The Numbers That Kill a Seed Round: Reading a Forecast Like a VC
Founders model hope. Investors model decay. Five numbers in your forecast quietly decide whether the rest of it gets read.
There are two ways to build a forecast. Founders build them forward: here is where we are, here is the growth rate we believe in, multiply for thirty-six months. Investors read them backward: here is the ending number you are claiming, here is what would have to be true every single month to get there, and here is the subset of those things you have actually shown me.
The gap between those two methods is where seed rounds quietly die. Not on the big number at the end, which everyone in the room knows is fiction, but on the handful of assumptions that produce it. Below are the five numbers an investor checks first. Get them honest and the rest of the model gets the benefit of the doubt. Get them wrong and nothing after slide one is believed.
Number one: the monthly growth rate you are quietly assuming
Every forecast contains an implied month-over-month growth rate, even when the founder never wrote it down. It is the engine under the curve. And because growth compounds, the difference between the rate you plan and the rate you hit is not additive, it is multiplicative. A miss that feels like rounding error in any single month becomes the whole story by month twenty-four. Drag the slider and watch a three-point monthly miss eat a third of your ending number.
A 3-point monthly miss is the kind of thing a founder rounds away as noise. Compounded over two years it leaves you 46% short of the number in the deck. That gap, not the slope of the line, is what an investor is pricing.
This is why an investor cares more about the slope you can defend than the height you are claiming. A lower rate you can actually sustain produces a smaller, fundable number. A higher rate you are hoping for produces a larger, ignored one.
Number two: CAC payback, the honest version
The blended customer acquisition cost is the most flattering number in most models, because it hides paid spend inside a pile of organic signups that will not scale. The investor recomputes it on paid customers only, then asks how many months of gross margin it takes to earn that cost back. Past roughly eighteen months at seed, the business does not compound, it consumes. You are not building a flywheel, you are buying revenue on a credit card.
Number three: net revenue retention, or the missing cohort table
A forecast without a cohort table is a forecast that quietly assumes nobody ever leaves. That assumption is doing enormous work, and the founder usually does not know it is there. Investors do. They look for the cohort curve immediately, and when it is absent they assume the worst, because founders with good retention lead with it. It is also the first thing a sharp reader hunts for when they open your data room.
The difference between a plan that compounds and one that leaks is not visible in month one. It is visible in the shape of the second year, once churn has had time to work against you.
Number four: the conversion rate that does not survive a funnel
Top-line growth is the product of two things founders tend to project independently: how many people arrive, and what fraction convert. The forecast usually improves both at once and never says so out loud. Lay the funnel out stage by stage and the hidden demand becomes obvious. To hit a number, you frequently need several times your current top-of-funnel volume and a better conversion rate, starting next month, simultaneously. Either one alone might be reasonable. Both at once is a wish.
Number five: what breaks if one input misses by thirty percent
The last test is the one founders almost never run on themselves. Take each major assumption and move it down thirty percent, one at a time, and see what happens to the runway. A healthy plan bends. A fragile plan snaps, usually because two optimistic inputs were silently stacked on top of each other and the whole thing depended on both landing. This is also the test that tells you whether you are really default alive or only alive on paper.
How to forecast like the person across the table
- Build it bottoms-up. Start from a single customer, a single channel, and a real conversion rate, then add channels. Never start from the market and divide.
- Show the cohort table before anyone asks. Retention is the most load-bearing number in the model, so lead with it.
- Separate organic from paid everywhere. Blending them flatters the present and lies about the future.
- Pre-shock it yourself. Move every key input down thirty percent and write down what breaks, so you are the one who found the breakpoint, not your investor.
- Reconcile the model against the deck. The story and the numbers should be the same story. See why investors pass for what happens when they are not.
A forecast is not a prediction. Everyone knows you will be wrong. It is a demonstration of how you think about your own business under pressure, and that is the thing being priced. If you want eight quantitative analysts to stress-test the curve, the funnel, the cohorts, and the breakpoints before an investor does, that is exactly what the forecast review inside Roast My Startup is built to do.
Find the holes before an investor does
Roast My Startup is a firm of AI analysts that tears apart your deck, model, forecast, and data room, then tells you exactly what an investor would use to pass. Brutal first, constructive second.