Sales pipeline stages: define them by the buyer

A VP of Sales I work with walked into a board meeting last year with $3.4M in stage-four pipeline and a quarter target of $1.1M. On paper he was covered three times over. He committed to $1.4M. He closed $680K. After the quarter, the board stopped trusting his forecast, and honestly, so did he.

We pulled the deals apart afterward. Stage four was called "proposal." It turned out four different reps moved a deal there for four different reasons. One did it when the buyer asked for pricing. One did it the moment he emailed a deck, before anyone had read it. One used it as a parking spot for deals he did not want to mark lost. The stage had a name and a probability attached to it, and it meant nothing, because nobody agreed on what had to be true for a deal to live there.

That is the problem with pipeline stages at most B2B companies. Not the number of them. Not the labels. The fact that the stages measure what the rep did instead of what the buyer committed to, so the forecast built on top of them is fiction. I have spent the last decade designing pipeline systems at companies between $2M and $80M ARR, and I want to give you the actual blueprint: the stages, the exit criterion for each one, and how to turn those stages into a forecast you can stake a quarter on.

Your pipeline stages are probably measuring the wrong thing

Open your CRM and read your stage names. If they sound like contacted, demo booked, proposal sent, negotiation, you have an activity pipeline. Every one of those describes an action your rep took. None of them describe anything the buyer did.

This matters because activity does not predict revenue. A rep can send a proposal to a buyer who will never sign. A rep can run a demo for someone with no budget and no authority. The action happened, the stage advanced, the forecast went up, and the deal was dead the whole time. You built a number out of motion and called it pipeline.

A buyer-action pipeline flips it. The stage advances only when the buyer does something that costs them effort: replies with a real next step, loops in a second stakeholder, gives feedback on price, agrees to a timeline. Those are commitments. Commitments cost the buyer something, and things people pay for, even in effort, are things they tend to follow through on. That is the entire principle, and most teams have it backward.

I dug into why activity stages wreck the weekly number in the pipeline management guide. This post is about the thing one layer underneath that: how to design the stages themselves so the management part even has a chance.

What a pipeline stage actually is

A stage is a checkpoint in the buyer's journey, defined by one observable thing that must be true before a deal is allowed in. That last part is the whole game. The "one observable thing" is the exit criterion, and if you cannot write it as a sentence a rep cannot argue with, the stage should not exist.

"Qualified" is not an exit criterion. It means something different to every person who reads it. "The buyer named a business problem, a rough timeline, and the person who owns the budget" is an exit criterion. You can look at a deal and answer yes or no in five seconds. No interpretation, no happy ears, no rep telling you the call "felt strong."

Here is the difference, side by side, for a few common stages.

Notice that the right column does not care how the deal got there. It does not reward the rep for effort. It only asks one question: did the buyer do the thing that signals real intent? That is what makes the stage worth a probability.

The six stages, and the exit criterion for each

Most B2B sales motions fit cleanly into six stages. You can run five, you can run seven, but six is the number I reach for first because it maps to how buyers actually move without giving deals extra places to hide. Here is the blueprint, with the buyer action that lets a deal leave each stage.

Stage one is lead, sometimes called prospect. This is a contact who matches your ideal customer profile and has shown a flicker of interest, but you have not spoken yet. The exit criterion is simple: the buyer agreed to a real conversation, a booked call with a time on the calendar, not "sure, send me some info." If they will not give you 20 minutes, they are not in your pipeline yet, they are in your marketing nurture. Keeping those two things separate is the single biggest cleanup I do on new engagements.

Stage two is discovery, or qualification. You are on a call digging into the buyer's actual problem, their timeline, and who controls the money. The exit criterion is that the buyer confirmed a problem worth solving and a rough timeframe, and you have identified the economic buyer. This is where frameworks like MEDDPICC earn their keep, and I broke that down in the MEDDPICC guide. If you cannot name the problem and the buyer in one sentence, the deal does not leave discovery, no matter how friendly the call was.

Stage three is solution validation, usually the demo or technical evaluation. You are showing how you solve the problem you uncovered. The exit criterion is not "we did a demo." It is that the buyer agreed the solution fits and pulled a second person into the deal. A single-threaded deal that never expands past your first contact is the most common way a quarter quietly disappears, which is why I am so blunt about multithreading. One champion who goes silent and your whole deal goes with them.

Stage four is proposal. Pricing is on the table and the buyer is evaluating it seriously. The exit criterion is that the buyer engaged with the proposal: asked questions, pushed back on a line item, requested a change, anything that proves they read it and are weighing it. A proposal that vanishes into silence is not a stage-four deal. It is a stalled deal wearing a stage-four costume.

Stage five is negotiation and commit. You are working through terms, redlines, procurement, and security review. The exit criterion is a mutual action plan with real dates that both sides have agreed to, the document that says "we sign on the 14th and here are the five steps to get there." If you do not have agreed dates, you have a verbal yes, and verbal yeses slip. I keep a live mutual action plan on every deal at this stage for exactly this reason.

Stage six is closed, won or lost. The only rule that matters here is that you mark losses honestly and capture why, because a pipeline full of deals nobody will call dead is the thing that broke that VP's forecast in the first place. Lost deals are also your best teacher, which is the whole point of running win-loss analysis.

The thing to take from this blueprint is not the exact six labels. It is that every stage has one buyer action attached, written so plainly that two reps looking at the same deal would agree on where it sits. Get that and your pipeline stops being a story.

Lead stages and deal stages are not the same thing

Here is a mistake I see at least once a month. A team crams their entire funnel into one set of pipeline stages, so "subscribed to newsletter" and "in legal review" sit in the same dropdown. Then they wonder why their conversion math is nonsense.

