requirement? Which exist because of a system limitation? Which exist because that's how someone solved the problem eight years ago?

And my personal favorite: Does anyone still need this?

Because there are few things more efficient than beautifully automating a process nobody actually needs.

The goal isn't to make the spreadsheet faster. The goal is to understand the business need underneath it.

Watch where people re-create information

One of the strongest signals of an automation opportunity is repeated information.

Someone already knows something. Then someone else has to figure it out again.

Information exists in one system but gets manually entered into another. A decision gets made upstream but has to be interpreted again downstream. A classification exists somewhere - but isn't carried forward.

These are more than efficiency issues. They can become data-quality issues.

Because every time a person has to re-enter, re-create, or reinterpret information, the result has an opportunity to change.

Automation isn't only about saving time

Imagine an organization that holds many different types of events. When each event is created, the people responsible for it know exactly what it is: a sporting event, a concert, a performance, a corporate event, or something else entirely.

But that classification isn't captured as structured information. Instead, much later in the process, someone in Accounting receives a description of the event and determines how the related expenses should be classified.

An experienced employee may become exceptionally good at it. They know the organization, the events, the terminology, and the accounting implications.

Then they leave. A new employee takes over. Same process. Same descriptions. Different institutional knowledge. And sometimes - different interpretation.

The process hasn't changed. But the data has.

If consistent data depends on consistent human interpretation, staff changes can become data changes.

Put the decision where the knowledge is

There's another way to design that process.

Instead of asking someone downstream to determine what the event was, capture the classification when the event is created. At that moment, the people closest to the event already know what it is.

Let them make the appropriate classification. Then preserve it.

The event type travels with the event - into operations, scheduling, timekeeping, payroll, and financial systems.

The downstream accountant no longer has to reconstruct the meaning of the event from its description.

The organization hasn't eliminated human judgment. It has moved judgment to the place where the best information exists.

Then technology does something people shouldn't have to keep doing: it remembers.

When information is known, capture it. When a business rule is established, preserve it. When data moves, carry its meaning with it.

Technology can become organizational memory

Organizations carry an enormous amount of knowledge in people's heads. Some of that absolutely belongs there. Experience matters. Context matters. Judgment matters. Relationships matter. Human beings understand nuance in ways systems cannot.

But established classifications, calculations, routing rules, relationships between data, and repeatable business logic often don't need to be rediscovered every time the work occurs.

Technology can preserve what the organization already knows.

That creates consistency. It reduces dependence on institutional memory. It makes transitions easier when people change roles. And it lets employees spend more time on the situations where their judgment actually adds value.

Good automation doesn't eliminate human judgment. It puts judgment where it adds the most value.

Look for the handoffs

The most interesting automation opportunities often don't exist entirely inside one person's job. They live between people, departments, and systems.

Operations creates something. Finance needs it. Payroll changes it. Accounting reconciles it. Leadership reports on it. Technology supports all of it.

Each team may be performing its piece perfectly. But every handoff creates a question: Does the next person receive what they need?

If not, somebody fills the gap. Often manually. They create the spreadsheet. Send the email. Reformat the report. Translate the code. Re-enter the information. Build the reconciliation.

That's why process improvement shouldn't only look at individual tasks. Follow the information across the organization.

You may discover that Janet's spreadsheet exists because two perfectly functional systems have never been taught to talk to each other.

Ask what technology should do - and what people should do

The goal of automation isn't to remove humans from work. It's to make better decisions about where human capability creates the most value.

Technology is very good at certain things: remembering, calculating, applying established rules, moving information, matching records, identifying exceptions, and repeating the same action consistently.

People are very good at other things: judgment, context, relationships, creativity, problem-solving, understanding unusual circumstances, and making decisions when the rules don't quite fit.

Which parts of this work require a person - and which parts simply require the organization to remember what it already knows?

And don't forget Janet

If Janet built the spreadsheet, talk to Janet before you replace it.

She may know why every column exists. She knows which exceptions matter. She knows why one formula looks completely illogical but absolutely cannot be changed. She knows that Finance needs Column H even though nobody else understands why. She knows which data can be trusted. She knows what she checks manually before she sends the file.

She may even have been saying for four years: “There has to be a better way to do this.”

Her spreadsheet isn't merely something to automate.

It's documentation of a problem someone has been solving for the organization.

Start with the boring stuff

Automation conversations can become exciting very quickly: AI, predictive analytics, agents, new platforms, sophisticated technology.

There are extraordinary possibilities ahead. But some of the highest-value opportunities in an organization may be considerably less glamorous.

The report someone manually reconciles. The information somebody enters twice. The email somebody sends 40 times. The classification someone has to interpret. The spreadsheet everyone depends on. The process held together by institutional knowledge.

Look there.

Not because every manual process should be automated. And certainly not because every spreadsheet should disappear.

Look there because repetitive work can reveal where people are compensating for gaps between systems, processes, information, and organizational needs.

Those gaps are worth understanding. And sometimes solving one of them gives hundreds of hours back to the organization while simultaneously improving the consistency of its data.

That's a pretty good return from a very boring spreadsheet.

A question worth asking

Take a walk through your organization. Don't start by asking: Where can we use AI? Don't even start with: What should we automate?

Ask something simpler:

What are our people doing over and over again that technology could reliably do for them?

Then follow the work. Follow the information. Follow the handoffs. Find the workarounds.

And when somebody says, “Oh, Janet has a spreadsheet for that,” you may have just found a very interesting place to start.