Forecasts you can defend
in front of a leadership team.

A forecast is not a prediction. It's a structured bet — a set of stated assumptions, combined with defined drivers, producing a range of possible outcomes. The value of a forecast is not its accuracy. The value is the discipline of making assumptions explicit, testing them, and exposing the logic behind the numbers.

If someone challenges your forecast, the right answer is never "the model said so." The right answer is: "here are my assumptions, here's the driver logic, here's what moves the outcome, and here's what would change my mind." A forecast you can defend is a forecast built on transparent assumptions.

What forecasts are for

In an operations context, forecasts answer one of three questions:

• Should we do this? — capital decisions, investments, hires, new lines of business
• Can we do this? — capacity, throughput, resource planning against known demand
• Will we be okay? — cash, runway, covenant compliance, risk posture

Pick the question before you build the model. A forecast built without a decision context is just a spreadsheet.

Almost every forecast you'll produce falls into one of three categories. Each has its own drivers, its own math, and its own failure modes.

The discipline of this module covers all three. The deliverable at the end is one forecast of your choice, but the method is the same.

Every forecast worth producing has four layers. Each layer feeds the next. If any layer is missing, the forecast can't be defended.

The assumption register

The single most important document in any forecast is the assumption register — a standing list of every input, with the value, the source, and a short rationale. Without it, the forecast is a black box. With it, every challenge to the numbers becomes a productive conversation about the assumptions.

Claude is exceptional at structuring a forecast: it will write your assumption register, describe your driver logic, produce the scenario narrative, draft the presentation slides, and generate the Excel formulas. There is one thing it must not do on its own.

That rule changes the workflow. You don't ask Claude "what's the base case revenue?" and paste the answer into a deck. You ask Claude to build the model structure, generate the formulas, explain the logic — then you open Excel, enter the assumptions, run the math, and use Claude to help you interpret and communicate the results.

• 1

  Define the question and the decision

  Write it in one sentence. "Should we approve the $42K Line 3 upgrade?" or "Can we meet committed demand in Q3 with current staffing?" This frames everything that follows.

• 2

  Draft the assumption register with Claude

  Give Claude your raw inputs — trailing data, known constants, gut estimates. Ask it to structure them into a register with values, sources, and rationales. Flag any assumption you haven't yet defended.

• 3

  Have Claude design the model structure

  Inputs sheet, drivers sheet, scenario sheet, summary sheet. Claude writes the layout and the formulas. You paste them into Excel and run the math there.

• 4

  Run scenarios in Excel — not in chat

  Change assumptions in the inputs sheet. Watch the summary move. Verify every major output by hand at least once — sanity check, not full re-calc.

• 5

  Return to Claude for the narrative

  Paste your final Excel outputs back. Ask Claude to draft the scenario summary, the assumption commentary, and the presentation slides. This is where Claude shines — articulating what the numbers mean.

The forecast-building prompt template

ROLE: You are a senior FP&A analyst supporting an operations consultant. You build defensible forecasts with explicit assumptions and scenario discipline.

CONTEXT: [The operation, the decision this forecast supports, the audience who will see it.]

QUESTION: [One sentence. Should we, can we, or will we be okay?]

RAW INPUTS: [Trailing data, known constants, estimates — paste it all, unorganized is fine.]

TASK:
1. Draft an assumption register — every input with value, source, and a one-sentence rationale. Flag any assumption that lacks a source.
2. Write the driver logic — the math that turns assumptions into outputs. Use plain formulas I can paste into Excel (no pseudocode).
3. Define three scenarios — base, upside, downside. For each, list which assumptions move and why. Each scenario must reflect a plausible world, not three versions of optimism.
4. Specify the summary outputs the decision-maker needs.

STANDARD: Do not compute final numbers in your response — I will run the math in Excel. Give me the structure, the formulas, and the logic. Flag every assumption that's not defensible yet. Your narrative should be something I can present to a leadership team without rewriting.

The most common forecast failure is the three-scenarios-are-actually-one-scenario problem. Base: we grow 10%. Upside: we grow 14%. Downside: we grow 6%. That's not three scenarios. That's one scenario with a 4-point range. It tells the decision-maker nothing about actual risk.

A real scenario varies specific assumptions and explains why. Each scenario has a name, a narrative, and a named set of assumption changes.

What disciplined scenarios look like

Illustrative only — numbers for demonstration. Your figures come from Excel.

Each scenario needs a narrative

• Downside narrative: "Economic softening tightens buyer budgets. Close rates drop 6 points, new pipeline falls 25%, churn lifts to 11%. Plausible if Q1 headwinds persist."
• Base narrative: "Continuation of trailing twelve months. No major macro change. The most likely world unless something shifts."
• Upside narrative: "Two committed enterprise deals close on schedule, existing accounts expand as planned, no unexpected losses. Requires execution, not just luck."

A forecast that lives in Excel is a spreadsheet. A forecast that gets a decision is a forecast and a brief. Module 05 taught you the brief. This section is about pairing the two.

When you present a forecast to a leadership team, you are not presenting the model. You are presenting:

• The question — one sentence
• The answer — the base case output with a range
• The key assumptions — three to five that drive the outcome
• The scenarios — one sentence each, with what triggers them
• The ask — the decision you need, and by when

The full model is the appendix. It's there if anyone wants to trace a number. The brief is what gets read.

The presentation prompt — after the model is built

ROLE: You are a senior FP&A analyst. You translate forecast models into executive presentations.

CONTEXT: [The decision, the audience, the venue — leadership meeting, board, client.]

INPUT: [Paste: assumption register · scenario outputs from Excel · scenario narratives.]

TASK: Produce a one-page executive brief in BLUF format (Module 05 structure), plus three presentation slides:
1. The base case output, with the key drivers
2. The three scenarios, side-by-side, with one-sentence narratives
3. The recommendation and the ask

STANDARD: Lead with the answer. Every number must be traceable to the assumption register. If an assumption is weak, say so — do not paper over it. Your draft should be something I can present without rewriting, just revising.

That's the full arc. Define the question → register the assumptions → build the model in Excel → run the scenarios → produce the brief and slides with Claude. Every step has its tool. None of them are skippable.