Lead stages and deal stages are two different objects doing two different jobs. Lead stages, often called lifecycle stages in HubSpot, describe a contact's relationship with your company before there is a real deal: subscriber, lead, marketing qualified, sales qualified. Deal stages describe an active opportunity moving toward a signature. A contact graduates from the lead world into the deal world at one specific moment: when a salesperson accepts them as a real opportunity worth working.

Blur that line and two things break. Your marketing team takes credit for "pipeline" that is really just a list of email subscribers, and your sales conversion rates get diluted by people who were never going to buy. Keep the two sets of stages separate, define the handoff point precisely, and your funnel math starts telling the truth. If you run HubSpot, I walked through the specific setup in the deal stages guide.

Turning stages into a forecast you can trust

This is where stages pay off. Once each stage has a clean exit criterion and reps actually obey it, you can attach a probability to each stage based on how often deals there actually close, and build a weighted forecast. The math is plain: multiply each deal's value by its stage probability, add it up, and you have a weighted pipeline number.

The catch is the probabilities. Most teams pull them out of thin air. Stage four is "75%" because someone decided that in 2019 and nobody questioned it. Real probabilities come from your own history. If deals that hit stage four close 40% of the time over the last year, the probability is 40%, not 75%. Pull your last 12 months of closed deals, calculate the actual conversion rate at each stage, and use those numbers. Recalibrate every quarter as the data shifts.

When you do this well, the payoff is real. Stage-based forecasting can land in the 85% to 95% accuracy range for the current quarter when the stages are clean and the probabilities come from history. When they are not, you get what most teams have: fewer than one in five sales leaders rate their forecast as predictable. The difference is almost never the software. It is whether your stages mean anything and whether your probabilities are honest.

One warning on probability. Reps will inflate it if you let them set a free-form "% to close" field on each deal. They are optimists by trade, and the number creeps up under quota pressure. Tie the probability to the stage, not to the rep's gut, and recalibrate from data. The rep's job is to record what the buyer did. The stage decides the probability. That separation is what keeps the forecast honest, and it connects straight to forecast accuracy and your pipeline coverage math.

How to rebuild your stages without breaking the quarter

If your current stages are a mess, do not blow them up mid-quarter and re-stage everything overnight. That just creates a different kind of chaos. Do it in order.

That last step stings the first time. When the VP and I re-staged his board off buyer actions, his stage-four number fell by more than half overnight. It felt like a loss. It was actually the first time in a year his pipeline told him the truth, and the next quarter he committed to a number he hit within 4%. A smaller honest pipeline beats a fat fictional one every single day.

Most of this is design work, not tooling. Your CRM already supports custom stages and exit criteria, whether you run HubSpot, Pipedrive, or Salesforce. The hard part is the discipline to define the criteria clearly and hold reps to them. That is the core of what we do inside CRM and RevOps engagements and go-to-market work, and it is the cheapest high-return fix in all of sales operations.

Frequently asked questions

How many sales pipeline stages should a B2B company have?

Five to seven for almost every B2B motion, and I default to six. Fewer than five and you lose visibility into where deals actually sit. More than seven and the extra stages become parking spots for deals nobody wants to call dead. The number matters less than the rule that every stage has one clear exit criterion based on a buyer action. If you cannot write that sentence for a stage, delete the stage.

What is a pipeline stage exit criterion?

It is the single observable thing that must be true before a deal is allowed into a stage. Good exit criteria describe what the buyer did, like "agreed to a mutual action plan with dates," not what the rep did, like "sent a proposal." The test is whether two reps looking at the same deal would agree on its stage. If the answer is no, the criterion is too vague and your forecast will inflate.

Should pipeline stages be based on rep activity or buyer actions?

Buyer actions, always. Rep activity tells you what your team has been busy doing, which has almost no relationship to whether a deal will close. Buyer actions cost the buyer effort, and effort signals real intent. A stage that advances on "we did a demo" predicts nothing. A stage that advances on "the buyer pulled a second stakeholder into the deal" predicts a lot. Build the stages around the buyer and the forecast follows.

How do I assign probability to each pipeline stage?

Use your own history, not default numbers. Pull your last 12 months of closed deals, calculate what percentage of deals at each stage actually went on to close, and set those rates as your stage probabilities. Recalibrate every quarter because the numbers drift as your market and motion change. Never let reps set a free-form close probability on individual deals, since it inflates under quota pressure. Tie probability to the stage and keep it honest with data.

What is the difference between lifecycle stages and deal stages?

Lifecycle stages, sometimes called lead stages, describe a contact's relationship with your company before a real opportunity exists: subscriber, lead, marketing qualified, sales qualified. Deal stages describe an active opportunity moving toward a signature. A contact crosses from one to the other at a single defined moment, when a salesperson accepts them as a real opportunity. Mixing the two in one pipeline ruins your conversion math and lets marketing claim subscribers as pipeline.

Build stages your forecast can stand on

Pipeline stages are not labels on a Kanban board. They are the load-bearing structure under your entire forecast, and most teams build them out of rep activity, then wonder why the number lies. Define each stage by one buyer action, write the exit criterion so plainly that nobody can argue it, calibrate the probabilities from your own history, and re-stage your open deals against the new rules. The pipeline gets smaller and a lot more honest.

If your stages have drifted into meaning nothing, book a free audit. We will re-stage a slice of your pipeline on buyer actions and show you the real forecast hiding underneath. You can also see how we approach this in CRM and RevOps and AI automation.