You have a spreadsheet open. Column A says "Month 1." You've been staring at that cell for twenty minutes because you have to type a number into it, and you genuinely don't know if it should be $0 or $40,000.
That paralysis is normal, and it kills more startups than bad products do. A sales forecast for a startup isn't a prophecy. It's a structured guess that tells you how much cash you can burn before the lights go out.
Here's what I've learned building forecasts for two companies that had zero customer history to lean on: the spreadsheet matters far less than the assumptions inside it. Get those wrong and you've built a beautiful machine for lying to yourself.
Key Takeaways
- A pre-revenue forecast is an argument, not a prediction. Build it from the bottom up, never from a market-size percentage.
- Pipeline-weighted forecasting beats every other method for startups once you have even 10 prospects in the funnel.
- Always build three scenarios — best, base, worst. Investors will ask for the worst one anyway.
- Update weekly, not quarterly. A forecast you don't touch for 90 days is already fiction.
- The formula that matters most: deals closed × average contract value, adjusted for churn and sales cycle length.
Why most startup sales forecasts are wrong on purpose
Founders build forecasts to impress investors, not to run the business. That's the core defect.
I watched a founder I'll call Marco project $80,000 in month three. He had one signed letter of intent and a warm intro to a procurement manager. The projection wasn't a forecast. It was a hope with a decimal point. When month three delivered $4,200, his runway math collapsed and he had to cut two hires he'd already promised.
The fix isn't pessimism. It's granularity. Every number should trace back to a named deal, a real conversation, or an observable conversion rate you've actually measured.
Is a sales forecast the same as a financial forecast?
No, and conflating them causes real damage. Your sales forecast covers revenue from customers — pipeline, contracts, renewals. Your financial forecast covers the whole business: costs, hiring, runway, burn multiple.
The sales forecast is an input. The financial model is the output. When people ask you for "the model," they usually mean the financial one, but if the revenue line is made up, everything downstream is decoration.
How do I make a sales forecast? Pick your method by stage
Most guides hand you one method and pretend it works everywhere. It doesn't. What works at 3 customers breaks at 300, and what works at 300 is impossible at zero.
Here's how I'd split it.
| Stage | Best method | Why it fits | What breaks it |
|---|---|---|---|
| Pre-revenue / idea | Top-down sanity check + bottom-up unit math | You have no data, only assumptions you can stress-test | Believing your own market share % |
| First 10 customers | Named-deal bottom-up | Every deal is hand-carried and knowable | Losing track of individual conversations |
| Repeatable sales motion | Pipeline-weighted | Conversion rates per stage become measurable | Stale deal stages |
| Scaling with a team | Cohort + historical conversion | You can extrapolate from rep-level performance | Sandbagging reps who want easy quotas |
One rule above all: if you're pre-revenue, never start from a percentage of a big market. That move feels impressive and tells you nothing.
What are the 7 steps of forecasting?
Frameworks vary, but the sequence that's survived every forecast I've rebuilt goes like this:
- Define the period and the unit — months, quarters, or deal count.
- List every active or probable deal with a value and a close date.
- Assign a probability to each stage in your funnel.
- Multiply value by probability to get a weighted total.
- Apply your historical conversion rate to shrink the optimistic number.
- Model churn and non-renewals against that total.
- Produce best, base, and worst cases, then feed the base case into your cash model.
Step five is where most founders cheat. They skip the shrinkage because they don't want to see the smaller number. But a forecast that doesn't survive contact with your own conversion data isn't a forecast — it's a wish.
Can you provide an example of a sales forecast?
Yes, and let's keep it small enough to check by hand.
Assume a B2B SaaS startup selling a $600 annual subscription to small agencies. In the current quarter it has:
- 4 deals at the demo stage, $600 each, historical close rate 30%
- 6 deals at proposal, same value, close rate 55%
- 2 verbal commitments, close rate 85%
Weighted revenue: (4 × 600 × 0.30) + (6 × 600 × 0.55) + (2 × 600 × 0.85) = 720 + 1,980 + 1,020 = $3,720.
Now here's the step everyone forgets. If your tested conversion rate is 70% of what you project as a pipeline, apply that haircut. Realistic base case: roughly $2,600.
The sales forecast formula in its plainest form: number of deals × average deal value × stage probability, then multiplied by your historical accuracy ratio. Nothing fancier is needed at this size.
Worst case? Assume half your verbal commitments fall through. Best case? Assume one deal refers a peer. Same model, three multipliers.
What is the best method to forecast sales?
For a startup, my honest answer is pipeline-weighted with a historical haircut, and I'd defend that position stubbornly.
Reason: it's the only method that forces you to look at reality deal-by-deal while still producing a number you can plan around. Top-down forecasting is fast but disconnected from your actual funnel. Pure gut feel works for about three months and then quietly drifts into fantasy.
How do you handle uncertainty without freezing up?
You quantify it. Build three columns in your sheet: worst, base, best. Give weight to the base case for planning, and keep the worst case in front of you when you decide how many people to hire.
I once ran a forecast where the worst case meant four months of runway left. That number lived on a sticky note on my monitor for a quarter. It changed how I spent every dollar. It also turned out to be the number that actually happened — which is a useful reminder about which scenario to trust.
The mistakes that cost me the most
Two years in, I insisted a big logo was "basically closed." I put it in the base case, hired a support person against it, and watched the deal sit in legal review for five months before dying. The lesson wasn't "don't be optimistic." It was never forecast a deal as closed until the signature exists.
Second failure: I forecast monthly but reviewed quarterly. By the time I noticed pipeline was thinning, I'd already committed to spend. Weekly review takes fifteen minutes and has saved me more than any dashboard tool.
A quick template structure that works
Rows: deals or deal cohorts. Columns: deal name, stage, value, probability, expected close date, weighted value. Below that: a scenario switch that multiplies the total by 0.6, 1.0, or 1.3. Freeze the top row, color-code probabilities, and update the close dates every Friday.
What actually matters when the spreadsheet is done
The number at the bottom is not the point. The point is the argument you had with yourself while building it — every probability you argued down, every close date you pushed out, every deal you finally admitted was dead.
That argument is what tells you how many months you have, and how many people you can afford. Build the forecast to run your business, and the investor version becomes easy. Build it for the investor, and reality will send you the invoice anyway — usually on a Friday afternoon, usually without warning.
