The annual budget assumes the coming year resembles the last one closely enough for a point estimate to mean something. When that assumption fails, most companies either produce a budget nobody believes or postpone the exercise entirely. There is a better third option.
Build three cases, and name them honestly
Not optimistic, realistic and pessimistic — those labels invite you to treat the middle one as the plan. Use descriptive names tied to conditions: restrictions lifted by spring, restrictions until summer, restrictions through the year.
Each case should have a revenue path and, more importantly, a cost structure that responds to it. The value is not in the forecast; it is in knowing what each scenario requires of you.
Decide the triggers in advance
For each case, identify the observable signal that tells you which world you are in, and the date by which you will know. Orders below a level by the end of February. Occupancy under a threshold by May.
Then write down what happens when a trigger fires: which spending pauses, which hiring stops, which facility gets drawn. Deciding this while calm is the entire point.
Protect the things you cannot rebuild
In a contraction, the temptation is to cut proportionally across everything. That is rarely right. Some capabilities — a key technical team, a certification, a customer relationship — take years to rebuild and should be cut last or not at all.
Identify them explicitly before the pressure arrives, because under pressure everything looks equally discretionary.
Re-forecast monthly, budget annually
Keep the annual budget as the frame for commitments and accountability. Run a rolling forecast against it monthly, and let the forecast, not the budget, drive operational decisions.
The budget answers what you intended. The forecast answers what is happening. Confusing the two is what makes a budget useless in a year like this one.