The Adjustment That Was a Negotiation
Some changes to a forecast are information and some are pressure. How to tell them apart, what to do when the request is pressure, and why the record is the defence.
Some requests to change a forecast carry information and some carry preference. The distinction is not whether the request comes from somebody senior; it is whether what is offered is a fact about the world or a view about what the number should say.
The adjustment process in “The Adjustment That Was a Negotiation” should preserve the unaltered model result, the human change and the reason for it. A team considering Monitask for attendance sheet template for attendance sheet template can add operational time and project context, while the forecast owner remains responsible for documenting the override and reviewing its effect.
Both arrive in the same meeting, in the same tone, and telling them apart is a practical skill rather than a moral one.
Teams documenting “The Adjustment That Was a Negotiation” may use Microsoft Power BI resources as an outside reference. Test the method against the real forecast horizon, known events and operational constraints.
The question that sorts it
Ask what changed. A request carrying information answers: a contract, a date, an announcement, a conversation with a customer. A request carrying preference answers with a reason the number is unwelcome.
- Information: "Procurement say the contract signs in June."
- Preference: "That number is going to cause problems."
- Ambiguous: "Can you look at the assumptions again?"
The question is neutral and it sorts almost every case.
What information sounds like
"Procurement told me the contract is going to be signed in June." "Marketing has moved the campaign to the following week." "The automation is going live earlier than planned."
These are facts with sources. They belong in the forecast and in the event calendar.
What preference sounds like
"That number is going to cause problems." "We can't take that to the board." "Let's not be alarmist." "Can you look at the assumptions again."
The last one is the most common and the most slippery, because looking at assumptions again is a legitimate thing to do.
Looking at the assumptions again
Do it, honestly, and report what happens. Sometimes a genuine problem surfaces. More often nothing changes, and reporting that nothing changed is the correct outcome.
What is not acceptable is looking again until the number moves, which is a search with a stopping rule.
Why it matters
Because a forecast adjusted toward what people want stops predicting and starts negotiating, and because the decisions taken on it are then sized to a preference.
The consequence usually arrives two quarters later as a staffing shortfall nobody can trace.
The separation that defuses it
A forecast row and a target row with different owners. The request to change the number almost always resolves to a request about the target, and once there is somewhere for it to go, it goes there.
Most of the pressure on forecasts exists because there is only one number in the room.
What to do when the request stands
Make the change, record it, and record what was offered as the reason. Refusing is rarely productive and the record does the work.
"Published figure adjusted from 141 to 128 at the request of the operations director; no specific information was cited" is a complete and neutral entry.
Why that entry is enough
Because the error record will, in due course, say whether the adjustment helped. The planner does not have to win the argument in the room; the record wins it over a year, without anybody having to be accused of anything.
This is the main practical reason to keep the three columns.
The adjustment made upstream
Harder: where the pressure is applied to an input rather than to the output. A driver's size is revised, an assumption is changed, a period is excluded.
The defence is the same — a dated record of the inputs as used — but it requires the inputs to be versioned as well as the output.
The organisation where this is routine
Where every forecast is negotiated, the series of published forecasts is a record of what was acceptable rather than of what was expected, and its error record cannot be interpreted.
Rebuilding starts from the date the two numbers were separated, and the first honest year will look worse than the ten before it.
What a planner can and cannot do
Cannot stop the adjustment. Can insist that the system number survives alongside, that the reason is recorded as given, and that the error record is published.
Those three things are within the gift of the planning function and they are sufficient.
The request that arrives as a question
The most common form, and the hardest to classify: "are you sure about that number?" It can be genuine curiosity or an opening.
Treating it as genuine every time is the right default. Answering with the basis — the method, the drivers, the band, the recent errors — settles it either way without anybody having to take a position.
The long view
Over two or three years the record changes behaviour without anybody confronting anybody. Requests that cannot be given a reason stop being made, because the reason column is empty in public.
That is a slow mechanism and it is the only one that reliably works.