Module 13 — Reading the Entrails: Forecasting

"A forecast isn't a prediction. It's a promise you make to people who will fire you if you break it. Predict accordingly, you goddamn optimist." — overheard from a CRO at 11:58 p.m. on the last night of Q3, four bourbons in, staring at a number that would not fucking move no matter how hard he willed it.

It is the last Tuesday before quarter-end. Somewhere a VP of Sales is doing math on a cocktail napkin with a shaking hand because the CRM says one thing, his gut says another, the board call is in forty hours, and The Number is short by $340K. Nobody — nobody — will tell him whether the Henderson deal is real or a beautiful hallucination held together with a junior champion's good vibes and a verbal "we're excited." This is forecasting. It is not science. It is augury performed on a dataset that lies to your face, performed by humans who lie to themselves, supervised by executives who cannot afford to know the truth. We are going to make it lie less. We are going to make it lie predictably, which is the only species of honesty this industry can metabolize.

THE JOB

Forecasting is you, on the goddamn record, telling a room full of people who can fire you exactly how much money is going to materialize — and then living or dying by whether the universe agrees. That is the whole job. You are not predicting the future for sport. You are issuing a commitment that finance will spend against, the CFO will tell investors, and the board will use to decide whether you keep your chair, your team, and your parking spot in the underground garage where the executive slots are heated.

Two things are true at once and you must hold both in your head without losing your shit:

A forecast is a number (what will close this period) AND a forecast is a trust instrument (whether anyone should believe you next time). The number can be wrong by 2% and forgivable. The trust, once broken, costs you the room forever. A leader who hits 100% of a forecast called at 100% is a minor deity. A leader who closes 120% but called 95% is a liability — because now nobody knows what your words mean, and a CRO whose words carry no meaning is just an expensive person in a Patagonia vest consuming oxygen and Erewhon smoothies at the company's expense.

Forecast accuracy is not a vanity metric. It is the measurement of whether your mouth works correctly. Protect it like a fucking organ.

THE PLAYBOOK

1. Learn the forecast categories. They are a language. Speak it precisely or shut up.

Every deal in the period lives in exactly one bucket. Not "sorta Commit," not "probably Best Case I think," not the weird no-man's-land your RevOps platform calls "Likely" because some VC-backed AI startup decided to invent a fifth category nobody uses. The standard four:

• Commit — "I will bet my job on this closing this period." High confidence. Champion locked. Economic Buyer engaged. No fatal open items. Next step confirmed. This is the floor of the forecast, the bedrock, the number your CFO will subtract from the board's expectation when she wakes up at 3 a.m. and checks the slide deck. If your Commit misses consistently, you have committed fraud against yourself and everyone downstream. This is not hyperbole. This is what the post-mortem says.

• Best Case — "This closes if the wind blows right and nobody's assistant takes a three-day weekend." Plausible. Real path. Live risk. The ceiling of what you'd celebrate, not the floor of what you'd promise. Best Case is where the happy ears live — keep them contained here, don't let them metastasize upward.

• Pipeline / Upside — "In the period, qualified, in-system, but I'm not calling it." Real opportunities with a close date inside the window that haven't earned Commit or Best Case yet. The raw material. The feedstock. The shit that could become money if enough things go right.

• Omitted — "In the system, in the period, and I am formally telling you to ignore it." Long shots. Slipped deals you haven't recategorized because despair is a form of procrastination. Garbage close dates from Q2 that nobody cleaned. The prospect your rep insists is "still engaged" because they opened one email eight weeks ago.

The two numbers that matter for the actual forecast call: Commit (the promise) and Commit + Best Case (the realistic ceiling). Everything else is context, background noise, a story you're telling yourself while the real number either closes or it doesn't.

2. Pick a forecasting method. Know what each one lies about.

There are four ways to generate the number. All four are wrong. They are wrong in different directions, which is the only goddamn reason any of this is tractable. Pick the right combination of wrong and you get close enough to run a company.

The weighted-pipeline formula, because you will be asked to defend it:

Forecast = Σ (Deal Amount × Stage Win Probability)

Example: $200K at 30% + $150K at 60% + $400K at 80% = $60K + $90K + $320K = $470K weighted.

That $470K is not in your bank. It is a probability cloud shaped vaguely like money. Treat it accordingly — as a floor estimate, a sanity check, not a fucking plan.

