Your forecast is a feeling with a spreadsheet around it

Every company I walk into with a forecasting problem believes they have a discipline problem.

The theory goes like this. The reps are optimistic. They do not update the CRM. If I inspected harder, ran more deal reviews, and made people justify their numbers, the forecast would tighten up. So they add inspection. The forecast gets worse, or it gets better for one quarter and then reverts, and everyone concludes they need to inspect even harder.

I have watched this cycle at companies with nine reps and companies with ninety. It almost never works, and the reason is that the diagnosis is wrong. It is not a discipline problem. It is a definitions problem, and no amount of inspection fixes a bad definition.

Look at your stage names

Open your CRM and read your deal stages out loud. In most organizations, they sound something like this: Discovery. Demo. Proposal. Negotiation. Closed Won.

Now notice what every one of those words describes. Discovery is something the seller ran. Demo is something the seller delivered. Proposal is something the seller sent. Negotiation is a thing the seller believes is happening.

Every stage in that pipeline is a description of seller activity. Not one of them describes anything the buyer did.

Which means a deal can sit in Proposal for eleven weeks. The seller genuinely did send a proposal. The stage is accurate. And the deal is completely dead, because the person who received the proposal left the company in week two and nobody on the buying side has opened it since.

Your forecast is not lying to you. It is faithfully reporting what your reps have been up to. It just has nothing to say about whether anyone intends to buy.

Why this always inflates in one direction

Here is the part that makes it structural rather than cultural.

Seller activity is fully under the seller's control. A rep can move a deal to Proposal unilaterally, at any time, without a single thing changing in the buyer's organization. The gate is on the wrong side of the table.

Buyer evidence is not under the seller's control. A rep cannot manufacture a second stakeholder joining a call. They cannot manufacture a customer confirming a budget line or naming a business consequence of doing nothing. Those things either happened or they didn't.

So a pipeline built on seller activity has no natural brake. Every incentive in the system, including entirely honest ones like wanting to show progress in a Monday meeting, pushes deals forward. Nothing pushes back. Over a quarter that compounds into a number that is not just wrong but wrong in a predictable direction, which is why your forecast probably misses high almost every time.

You cannot inspect your way out of that. You can only redefine it.

Buyer-verifiable exit criteria

The fix is to make every stage transition require evidence that came from the buyer, not from the seller.

Not a judgment call. Not a confidence level. A specific, checkable fact that either exists or does not, and that a manager can confirm in ten seconds without a conversation.

The exact criteria depend on your motion, but the shape looks like this.

A qualified deal is not one where you ran discovery. It is one where the buyer has named a business consequence of not solving the problem, in their words, and you have it written down. If you cannot state what happens to their business if they do nothing, you have not qualified anything.

An evaluating deal is not one where you gave a demo. It is one where more than one person on the buying side has engaged, and you know who else has to say yes. Single-threaded deals are not evaluating. They are a conversation with one interested person.

A committed deal is not one where you sent a proposal. It is one where the buyer has told you their process, their timeline, and who signs, and has confirmed a next step with a date attached that they own rather than you.

The moment you write criteria this way, something uncomfortable happens. The first time you apply them honestly, a large portion of your pipeline moves backward. At one company it was slightly more than half.

That is not a disaster. That is the first accurate picture of the business anyone there had seen. The pipeline did not shrink. It was always that size. You were just measuring it wrong.

Three questions that will do most of the work

If rebuilding your stage definitions feels like a project you will not get to, start with three questions in your next pipeline review. Ask them about every deal above a threshold you care about, and ask them the same way every time.

Who else has to say yes, and have you met them? Single-threaded deals fail at a dramatically higher rate than multi-threaded ones, and the failure is usually invisible until it is total. If the rep cannot name the other people, the deal is not where they think it is.

What happens to their business if they do nothing? If the answer is a version of things stay the same, there is no deal. Not a slow deal. No deal. Most stalled pipeline was lost to the status quo rather than to a competitor, and the status quo wins when nothing bad happens by choosing it.

What is the next step, what is the date, and who owns it? The word who is the whole question. If the rep owns every next step in the deal, the buyer is not participating. They are being pursued. Those are different situations and only one of them closes.

Three questions. Most reps will fail at least one on most deals the first few weeks. That failure rate is the actual state of your pipeline, and it is worth more than any weighted probability calculation you could run on top of the old data.

The thing that makes it stick

One warning, because I have watched this fail more than once.

If you redefine your stages and then treat the resulting pipeline drop as a performance problem, you will have taught your team exactly one lesson: honesty is punished. Within a quarter the numbers will drift back to comfortable and you will have spent political capital for nothing.

Say plainly, before you start, that the number is going to look worse and that this is the point. Then hold the line for a full quarter, because the first accurate quarter is the one that establishes whether people can trust the new rules. The forecast accuracy you get afterward is worth more than the pipeline number you gave up, because you can finally make decisions on it. Hiring, spend, and what you tell your board all depend on the number meaning something.

What to do in the next thirty days

Week one, write down what buyer evidence should be required at each stage. Not what your reps do. What the buyer must have done. Keep it to one checkable fact per stage.

Week two, apply it to your current pipeline and let deals fall where they fall. Do not soften it. Do not grandfather anything.

Week three, tell the team what changed and why, and say the part about the number looking worse out loud.

Week four, run your pipeline review with the three questions and nothing else.

Do that and you will have something most companies at your stage do not: a number that means the same thing every time somebody says it.