Every investor has a version of the same first move: open the model, ignore the revenue line, and go straight to the assumptions tab. If there isn't one, the meeting is already in trouble.
After reviewing models across six cohorts and dozens of Idea Challenge pitches, the same five assumptions get attacked almost every time. Not because investors are predictable, but because founders make the same mistakes.
1. Customer acquisition cost — and whether it ever stops climbing
Most models show CAC dropping over time as 'the brand builds'. That might happen. It also might not. An investor who's seen a hundred models has seen this exact line in ninety of them, and the question is always: what specifically causes CAC to fall? If the answer is vague, the assumption is vague, and the model is doing heavy lifting for you.
2. Revenue-per-customer growth without a mechanism
Expanding revenue per customer requires a product roadmap, upsell triggers, or pricing power — not just a spreadsheet cell that grows 15% a year. Model what actually drives the expansion: new features, usage tiers, price increases. If none of those exist yet, don't model them.
3. Headcount that doesn't match the revenue it's supposed to produce
A company planning to reach $2m in revenue with three salespeople needs to show $660k per rep. That's an extraordinary quota for an early-stage product with no brand. The fix is simple: map each hire to a revenue contribution and ask whether that contribution is realistic.
4. Working capital that doesn't exist
If your customers pay in 60 days and your suppliers want payment in 30, you need working capital. Most startup models don't have a working capital line at all. The cash balance in the model is therefore wrong — sometimes by months of runway.
5. The break-even month that never moves
The model says you break even in month 18. Everything is built toward that point. But in the downside scenario, break-even moves to month 26 — and the raise doesn't cover month 26. Investors run the downside for you if you don't, and the conversation gets awkward when the gap appears.
The model isn't a prediction. It's a tool for thinking clearly about what has to be true for the business to work.
What to do about it
Label every assumption. Stress-test the five above before anyone else does. Build a sensitivity table on the two or three variables that actually swing the outcome. And when you don't know something, say so — an honest 'we're assuming X until we have data' is infinitely better than a confident number with no basis.