The professional answer is don't pick one — triangulate, you lazy bastard. Run weighted/stage-based for the bottoms-up sanity number. Overlay rep judgment via the category call for the human truth. Let the AI model referee both and flag every deal where the rep's category and the model's score are having a fistfight. Where they disagree is where you inspect. The agreement is just comfort food. Comfort food doesn't close deals.

3. Name the two original sins: sandbagging and happy ears. One of them is in your pipeline right now.

• Sandbagging — the rep hides real, closeable pipeline so he can blow past a soft number, bank a beat, and emerge from the quarter smelling like a hero while you've been staring at a forecast that looked like shit for eight weeks. A forecast so sandbagged it qualified as federal flood infrastructure. It feels safe. It is poison: it makes your number meaningless, hides true capacity, screws the hiring plan, and ensures finance will never trust your top-of-funnel math again. The Brilliant Jerk does this. You know who he is. You've been tolerating it because he's 140% of his number and nobody wants to have the conversation. Have the fucking conversation.

• Happy ears — the rep believes every "this looks great, send the contract" is a signed deal and commits a pipeline held together by hope, a single junior champion who just updated his LinkedIn to "open to work," and one call where the prospect didn't immediately say no. Happy ears are worn by every rep who has ever missed by 40% and said "I really thought we had it." You thought wrong. You committed wrong. You are wrong in public, every quarter, wearing the same surprised face.

Both are forecast accuracy crimes. You fight them the same way: category discipline, exit criteria, and historical conversion rates that do not care about anyone's feelings. If a rep's Commit category closes at 65% historically, their Commit isn't a Commit — it's Best Case in a costume, and you discount it on sight, professionally, without drama, by the math alone.

4. Run the weekly forecast call. Here is the cadence. Do not improvise.

Same time every week. 30–45 minutes. No exceptions for offsites, for someone's birthday, for "it's a slow quarter." Reps come pre-submitted — the call is for inspection, not data entry, not group therapy, not a ritual where everyone pretends the number looks better than it does.

Step 1 — Submit before the call. Every rep locks categories in the CRM the night before. The number is frozen. If they haven't submitted, they don't talk on the call. That's the rule and you will enforce it or it will dissolve inside of three weeks.

Step 2 — Open with the delta. "Last week your Commit was $X. This week it's $Y. What the hell changed." Movement is the agenda. A deal that hasn't moved in three weeks is either closed or dead, and your rep's silence on which one is a data point in itself.

Step 3 — Inspect Commit, deal by fucking deal. For each: next step, next step date, who is the Economic Buyer and have they been engaged, what is the open risk that could kill it, what happens if the champion leaves. If they can't answer these cleanly, the deal is not a Commit. Move it down. Do this out loud in front of the team. The standard has to be public or it doesn't exist.

Step 4 — Pressure-test Best Case. What single thing converts these to Commit before close? That thing is this week's job. Write it down. Check it next week.

Step 5 — Roll up live. Rep → manager → VP → CRO. Each level applies its own judgment haircut based on that person's documented historical accuracy. A manager who calls 95% and delivers 95% gets her number. A manager who has missed six quarters running gets a haircut whether she likes it or not.

Step 6 — Record the call. Score it next week. What each rep committed goes into a log. Next week, you score it against what actually closed. This is how a forecast call becomes a trust ledger instead of a weekly ceremony where everyone says things and nothing is ever verified.

5. Build the roll-up. Keep it honest or it eats you.

                          CRO Forecast (to board)
                                   ▲
                      + CRO judgment / risk haircut
                                   │
    ┌──────────────────────────────┼──────────────────────────────┐
VP Sales Commit               VP Sales Commit               VP Sales Commit
    ▲                                                              ▲
+ Mgr haircut                                                 + Mgr haircut
    │                                                              │
Rep Commits  ─────────────── Rep Commits  ─────────────── Rep Commits

At every tier: Commit, Best Case, and Pipeline coverage. You want 3x coverage — open pipe against remaining gap — invoked like a blood ratio, like a covenant, like something you'd carve into the conference room wall if legal would let you. The CRO's number to the board is not the sum of rep commits. It is the sum, minus the haircut history demands, plus or minus what the CRO knows that the CRM doesn't. This is called judgment and it is the only part of the job that cannot be automated yet, though the machines are working on it and they are making unsettling progress.

