Give every pipeline stage an exit criterion.

Open your CRM and read the stage names out loud. If most of them describe something your team did rather than something the buyer did, your forecast is measuring your own activity and reporting it as likelihood.

That isn't a configuration mistake somebody made at your company. It's what the software ships with.

What are CRM pipeline stages?

A CRM pipeline stage is a named position a deal occupies between created and closed. Each stage usually carries a probability, and most CRMs use that probability to compute a weighted forecast.

That last sentence is the one that matters, and almost nobody reads it. The stage is not a label for humans. It is an input to a calculation.

Your CRM multiplies money by the stage name

HubSpot's documentation is explicit about the mechanic. The weighted amount shown in board view is, in its own words, "calculated by multiplying the total amount in each stage by the stage probability."

So the forecast is deal size times a number that comes from the stage. Which means the stage names are doing arithmetic, and it's worth looking at what they actually say.

This matters past the forecast too. The weighted number is one of the few figures a CRM computes for you unasked, and several of the ones you actually want it cannot compute at all.

Look at what moves a deal from 20% to 60%. Somebody put a second meeting in a calendar. The buyer may have agreed to attend it out of politeness, or because a junior analyst is collecting vendors, or because they always take the meeting and never buy. Nothing about their intent has changed, and the forecast has tripled.

Then look at 90%. We sent a contract. Every person who has ever run a pipeline knows what sits in the gap between sending a contract and having one signed: procurement, legal, security review, a budget that turns out to belong to a different department, and the quarter ending.

Exactly one of those five open stages describes something the buyer did. It's the one at 80%, and it's the only one I'd argue with less.

The vendors recommend it, which is why everyone does it

I went looking for the default stage names the other major CRMs ship, expecting to build a comparison table. I couldn't, and the reason turned out to be more interesting than the table.

Pipedrive doesn't publish its defaults. Its knowledge base says a new pipeline arrives as "a sample five-stage pipeline that you can edit" and never names the five. The screenshots on the page carry empty alt attributes, so the names aren't in the markup either. The list you'll find quoted everywhere else traces to no Pipedrive page I could open, so I'm not repeating it here.

What that page does carry is guidance, and the guidance is the finding. Pipedrive tells you to name stages after real actions, and the two examples it gives as good practice are "Meeting scheduled" and "Proposal sent". Both describe the seller.

Salesforce's documentation wouldn't render at all. Their help article and their object reference both load as empty JavaScript shells, and the help article ID for opportunity stages returns a 404. The default list everyone quotes for Salesforce appears only in secondary sources, so it stays out of this post.

That leaves a smaller claim than I wanted and a firmer one than I expected: the one vendor that documents its defaults ships a pipeline named mostly after seller activity, and the one vendor that documents its advice recommends exactly that. This isn't a mistake spreading between companies. It is the default, and it is taught.

A stage without an exit test is an opinion

Here's the practical version of the problem, and you can check it this afternoon.

Take a deal sitting in your third stage. Ask the rep who owns it what specifically has to be true for it to be in that stage. Then ask a different rep the same question about one of theirs. If the two answers don't match, your stages aren't stages. They're a shared vocabulary that everyone pronounces differently, and every report built on top of them inherits the disagreement.

This is why forecasts move week to week without anything happening in the market. The deals didn't change. Somebody's interpretation did.

An exit criterion is the fix, and it has to satisfy three things at once or it doesn't hold up.

1. 1.
   Binary, not a judgement

   Yes or no, with no middle. "The buyer has given us a date they need this working by" is binary. "The buyer seems engaged" is a feeling wearing a field name. If two people can look at the same deal and disagree, it isn't a criterion.

2. 2.
   Produced by the buyer, not by us

   The test is who had to act. We can schedule a meeting without the buyer doing anything except not declining it. We cannot get them to name a budget holder, forward an internal requirements doc, or put us in front of the person who signs. Those cost them something, and things that cost the buyer something are the only reliable signal you get.

3. 3.
   Stored in a field the CRM can check

   A criterion that lives in a rep's notes is not enforceable, and a criterion nobody enforces decays within a quarter. It needs a field, and the field needs to be required before the stage can change. This is the part every article on this topic skips.

Six stages, six exit tests, and the field that enforces each

This is the part that isn't published anywhere else. Every page ranking for this topic says exit criteria should exist. None of them shows the field and the rule that make one real.

Two things about this table are the point.

Every exit is something the buyer had to do. Reply, name a budget owner, hand over a constraint, come back with an objection, name a signatory, open their own process. None of them can be manufactured by a rep having a productive week.

Every exit has a field, and every field is required. Not "we agreed the definition in a workshop." A required field, checked on stage change, in the CRM. HubSpot, Pipedrive and GoHighLevel all let you require a property before a deal moves; the setting is in the pipeline or stage configuration rather than on the field itself, which is why people miss it.

Stage 4 is the one worth arguing about. Waiting for an objection feels like waiting for bad news, but a proposal that comes back with no response at all is the single most reliable predictor of a deal that never closes. Silence is information, and the field records it either way.

What to check in your own pipeline this week

Three checks. Each has a failing condition, so reading them gives you an answer rather than a to-do list.

Count the stage names that describe your side. Open the pipeline settings and read the list. It fails if more than two of your open stages describe something your team did. That is the same defect as the shipped default, and you've inherited it rather than chosen it.

Ask two reps to define your middle stage. Not the first, which everyone agrees on, and not the last. It fails if the two answers are not the same sentence. You do not have a definition, you have a word.

Try to move a deal forward with every field empty. Pick a test deal, clear nothing, just drag it. It fails if the CRM lets you. Every exit criterion you believe you have is advisory, and advisory criteria are the ones that quietly stop being applied in the second month.

If all three pass, your stage data is better than most and your forecast problem is somewhere else, probably in what the records themselves are worth or in who is allowed to change them.