The number you make up
Someone asks what next quarter looks like. You say something. You've said it with enough conviction that nobody probed. But you know exactly what happened in your head when you said it: you thought about three deals you feel good about, applied a private discount for optimism, and produced a number that was mostly a mood.
Then the quarter happens and the number is wrong — not always low, which would at least be a pattern. Sometimes it's high because a deal you'd written off closed. That randomness is the tell. A forecast that's consistently 30% optimistic is a usable instrument with a known bias. A forecast that's wrong in both directions isn't an instrument at all.
And you make real decisions on it. Whether to hire. Whether to take the deal you don't want. Whether you can afford the build that would fix the thing that's been broken for two years. Every one of those decisions is sitting on a number you invented, and some part of you knows it, which is why you keep deferring them.
Forecasting is arithmetic, and you have no operands
A forecast isn't a prediction in the psychic sense. It's arithmetic, and it takes exactly two inputs. First: how often does a deal at this stage historically become revenue? Second: how long does it take? Give me those and I can forecast your business without ever meeting one of your clients. Take either away and nobody can, including you, including with twenty years of instinct.
So look at what you actually have. Your stages — are they gates or are they labels? A gate is a fact that either happened or didn't: they've told us the budget, we've met the person who signs, there's a date. A label is a word somebody picked. Most CRM stages are labels, which means "Proposal Sent" contains a deal that's genuinely closing and a deal that ghosted you in April, and they're indistinguishable. A stage that contains both is not a stage. It's a folder.
Second input: time. Nobody wrote down when the deal entered the stage, or when the first call was, or when it closed. So you can't compute a cycle. And without a cycle you can't age anything, which means you can't tell a live deal from a corpse — which is why your pipeline number is always inflated, because corpses never leave. They just sit there, propping up a total you don't believe.
You're not bad at forecasting. There's nothing to compute. The number came out as a feeling because a feeling was the only instrument available.
What an unpredictable number costs
The first cost is that you can't hire. Every capacity decision requires a floor you believe in. Without one you stay understaffed on purpose, which means you personally absorb every surge, which means the surge weeks generate nothing — and now the pipeline is emptier, and the number is even less predictable. The forecast problem manufactures the capacity problem that makes the forecast worse.
The second cost is that it wrecks your pricing. A founder with no visible next conversation prices from fear. You discount, you widen scope, you take the bad-fit deal. Not because you're weak — because your position is genuinely weak, and weak positions are visible from across the table.
The third cost is that you can't improve anything. Where do deals actually die? You don't know. So every change you make to how you sell is unmeasured, which means you'll keep changing things based on the last painful loss rather than the pattern. Revenue is the least useful number you own, because it's a trailing indicator of decisions you made months ago. It tells you the weather, not the climate. The metrics that matter are upstream of it — the ones that move first and give you time to act.
The fourth cost is personal, and it's the one that shows up at 4am. You cannot plan a life on a business you cannot see. That's not a business problem. That's why you're tired.
Build the operands
You don't need a forecasting model. You need the two inputs, and both are buildable in a week.
Make stages into gates. For each one, write the fact that must be true to enter it — a fact, not a vibe. Something you could argue about with evidence. Then a deal in that stage means something, and the historical conversion rate of that stage becomes a real number instead of an average across a folder full of unrelated objects.
Make time exist. Record when a deal entered each stage. That's a field and a trigger, and it's the entire foundation of every useful thing downstream: cycle time, aging, and the ability to say a deal that's been in Proposal Sent for nine days when your average is three has a problem right now. That aging rule is what stops corpses from inflating the number.
Then require a next step with a date on every open deal, and make its absence visible. A deal with no next step isn't a deal. It's a memory. Purge on age, deliberately, because a forecast built on things you're being polite about is a fiction you'll act on.
And build it as you do the work, not instead of the work. The next deal is the one where you write the gate down as you pass through it.
Pipeline, Telemetry, and the honest offer
This lives in Pipeline — the stages, the gates, the next steps. But the reason you feel it is Telemetry: you are not short on data, you are short on signal. The business generates the information constantly and almost none of it reaches you until it's already become a problem worth interrupting your day.
There's a second benefit to a real scoreboard that only shows up over time: it's what makes delegation safe. You can't hand a function to someone if the only instrument for judging it is your gut sitting in the room. That's why so many founders take sales back and conclude nobody else can do it. It wasn't the person — performance was only ever legible to you.
The honest offer. If your pipeline is otherwise healthy and you just need gates and dates, that's a week of work and you don't need to buy anything to do it. What we see more often is that the unforecastable business has a deeper shape problem: every deal routes through the founder, so the stages describe your availability rather than the buyer's progress, and no amount of field configuration will fix that.
The OPERATE Report is a $1,997 diagnostic across all seven pillars, for the founder who can feel that they can't see their own business. You'll get the binding constraint in writing and the specific thing that has to exist. The goal isn't to watch everything — it's to know the right things to watch.
A forecast is arithmetic on conversion rates and cycle times. If your stages are labels and your dates were never recorded, you don't have a forecasting problem — you have no operands.