Sales forecasting for startups. Built for a small pipeline, not a big one.

The short answer. A cash-runway signal, not a quota scoreboard.

Sales forecasting for startups is a different job from sales forecasting at a company with fifty reps and five years of closed-deal history. At an early-stage company the forecast isn't primarily there to hold a VP of Sales accountable to a number set by the board. It's there to answer one question the founder needs answered every month: given what's actually in the pipeline right now, does the business have enough contracted or near-certain revenue to cover what it's spending before the next round closes. Paul Graham's 2015 essay for Y Combinator on being "default alive" versus "default dead" put a name on the test most early forecasts exist to serve: whether current growth and burn, projected forward with no further fundraising, gets the company to profitability, or leaves it dependent on a raise that may not arrive on schedule. A sales forecast at this stage is one of the two or three inputs that answer that question honestly.

That reframing changes what "good" looks like. CB Insights' recurring analysis of startup post-mortems has found running out of cash among the most commonly cited reasons founders give for shutting down, and a forecast that overstates near-term revenue by even a modest margin can turn a genuine six-month runway into a false sense of nine. Mature B2B forecast-accuracy work is mostly about narrowing a process gap that only shows up once a team has enough historical deals to measure against. Forecasting for a startup is a different exercise: catching the gap between what the pipeline looks like and what the bank balance can survive, early enough to act on it.

How it works in practice. Bottom-up, not a model you can't trust yet.

With fewer than roughly twenty or thirty open deals a quarter, which describes almost every startup sales pipeline, a single lost or won deal swings the forecast by far more than it would for a team running hundreds through the funnel every month. That's the practical reason a startup should use a bottom-up method rather than anything more sophisticated: list every open deal, apply a realistic probability of closing this quarter to each one based on its actual stage, and sum the weighted total. A regression model or a multivariate forecasting tool needs a volume of historical closed-won and closed-lost data a young pipeline simply doesn't have yet. Running one on a dozen data points produces a number that looks precise and isn't.

The weighting itself should start conservative and get calibrated as real outcomes come in. Before a founder has closed enough deals to have genuine stage-by-stage win rates, a workable starting point is roughly 10 percent for a deal that's had one meeting, 30 percent once there's been a proper discovery call and the budget holder is confirmed, 60 percent once a proposal is out and being actively discussed, and 90 percent only once verbal agreement has been given and a contract is in legal review, never before. Once fifteen or twenty deals have closed either way, replace those starting numbers with the startup's own actual conversion rate at each stage. Most founders find their early guesses were too generous somewhere in the middle of the pipeline.

Roll the weighted pipeline total up alongside two other lines: revenue already signed and contracted, which counts at 100 percent regardless of when it's collected, and any known or likely churn among existing customers, which should be subtracted rather than ignored. The three lines together, signed revenue, weighted pipeline, expected churn, give a single monthly number worth comparing against burn. A spreadsheet, or the native reporting inside whatever CRM the founder is already using, is sufficient for this. Dedicated forecasting software earns its cost once the team is large enough that no one person can hold the whole pipeline in their head, not before; buying a tool ahead of that point usually just moves the guesswork into a more expensive interface.

Cadence matters more than tooling at this stage. A weekly look at the raw pipeline, what moved, what stalled, what's new, keeps the underlying data honest. A monthly reconciliation against actual burn and runway is the moment that matters for the business: pulling the weighted forecast, the actual bank balance, and the current monthly burn into one view, and asking whether the gap between the two has widened or narrowed since last month. This is the discipline a properly built sales dashboard is meant to support once the manual version becomes too slow to trust.

The moment to move past a one-person spreadsheet is usually the second sales hire, not any particular revenue milestone. Once two people are working the same pipeline, attribution gets murky, a deal either founder touched can quietly get counted, or lost, in both people's heads at once, and the written stage probabilities stop being a nice-to-have and become the only way anyone outside the room can trust the number. This is also the point where it's worth deciding, in writing, who owns updating the forecast each week, since a shared spreadsheet with no clear owner drifts out of date within a month.

Forecasting for a startup also has to answer a second question the mature process rarely asks, because mature teams have already answered it: how many people does this pipeline justify hiring next. A founder who forecasts revenue without also forecasting the pipeline volume needed to hit it risks hiring a second or third salesperson into a pipeline that was never big enough to support one extra quota in the first place. The same weighted-pipeline number used to check runway should feed the hiring plan too, since both questions come from the same underlying data.

What good looks like. A worked example.

An eight-person, seed-stage vertical SaaS company I worked with had a founder personally running sales, tracking deals in a spreadsheet with a single "likely to close" column he filled in from memory before board meetings. The number was consistently optimistic, not through dishonesty, but because every deal in an active conversation felt further along in his head than it actually was in the buyer's process. Investors noticed the gap before he did: two consecutive quarters where the forecasted number missed the actual closed number by more than a third.

The fix wasn't a new tool, it was the three-line structure above, built into the same spreadsheet he already used. Every open deal got tagged to one of four stages with a fixed starting probability, signed revenue was pulled into its own line, and known churn from two accounts that had gone quiet was subtracted rather than left out. The first month running the new format, the forecast came in within 12 percent of the actual closed number, and it stayed within that range for the following two quarters as the stage probabilities got calibrated against real outcomes. Nothing about the underlying sales process changed. The forecast simply stopped lying to him, and the board conversation shifted from why he'd missed again to a genuine discussion about runway and hiring.

Pitfalls to avoid. Where founder-led forecasts go wrong.

The most common failure is forecasting from gut feel rather than a fixed stage probability, which is exactly what caught the founder above out. A deal the founder likes, or has spent the most time on personally, tends to get an optimistic mental probability regardless of where it actually sits in the buyer's process. Fixed, written probabilities per stage remove most of that bias, even before there's enough data to calibrate them precisely.

The second is conflating a verbal yes with a signed contract. "They said yes on the call" is a real signal and belongs somewhere in the pipeline, but it belongs at 60 or 70 percent, not 100, until a contract is actually signed. Enough verbal agreements fall through procurement, legal review or a change in budget that treating them as closed revenue overstates the number every single quarter.

The third is running too thin a coverage ratio. A mature sales organisation with a large, stable pipeline can sometimes forecast reliably from pipeline value close to the target number itself. A startup pipeline, with far fewer deals and much higher variance per deal, needs a pipeline several times the size of the number it's meant to produce, since losing any single deal in a small pipeline removes a disproportionate share of the total.

The fourth is never tying the forecast explicitly back to runway. A forecast that lives in a CRM tab, disconnected from the bank balance and the burn rate, is an interesting sales metric and not the cash-management tool a startup actually needs it to be. The monthly reconciliation described above is what closes that gap.

The fifth is letting one large deal carry the whole quarter. It's common for an early pipeline to include one enterprise-sized deal that, on its own, would make the quarter's number. Treating that deal as in the bag anywhere earlier than a signed contract concentrates the entire forecast's risk in a single buyer's internal timeline, one the startup does not control.

My rule with founders building their first forecast: build it to be checked against the bank balance every month before you build it to be presented to a board every quarter. A forecast a founder trusts enough to make a hiring or spending decision against is worth more, this early, than one built to look sophisticated in a deck.

The sixth, and the one that catches teams a little further along, is leaving the forecast process behind once the first dedicated sales hire arrives. A founder who built good habits alone sometimes assumes a new rep will inherit them by osmosis. They rarely do without the stage probabilities and the weekly update habit being written down and handed over explicitly, which is a smaller task than it sounds and pays for itself the first time a rep's forecast and the founder's gut instinct disagree.