Ask an early-stage sales team how their forecast is built and you will often hear some version of: "Each rep tells us what they think will close." That is not a forecast. It is a collection of hopes, weighted by each seller's optimism.
A good forecast is a judgement built on evidence. It should tell you what will close, what might, and why, in a way anyone on the leadership team can check. Getting there does not need expensive software. It needs discipline about what counts as proof. The problem is widespread: in Gartner's State of Sales Operations survey, only 45 percent of sales leaders and sellers said they had high confidence in their organisation's forecast accuracy.
Why forecasts miss
- Stages are based on seller activity. A deal moves to "proposal" because a proposal was sent, not because the buyer agreed to anything.
- Probability is assigned by stage. Every deal in stage three gets 40 percent, whatever the evidence says.
- Close dates come from the quarter. The date is the end of the period, not a date the buyer has agreed to.
- Nobody challenges the call. Pipeline reviews become status updates, so weak deals are never questioned until they slip.
Define stages by buyer evidence
Rewrite every stage so it can only be passed when the buyer has done something verifiable, using exit criteria written as buyer actions. For example:
- Qualified: the buyer has confirmed a problem worth solving and who else is involved.
- Validated: the economic buyer has agreed the business case and a budget range.
- Committed: the buyer has agreed a mutual success plan with named owners and dates.
- Contracting: procurement and legal are engaged and working from your paper.
Now a stage means something. When a deal is in "committed", everyone knows what has actually happened.
Use three categories, not percentages
For each deal expected this period, ask the seller to call it one of three things:
- Commit: you would bet your bonus on it. Every step to signature is known and agreed.
- Best case: it can close this period if specific, named things happen.
- Pipeline: real, but not this period.
Track how accurate each seller's commit calls are over time, and use the categories to keep the weekly forecast call short. Within two or three quarters you will know whose forecast to trust, and who needs coaching.
Run deal reviews, not pipeline reviews
A pipeline review walks through every deal and asks "any update?" A deal review takes the few deals that matter most and tests them properly. Five questions do most of the work:
- What problem is the buyer solving, and what happens if they do nothing?
- Have we met the person who signs, and what did they say?
- What are the remaining steps to signature, and who owns each one?
- What could stop this deal, and what are we doing about it?
- Where did the close date come from?
If the seller cannot answer one, the deal is not in commit. That is not a criticism of the seller. It is useful information, found early enough to act on.
Watch the trends, not just the number
Beyond the deals themselves, a few simple measures show whether the forecast is healthy: how many deals slip each period, how long deals spend in each stage, how much pipeline you have against target, and how accurate last quarter's commit was. A sudden rise in slipped deals or time in stage is an early warning, usually weeks before the number misses. Slippage is more common than most teams assume: Ebsta’s 2024 benchmark data found that between 34 and 46 percent of deals slipped past their close date each quarter in the first half of that year.
Make honesty safe
None of this works if sellers are punished for bad news. If moving a deal out of commit leads to a difficult conversation, sellers will keep it there until the last day. Reward accurate calls, including the ones that come down. A team that tells the truth early gives leadership time to fix things, and that is the whole point of a forecast.