# Fix the roster before the forecast · Insights · ZAC > When plans keep breaking, the problem is usually not the forecast but the data about who will actually be there to do the work. Fix how availability is reported first, and the plan starts to hold. Source: https://zanatelloacosta.com/insights/fix-the-roster-before-the-forecast Insights · Operations 14 October 2026 · 3 min read # Fix the roster before the forecast When plans keep breaking, the problem is usually not the forecast but the data about who will actually be there to do the work. Fix how availability is reported first, and the plan starts to hold. ![Portrait of Bruno Zanatello Acosta](https://zanatelloacosta.com/_astro/portrait-1x1.CHsNpeNf.png) By [Bruno Zanatello Acosta](https://zanatelloacosta.com/about) Zanatello Acosta Consulting **Related service:** [Diagnose](https://zanatelloacosta.com/services#diagnose) **Related case:** [National capacity planning](https://zanatelloacosta.com/work/national-capacity-planning) When a plan keeps breaking, most teams try to improve the forecast. In my experience the better first move is to fix the roster: the simple, unglamorous record of who will be available to do the work, and when. A plan can only be as good as its inputs, and availability is the input that is usually worst. ## Why do good forecasts still produce bad plans? Because a forecast tells you how much work is coming, not who will do it. Most planning errors I have seen came from the second half of that equation. At Amazon Spain I planned capacity for a national delivery network of about 2,500 partner stores. The volume forecasts were reasonable. The plans still broke, week after week, because the network could not see reliably which partners would be open and able to deliver. Some reported late, some not at all, and some reported one thing and did another. Every downstream decision, from routes to vans to staffing, sat on top of that uncertainty. The same pattern shows up in much smaller operations: a field-service company that schedules jobs before technicians confirm their week, a warehouse that plans shifts around temporary staff who may not turn up, a clinic that books rooms before practitioners send their availability. The model is fine. The inputs are guesses. ## What does fixing the roster actually involve? It is a process change before it is a technology change. Three things mattered most: 1. **One way to report availability.** A single, simple channel with a fixed deadline, instead of messages, calls and spreadsheets in different places. 2. **Clear rules for what happens when someone does not report, or reports and does not show up.** We designed the compliance mechanism with the legal team so it was fair, proportionate and enforceable. Rules nobody can enforce are just suggestions. 3. **Visible results.** Weekly compliance reporting, by site, shared with the people who could act on it. Once people can see the numbers, most of the improvement happens on its own. Only then did we build the capacity and volume forecast on top of the new data, replacing a set of scattered spreadsheets with one tool. ## How do you roll it out without breaking the operation? In phases, with plain communication. Partners needed to understand what was changing, why, and what was expected of them, in short messages rather than long policy documents. We trained people before each phase, watched adoption every week, and fixed the friction points as they appeared. When the underlying technology changed halfway through, we adapted the process to the new tool instead of starting again. Because the rules and the reporting were already in place, the change of tool barely moved the results. Roster compliance rose from 85% to 97%. The forecast did not get more sophisticated. It got honest inputs. ## How can a smaller team apply this? You do not need a national network to have this problem. If your plans break more often than your forecasts are wrong, start here: - **Measure the gap.** For two or three weeks, compare who you planned with who actually turned up or delivered. If the gap is bigger than your forecast error, availability is your problem. - **Pick one channel and one deadline.** A shared form, a calendar or a simple app is enough. What matters is that there is only one, and that it closes at a fixed time. - **Write down the rule for late or missing reports.** Keep it short, make it fair, and apply it consistently. - **Publish the number every week.** Compliance by team or site, visible to the people who own it. - **Then improve the forecast.** With honest inputs, even a simple spreadsheet model starts to hold. ## What is the broader lesson? Most operational problems that look analytical are really process problems upstream. Before building a better model, check whether the data feeding it describes reality. Often the smallest intervention that matters is not a new tool but a clear rule about who reports what, and when. That is how I start every engagement: map the people, decisions and handoffs behind the visible problem before designing anything. If your plans keep breaking, a short diagnosis usually finds where. [All insights](https://zanatelloacosta.com/insights) [Next: Test the route before you drive it](https://zanatelloacosta.com/insights/test-the-route-before-you-drive-it) [Discuss a problem](https://zanatelloacosta.com/contact) --- Zanatello Acosta Consulting · Org.nr. 938 525 390 · Øyer, Innlandet, Norway · post@zanatelloacosta.com · https://zanatelloacosta.com