The spreadsheet isn't the problem — the gut feeling it papers over is. What planning in Excel really costs, and where the line runs.
In most Swiss SMEs, the central planning file is an Excel spreadsheet. That is not an oversight but a sound choice: Excel is flexible, everyone understands it, it needs no sign-off from IT and no licensing debate. For the question "What did last month look like?" it is unbeatable. The problem begins elsewhere — where the spreadsheet stops documenting the past and starts asserting the future.
The spreadsheet isn't the problem
It is worth being precise here, because the widespread mockery of "Excel" leads astray. Excel reliably calculates whatever you tell it. The break occurs where a planning figure is created — and that figure almost never comes from the spreadsheet. It comes from the head of the person who maintains it. "Next year we're counting on five percent more" is not a forecast but an extrapolation with a gut feeling in front of it. The spreadsheet merely lends this hunch the aura of a calculation.
«A planning figure in Excel is only as good as the gut feeling that produced it — the spreadsheet just makes it look tidier.»
What the spreadsheet really costs
The costs of planning in Excel appear on no invoice, which is why they are easy to overlook. They show up in four places:
- Time no one counts. The monthly consolidating, updating and formatting of planning files regularly costs hours — dispersed and invisible, but real. It is exactly the kind of recurring manual work that, over a year, adds up to a considerable sum.
- A single head as a single point of failure. Often only one person truly understands the logic of a spreadsheet that has grown over time. If that person is out or leaves, the planning knowledge leaves the company — and with them, traceability disappears.
- Errors that stay silent. Shifted references, overwritten formulas, a wrongly copied row: in spreadsheets that have grown over time, such errors are the rule, not the exception — and they often only surface once the decision has already been made.
- Decisions on an outdated basis. Planning done once a year remains a snapshot for eleven months. What has changed since then — season, order book, liquidity — is not in it until someone opens the spreadsheet again.
Where the line runs
The line is not the size of the company but the nature of the question. As long as it is about bookkeeping and hindsight — what was, what is — the spreadsheet remains the right tool. The moment it is about looking ahead under uncertainty — how much we will sell in the third quarter, how liquidity will develop over the coming months, when we need to build capacity — extrapolation reaches its limit. This is precisely where bookkeeping parts ways with forecasting.
A data-driven forecasting model does something the spreadsheet fundamentally cannot: it learns from its own history, it recognises seasonality and trends, and it corrects itself as new figures arrive. It does not replace the entrepreneur's judgement — it gives them an honest starting point they can dispute on solid grounds.
Why this matters especially in Switzerland
Swiss mid-sized businesses are particularly exposed to the silent Excel trap, and for good reason. Many SMEs have grown over decades, their processes with them — and their spreadsheets with those. They are often family businesses with deep trust in proven tools and a healthy scepticism toward every new system that promises to do "everything better". That scepticism is justified. Too often software has been sold that created more work than it saved.
At the same time, it is precisely these businesses that operate under real planning pressure: seasonality, supply bottlenecks, fluctuating exchange rates, and the liquidity question that, in case of doubt, decides their ability to act. Whoever reacts here one planning round too late pays not with a metric but with expensive inventory or a bottleneck at the worst possible time. The spreadsheet reflects this pressure — but it does not help to get ahead of it.
Forecasting does not mean a black box
The legitimate objection to "AI forecasts" is the loss of control: a figure that falls out of a black box and that no one can question is more dangerous than an honest gut feeling. That is why we apply the opposite rule. A usable model shows which drivers move the forecast; it can be tested against the existing planning on your own historical data; and it makes its hit rate measurable. For context: in one of our forecasting examples, the mean absolute percentage error came in at around 6.1 percent — a figure you must know and be able to trust, rather than merely believe.
Just as important is the honest flip side: if a model does not beat the existing planning on the real data, then we say so — and you keep your spreadsheet. A model that brings no demonstrable advantage is not a solution but extra work.
The first step is a comparison
The most sensible entry point is therefore not a change of system but a comparison: the existing Excel planning against a model, computed on your own historical figures. The result is either a demonstrable reason to place your planning on a more robust footing — or the reassuring confirmation that the spreadsheet is good enough for your case. Both are a win, because both replace a hunch with a figure.
What matters is what such a change does not mean: the end of the spreadsheet. Excel remains the right tool for everything that is hindsight and ad-hoc calculation. What shifts is solely the basis for looking ahead — away from silent extrapolation, toward a model whose assumptions lie open and whose hits you can recount. In the end it is not about technology but about a plain question of honesty with oneself: on what basis do we actually make our most important decisions — and would that basis hold up to sober scrutiny?
