---
name: forecast-assumptions
description: "Build a forecast that documents key assumptions, stress-tests them, and surfaces what must be true — so the model survives board scrutiny instead of collapsing under the first question."
---
# /forecast-assumptions

A model without documented assumptions is a black box that produces confident numbers with no visible foundation. When the CFO asks "where does the 15% growth rate come from?" and the answer is "it's in the model," the meeting stops being about strategy and starts being about credibility. Undocumented forecasts also decay silently — the assumption that was reasonable in January becomes wrong by April, and nobody notices until the miss. This skill forces every material assumption into the open, classifies what's data-backed versus guessed, and tests the ones that break the business.

**The 5-7 Key Assumptions — Name Them All**
For each assumption:
- State it precisely: not "growth will continue" but "new logo ARR grows at 12% QoQ, driven by 8 new AEs ramping to $180K quota in Month 4"
- What type is it? Revenue driver, cost driver, unit economic, timing assumption, or market assumption?
- What's the time horizon where this assumption applies?
- Are there interdependencies — does Assumption 3 only hold if Assumption 1 also holds?

Common assumptions to force yourself to name:
- New customer acquisition rate and ramp time
- Churn rate and expansion rate by cohort
- Sales capacity: headcount, ramp, quota attainment rate
- CAC and payback period trend
- Gross margin at scale vs. today
- Market growth rate and your share capture

**Data vs. Guess — Classify Each One**
For each assumption, assign one of three classifications:
- Data-backed: you have 3+ quarters of your own data supporting this; name the source
- Benchmarked: you're using industry data or comparable company data; name the specific source and acknowledge it may not apply
- Assumption: you don't have data; this is a judgment call; name the person making the judgment and what would make them update it
- Treat any model where more than 3 of 7 assumptions are in the "Assumption" category as a narrative, not a forecast

**Sensitivity Analysis — Test the Ones That Move the Needle**
- For each assumption, answer: if this is 5 percentage points worse than base case, what happens to EBITDA (or runway, or ARR)?
- Which 2-3 assumptions have the largest impact on the output metric? These are your key sensitivities.
- For each sensitive assumption: what leading indicator would tell you it's tracking wrong within 30 days?
- What would you change operationally if you knew by Month 2 that a key assumption was off?

**Three Scenarios — Base, Bear, Bull**
- Base: most likely outcome given current data and trends
- Bear: the 2 key assumptions that are wrong simultaneously — what does the business look like? Is it survivable?
- Bull: the 2 assumptions that are conservative and both turn out better — what does that unlock?
- Each scenario must produce a specific output number (ARR, EBITDA margin, burn multiple, cash out date) — not just a narrative
- The range between bear and bull is your honest confidence interval — if it's 40% wide, say so

**The One Assumption That Breaks Everything**
- Name the single assumption that, if wrong, invalidates the entire forecast — not just weakens it
- Example: "If quota attainment stays at 62% instead of hitting 80% in Q3, the revenue model doesn't close and burn extends by 7 months"
- What's the probability this assumption holds?
- What early warning signal — before the damage is done — would tell you it's failing?
- What's the contingency plan?

**Rules**
1. Every material assumption must be named — "the model assumes" is not documentation
2. Data vs. guess classification is mandatory — board members who find hidden guesses lose confidence in everything
3. Sensitivity analysis must focus on the 2-3 high-impact assumptions, not every cell in the spreadsheet
4. Three scenarios must each have a specific output number — narrative scenarios without numbers are not scenarios
5. The one breaking assumption must be named explicitly — if you refuse to name it, you haven't stress-tested the model
6. Every assumption must have an owner who is accountable for monitoring it against actuals

This output is a documented assumptions register, a sensitivity table for the key variables, three scenario outcomes with specific numbers, and a clear statement of what must be true — the foundation for a board conversation that builds credibility instead of eroding it.