When should a founder stop running operations out of spreadsheets?

When the spreadsheet only works because a specific person is watching it. That is the signal. Not the row count, not the file size, and not how sophisticated the formulas have gotten.

A spreadsheet that anyone on your team could pick up and run correctly on Monday is a tool. A spreadsheet that produces the right answer because one experienced person knows which cells to check, which exceptions to catch, and which rules were never written down is not a tool. It is a person, doing a job, inside a file.

What is actually wrong with the spreadsheet?

Usually nothing. Spreadsheets are excellent, and most of the ones I get asked to replace were the correct decision when they were built. Somebody needed to move fast, they built the thing in an afternoon, and it worked.

The failure is never the spreadsheet itself. It is that the business kept growing and the spreadsheet quietly took on responsibilities it was never designed to hold: scheduling, rules, exception handling, tracking, and institutional memory. At that point you do not have a spreadsheet problem. You have a process with no system behind it, and a file where the system should be.

I ran into this from the other side for seven years. The short version: Co-founded and scaled a multi-channel eCommerce business past $10M in annual revenue. Nearly every operational failure we had traced back to a manual step that worked fine at one volume and stopped working somewhere around three times that volume. It never announced itself. It just started producing errors we had not seen before.

What are the signals that it is time?

Four, in rough order of how seriously I take them.

One person is the process. If that person is out for two weeks, does the work stop, or does it just get done more slowly? If it stops, you are one resignation away from an operational crisis, and you are paying for that risk whether you have priced it or not.

The founder or CEO is still doing it. This is the most common version I get called about. It is also the most expensive, because the hourly cost of a CEO doing manual data work is enormous and completely invisible on the P&L.

Growth requires headcount that should not be necessary. If the plan for doubling volume is doubling the number of people touching the process, the process is the constraint, not the staffing.

The exceptions come to a human every time. Routine cases should move on their own. If everything routes to a person because the system cannot tell a routine case from an unusual one, there is no system.

If two of those are true, it is time. If three are true, it has been time for a while.

Does this mean you need custom software?

Usually not, and this is where founders talk themselves out of fixing it.

The mental leap from "our spreadsheet is failing" to "we need to build a platform" is a large one, and it is why a lot of these problems sit unsolved for years. The real answer is almost always smaller. Most of the work I do here lives inside tools the business already owns and already pays for. Microsoft 365 and the Power Platform cover an enormous amount of this ground. So does connecting two systems that were never talking, so people stop retyping the same information between them.

Buying something off the shelf is frequently the right answer too. Custom software earns its cost in specific situations, and a founder-run operational process is often not one of them.

What does the replacement actually look like?

I am working with a founder-led specialty healthcare provider right now where this was the whole engagement. The CEO personally ran patient logistics every week out of spreadsheets. It produced correct results, because they were watching it closely, which meant the ceiling on volume was their calendar.

The constraint was not the spreadsheet. The workflow had no operating system behind it. Scheduling, fulfillment, tracking, group rules, and exceptions all depended on spreadsheets and individual attention.

So we built the thing that was missing: an automated workflow covering intake, group rules, fulfillment, tracking, and exception handling. The rules that used to live in the CEO's head are encoded in the system. Routine cases move without anyone touching them. The unusual ones surface as exceptions instead of arriving as a weekly pile of work to sort through.

The weekly process no longer depends on the CEO. Worth noting: they had already tried to build the replacement themselves, and it had not gotten to something usable. That is common, and it is not a competence problem. It is that the person who understands the process best is also the person with the least uninterrupted time to rebuild it.

Where do founders get this wrong?

Two places.

They wait for a crisis. The spreadsheet does not fail loudly. It degrades, and the degradation gets absorbed by the person running it, until that person leaves or burns out or the business tries to grow and cannot.

And they scope it too big. The instinct is to fix the whole operation at once, which turns a three-week project into a nine-month one that never starts. The better move is to take the single worst process, replace it properly, measure what it gave back, and use that result to decide what comes next.

If a process in your business only runs because someone is personally holding it together, that is worth a conversation.