Twenty years of planning platforms, and the market's best-kept open secret is this: a survey of 164 S&OP professionals across 54 countries found 81% of companies still run the process in Excel or Google Sheets (AbcSupplyChain, November 2025). A modest sample, but an unambiguous signal, and many of the respondents own top-tier planning systems at the same time. They paid for the platform. They plan in the spreadsheet.
The standard explanation is resistance to change, followed by a training budget. The standard explanation has now failed for two decades, which suggests it was never the explanation.
People do not retreat to Excel out of nostalgia. They retreat to it because it is the last place their judgement is allowed to operate.
What the spreadsheet actually offers
Watch a planner or an analyst work and the pattern is consistent. The system of record holds the official numbers. The spreadsheet holds the thinking: the promotion that was cancelled yesterday, the customer who is quietly switching supplier, the production line that goes down for maintenance next month, the regulation that changes the labelling in March.
None of that is in the master data. All of it belongs in the plan. The enterprise system offers no way in, so the knowledge goes where it can go: a tab, a column of manual overrides, a side model held together by one person's discipline.
That is not indiscipline and it is not a skills gap. It is a verdict on the tools, not on the team.
The shadow copy is a requirements document
Every large organisation runs what amounts to a thousand and one planning systems: the official one, and a thousand spreadsheets correcting it. Most software vendors read this as a compliance problem and respond with lockdown, closing the exports, mandating the platform, retraining the users.
Lockdown fails, and it deserves to. The shadow spreadsheet estate is the most honest requirements document the organisation owns. Each workaround marks a place where the official tool could not hold something a person knew to be true. Confiscating the workaround does not transfer the knowledge into the system. It deletes the knowledge.
The best tool is the one your team actually uses. A platform that is bypassed is not a platform. It is a licence fee.
Upgrade the thinking, not the tool
The way out is not to defeat the spreadsheet but to give the judgement that lives there a proper home: one that keeps the freedom and adds what freedom alone cannot provide.
Three upgrades matter:
- Memory. An override typed over a forecast should be recorded as a decision: who adjusted, by how much, on what reasoning. Research covering roughly 147,000 forecasts found that judgemental adjustments improved accuracy for only just over half the items examined (Fildes, Goodwin and De Baets, International Journal of Forecasting, 2024). Half the forecast is a person, and nobody scores that half.
- Scoring. Once overrides are remembered, they can be scored. The planner whose adjustments beat the model keeps making them with evidence behind her. The adjustment class that consistently destroys value gets retired. Nobody's freedom is removed. Its results simply become visible.
- An audit trail. When the number feeds a board pack or a regulator, the chain from data to decision has to survive scrutiny. A spreadsheet cannot provide that. A decision layer can, without taking the spreadsheet away.
Exploration stays free. It simply stops being invisible.
The endpoint is not a world without Excel. It is a world where the spreadsheet is a window onto a system that remembers, scores and audits what the person in front of it decides. Nobody is asked to leave the place they think.
This is how Prophesee treats the people it works with. Judgement is an input to be captured and scored, not a behaviour to be trained away. See it on your own data. Start here.