6. Track forecast accuracy like it's the only metric that matters. Because it is.

Forecast Accuracy = Actual Closed ÷ Forecasted (Commit)

You want 95–105%. Inside that band, the board trusts you, CFO stops auditing your deal-by-deal, and the quarterly call is a professional conversation rather than a public execution. Outside it — chronically over (sandbagging; leaving growth and headcount on the table; your own team doesn't know what you'll spend) or chronically under (happy ears; you're a predictable liar with good slides) — and every forecast call becomes a goddamn deposition.

Accuracy beats magnitude. A predictable 95%-attainment org is worth more money than a wild 130%-then-70% org because the predictable one can be financed. You can hire against it. You can plan around it. The board can sleep. Sandbagging and happy ears both destroy the predictability that is the actual value you are delivering. This is not opinion. It is math.

HOW IT GOES TO HELL

The VP of Vibes runs the forecast call on adrenaline and raw feeling. "I'm feeling really good about Q3, team." No categories. No deltas. No historical context. Just vibes, issued confidently to a roomful of people who desperately want to believe them. His Commit is a mood. It misses by 22% and he is genuinely, authentically shocked — the exact expression of a man surprised by his own front door, every quarter, for four consecutive quarters, until the board asks whether he is perhaps the problem. He is.

The Brilliant Jerk carries a third of the number and updates the CRM when he feels like it, which is never, because he has been performing close magic for ten years and sees no reason to explain himself to a database. His forecast is a verbal rumor delivered with a smirk. You either build a process that survives his contempt — exit criteria that require artifacts, not feelings — or you let one closer hold the entire company's credibility hostage to his mood. The manual has opinions. Inspect the deals he won't discuss in the call. Especially those.

The Founder Who Still Thinks He's the Best Salesperson in the Building overrides the roll-up with a deal he "knows" will close because he golfed with the prospect's CEO and the energy was terrific. It slips two quarters. He overrides again. The forecast is now his autobiography — a document of hope that has never once matched the ledger of reality, maintained at enormous expense to everyone's sanity and the CFO's blood pressure.

The Guru — Lance "The Closer" DiMarVo — sells a webinar for $297 called "Throw Out Your Pipeline. A Real Closer Forecasts from the Heart." My attorney advises that forecasting from the heart is, in a board context, legally indistinguishable from securities fraud, and that the discovery process is unkind to people who used the word "vibes" in a Slack channel between weeks four and eight of the quarter.

The RevOps Martyr quietly maintains the only accurate model in the building — a beautiful, correct, data-grounded triangulation that she built over three weekends she will never get back — gets overruled by every VP's gut call, and watches the org miss the number she called to within $12K eleven weeks ago. Thank her. Actually use her number. Give her the meeting time. Let her present first. Or keep bleeding and wonder why.

The AI Model flags seventeen deals where the rep's Commit category conflicts with the model's probability score. Nobody looks. Nobody looks because the VP is in a hurry and the model is "just a machine" and the machine is right about fifteen of them. Three of those fifteen slip. One of them kills the quarter. The machine says nothing, updates its weights, and becomes better for next time, which is the most unsettling sentence in this entire manual.

FIELD RULES

• RULE No. 41: Commit is a promise, not a hope. If you can't bet your job on it, it's Best Case. If you're calling Best Case as Commit, you're committing perjury against yourself in a room full of witnesses.

• RULE No. 42: Accuracy beats attainment. The board can finance a predictable 95%. It cannot finance a goddamn surprise. Surprise is not a go-to-market strategy.

• RULE No. 43: Sandbagging and happy ears are the same crime from opposite directions — a forecast that doesn't mean what it says. Both rot the trust that is the whole point of doing this.

• RULE No. 44: Triangulate. Weighted for the floor. Rep judgment for the human truth. AI to referee the disagreements. Where they all agree, relax. Where they fight, inspect immediately.

• RULE No. 45: The forecast call inspects reality. It is not a data-entry séance. Submit before, discuss live. If they haven't submitted, they don't speak.

• RULE No. 46: Score every forecast against what closed. A forecast you don't grade is a wish you didn't keep, and you will make the same wish forever.

---

From the field: I once watched a CRO call Commit to the board on a deal that had already gone dark — the champion had quit, the champion's manager had quit, the whole company was in a hiring freeze — because admitting the slip felt worse than the lie. The slip cost him a quarter. The lie cost him the company. Reading the entrails is brutal, thankless, unglamorous work. The entrails do not have happy ears. They do not care about your vest. File this under: the truth is always cheaper than the cover-up, and the cover-up is always what shows up in the postmortem. Bourbon: $90. The truth: free, and nobody wanted it.