The deal you cannot prove.

A deal’s forecast risk is not its stage. It is not the percentage sitting next to its name in the CRM. Forecast risk is a count: the number of things about that deal that nobody has actually verified.

Between 40 and 60 percent of qualified B2B deals end in no decision. Not a loss to a competitor. Not a win either. A deal that simply stops moving. We train hard to beat the other vendor and spend almost no time preparing for the opponent that kills more deals than any competitor does, which is nothing at all. The most common way a deal dies is that nothing anybody could point to was ever actually confirmed to be true.

I have sat in enough forecast calls to know what an unverified deal sounds like from the outside. It sounds exactly like a verified one. Confidence is not evidence, and closing that gap is what this issue is for.

How do you know whether a deal is real?

Deal inspection is testing one live deal, claim by claim. Every claim behind that deal needs a name and a date attached to it: who told you, and when they told you. Whatever cannot be named becomes the deal’s forecast risk. Not a feeling. A count.

Three tools do three different jobs here, and managers tend to collapse them into one word and lose all three. One works on the pipeline: is there enough of it, and where did it come from. That is pipeline coverage quality. One works on a single opportunity: is this one real, and how would you know. That is deal inspection. One works on the total: does the number still hold when a real deal moves anyway. That is forecast survivability. The middle one feeds the last, and you cannot answer the number honestly without first answering the deal.

Philip E. Tetlock and Dan Gardner, Superforecasting: The Art and Science of Prediction, 2015

Tetlock spent years running large forecasting tournaments, comparing people who guessed against people who consistently got it right. The ones who got it right were not the most confident and were not always the most experienced. They broke a big vague question into small checkable pieces, and they updated the moment new evidence arrived.

A commit percentage is a treasure being guarded. Nobody wants to open it up and look inside. The six lines below are the same deal treated as a hypothesis instead.

1. The signature. Who signs. A name, and the date they confirmed it to you, or the word unknown.
2. The money. Where the budget comes from, whether it is approved, and who told you so, on what date they said it.
3. The date. Not the date they gave you. What event on their calendar produces it.
4. The alternative. What they do today instead of buying, and what that costs them to keep doing. This is where no decision lives. If doing nothing is free for them, doing nothing is what they will do.
5. The dependency you do not control. What has to be true on their side for this to close, and who on your side verified that it is.
6. The behavior. What the customer did since last week. Not what they said. What they did, with a name and a date attached to the doing.

Six questions aimed at a live deal read as scrutiny to the exact people you most need talking to you honestly. If nobody says that out loud, the whole routine lands as a manager who does not trust his reps, and every good rep in the room already knows that sound from somewhere else in their career.

The fix is not tone. It is structure. The six questions are addressed to the deal, not to the rep. Same six, same order, every deal, every time, whether it is somebody’s best quarter or their third bad month. Nobody in that room is being interrogated. The deal is being tested. That distinction is the entire reason anybody keeps telling you the truth about a deal after the first time you ask.

Run that way, accountability is not suspicion. It is respect. You are telling every rep that a deal has to earn its number the same way every other deal does, including the ones you like best.

The spike and the system.

Inspecting the deal that is keeping you up at night is a spike. Any manager can do that once, under pressure, with a bad feeling and a highlighter.

Inspecting every deal the same way, in the same order, whether or not it worries you, is a system. A team that only inspects under pressure teaches itself something true and unhelpful: that inspection is what happens right before something bad. Reps learn that fast, and then they learn to hide the early signs instead of bringing them to you.

That is Isolate, Refine, Compound at the level of one habit. Isolate the six questions. Refine the routine on the deals that do not scare you yet. Compound it into something you run without having to decide to.

Thirty-plus years carrying a number, a lot of those years now spent coaching managers rather than reps, and the pattern repeats in every room I have been in. The managers who inspect only the frightening deal train their team to hide the early signs of a frightening deal. The managers who inspect everything, on schedule, get told about problems while the problems are still small.

Run this.

The Ten-Minute Inspection. One deal. Ten minutes. Six lines. One question, asked six times. Run it on the first Monday of the quarter, and on every deal the moment it enters commit.

1. Step 1. Pick the deal carrying the most of the quarter. This is single-deal exposure: when one deal carries more than 40 percent of the quarter, it fails to close about four times out of five. That is the deal to inspect first, and it is the one nobody wants to touch.
2. Step 2. Write six lines on one page, in order: the signature, the money, the date, the alternative, the dependency you do not control, the behavior. A name and a date on every line, or the word unknown. No adjectives. “Strong relationship” is not an answer. A name is.
3. Step 3. Ask the only script you need. “How do we know that?” Once per line, same flat tone, addressed to the deal and not to the rep. No second question. No softer version for the deal that worries you. Let the silence do the work. Asking it identically every time is what makes this a routine instead of a verdict.
4. Step 4. Count the unknowns and write the number at the top of the page. That number is the deal’s forecast risk. Not the stage. Not the percentage. Zero unknowns is a commit. Three or more is not a commit, it is a project.
5. Step 5. Give every blank an owner and a date this week. The inspection is not finished when the questions are asked. It is finished when the blanks have names next to them.
6. Step 6. Re-inspect on change only. Run the full six once, at commit. After that, reopen a single line when that line moves. The weekly one-line version of this is the Tuesday question, pointed at whichever line changed.

What it changes.

1. Time. The forecast call stops being an argument, because you are comparing counts instead of opinions. A zero-blank deal takes a minute.
2. Opportunity. The blanks are the week’s actual work. An inspection produces a to-do list where a confidence percentage produced nothing at all.
3. Compensation. A blank found in week two is a deal you can still save. The same blank found in week twelve is a quarter you cannot.

The short version.

1. Forecast risk is a count of unverified claims. Not a stage, not a percentage.
2. Deal inspection tests one live deal, claim by claim. A name and a date, or the word unknown.
3. Between 40 and 60 percent of qualified B2B deals end in no decision, and inspection is built for that opponent rather than for the competitor.
4. Pipeline coverage quality, deal inspection and forecast survivability are three tools working at three different levels. Inspection is the input to survivability.
5. The six lines: the signature, the money, the date, the alternative, the dependency you do not control, the behavior.
6. One question, asked once per line, in the same flat tone, addressed to the deal and never to the rep.
7. Zero unknowns is a commit. Three or more is a project wearing a commit’s clothes.
8. A frightening deal inspected once is a spike. Every deal inspected the same way, every time, is a system.

Here is your homework: before your next forecast call, pick the one deal carrying the most of your number and run the Ten-Minute Inspection on it. Six lines on one page, a name and a date on every line or the word unknown, and one question asked flat, once per line: how do we know that? Count the unknowns and write the number at the top. Then give every blank an owner and a date this week. A deal has to earn its number the same way every other deal does, including the ones you like best.

This essay ran as edition 17 of The Frontline Sales Forecast, the weekly newsletter. One issue a week, published here on the Friday, then emailed and posted to LinkedIn the same morning. Same piece everywhere. Get the next one. Disagree with it? Reply to any issue, or find me on LinkedIn.