Why I Stopped Selling Forecasts
Consulting sells certainty. Clients need an order of action under uncertainty. After one workshop I replaced the entire deliverable — including the forecasting model I had built for six years.
The turning point came on the third day of a strategy workshop.
The first two days were good work: the industry split into five segments, share, growth and profit pool calculated, and a structural forecast chart for the next three years. On day three there was one agenda item — whether to enter a new business. Halfway through, the CEO said: your chart assumes our competitors are still these three. But if the platform upstream decides to do this itself, does the chart still mean anything?
It did. And it did not. It described, precisely, a world that was not going to happen.
1. The hidden trade inside a forecast
When I reviewed that delivery later, the problem was not the model. It was the transaction.
A forecast report completes an implicit trade: the client pays for a portion of certainty about the future, and the consultant takes on the professional risk of being wrong. In calm periods this is a good trade, because the forecast mostly holds and the report is actually used.
At a turning point it is bad for both sides. The client receives a rigorous-looking document that describes the old world; the consultant is asked to stand behind something unpredictable, and can only add complexity and appendices to look credible. Complexity becomes a disclaimer.
2. Replacing the deliverable
There is no "three-year forecast" section in what I deliver now. In its place is a scenario brief, in four parts.
- Structural assumptions: which rules we believe currently organize the industry. This is the first thing tested, and the first thing to break.
- Generative mechanism: who is driving the change, with what resource, and how costs and returns move with scale.
- Three paths: not base/upside/downside, but three different ways of organizing, each with trigger conditions.
- Early signals: three to five observable indicators per path, who watches them, how often, and what happens when they fire.
The last column is the most useful. It turns strategy from "a judgment" into "a monitoring and response mechanism". The client leaves with something that can be updated at the quarterly meeting, not a conclusion.
3. The cost
This shift has real costs, and I do not want to make it sound painless.
Longer sales cycles. Clients who want one answer leave. Those who stay usually have a live decision and are more willing to invest time.
Heavier preparation. Scenario generation needs material: real operating data, frontline judgment, records of past decisions. Without it, the output can only be generic.
Blurrier accountability. Without a point forecast, the clean "we were right, they were wrong" score disappears. The test becomes: did the early signals appear, and did the decision change because of them?
4. What I kept
Not everything was thrown away. Three things stayed.
Review. After each decision, go back: which assumption was it based on, which one was falsified, what would we do differently. It is the only reliable source of calibration data.
Quantification. Scenarios need numbers or they cannot be compared. The numbers just shift from "forecast" to "conditional" — given this path, what do they look like.
Being in the room. The first-hand signal of structural change often lives in how people talk, not in a report. Workshops remain my most important research site.
5. A more honest promise
What I promise now is one sentence: I will not give you a certainty I cannot carry, but I will give you an order of action that still works when you do not know.
That sentence is not an advantage in a proposal meeting. What it buys is this: a year after the project, the client is still using the signal list — instead of filing the report in a drawer.