Deal Stages for a B2B SaaS Startup (With Exit Criteria)

Open the CRM at most seed-stage startups and you will find a stage called "Negotiation" holding eleven deals. Ask the founder how many of those buyers have seen a price, and the answer is four. Ask how many have a signer identified, and it drops to two.

That gap usually has a simple cause. The stages have names but no definitions, so everyone moves a deal forward when it feels like it moved, and the forecast inherits all of that optimism.

The fix is exit criteria: a short list of things that must be true before a deal leaves a stage. This guide gives you a set of deal stages that works for most B2B SaaS startups selling contracts between $5k and $100k a year, with the criteria written out.

Why deal stages need exit criteria

Without exit criteria, a stage measures the rep's mood. With them, it measures what the buyer has done. That difference shows up everywhere:

• Forecasts get more accurate, because probability by stage starts to mean something.
• Pipeline reviews get shorter, because "why is this in Proposal?" has a factual answer.
• New reps ramp faster, because the stages tell them what to do next.
• You find the real bottleneck. If 30 deals pile up in Discovery and few leave, the problem is qualification or the demo, not closing.

A six-stage pipeline for B2B SaaS

Stage	Exit criteria (done when)	Starting probability
Qualified	A real problem, a plausible buyer, and a first call booked	10%
Discovery	Problem, impact and decision process confirmed with the buyer	20%
Evaluation	The buyer has seen the product against their own use case and agreed what success looks like	40%
Proposal	Pricing sent and discussed, economic buyer known, a target signing date agreed	60%
Negotiation	Verbal yes, contract or order form in review, legal and security underway	80%
Closed won or lost	Signed, or a clear no with a reason recorded	100% or 0%

The probabilities are starting points, not statistics. After two quarters, replace them with your own conversion rates.

Qualified

The deal exists. Someone with a real problem has agreed to talk.

• The company fits who you sell to (size, industry, use case).
• You know the contact's role and why they took the call.
• A first meeting is on the calendar and accepted.

Discovery

You understand the problem well enough to explain it to someone else.

• The buyer has described the problem and what it costs them, ideally with a number.
• You know how they will make a decision and roughly when.
• You know who else is involved, by name or at least by role.
• The buyer agreed to a next meeting to see the product.

Evaluation

The buyer is testing whether your product solves their problem.

• They have seen a demo built around their use case, not a generic tour.
• If there is a trial, it has a start date, an end date and written success criteria.
• At least two people at the buyer have been involved. (More on why in this guide to single-threaded deals.)

Proposal

The conversation is now about terms.

• Pricing has been sent and discussed on a call. Sending a PDF does not count.
• You know who signs and they know the deal exists.
• There is a target signing date that the buyer agreed to, not one you picked.

Negotiation

You are working out details, not whether to buy.

• The buyer has said yes in principle.
• Contract, order form or procurement paperwork is in their hands.
• Legal and security review have started, with dates.

Closed won or closed lost

Closed won means signed. Closed lost means a clear no or a deal that has gone silent for longer than your sales cycle. Either way, record the reason.

Write exit criteria as things the buyer did

The most common mistake is writing criteria about the rep's activity: "sent proposal", "did demo", "followed up twice". Those are inputs. A rep can send a proposal into a void.

Compare:

• Rep activity: "Sent trial invite."
• Buyer action: "Buyer started the trial and agreed to success criteria."

The second version is harder to fake and much better at predicting whether the deal will close. When you review a deal, you can check the evidence: an email, a meeting note, a signed form.

How many deal stages do you need?

Five to seven open stages covers almost every early-stage B2B sales process. Fewer than four and the stages stop telling you anything. More than seven and reps stop updating them, because the difference between "Proposal sent" and "Proposal reviewed" is not worth the click.

If you run more than one motion, use separate pipelines rather than one long list. A self-serve upgrade that closes in a week does not need the same stages as a six-month enterprise deal. Renewals and expansions usually deserve their own pipeline too.

When to override the stage probability

Default probabilities work across a portfolio of deals. For a single deal they are often wrong, and that is fine. Override when you have specific evidence:

• Lower it when the close date has already moved once, when only one person is engaged, or when the economic buyer has never been on a call.
• Raise it rarely, and only with something written: a signed order form waiting for a countersignature, for example.

Watch for deals where the rep's probability is well above the stage default. That is often the first sign a deal is about to slip.

Closed lost reasons worth tracking

Keep the list short so people actually pick one:

• No budget or budget moved
• Chose a competitor
• Chose to do nothing or build it themselves
• Timing (revisit later)
• Not a fit
• Went silent

After a quarter, look at where lost deals died. If most of them die in Evaluation with "chose to do nothing", your problem is the size of the pain, not your pricing.

Rolling out new stages to a small team

1. Write the stages and exit criteria in one shared document and walk through them in a single 30-minute meeting.
2. Re-stage every open deal against the new criteria in one sitting. Expect a lot of deals to move backwards. That is the point.
3. Put the criteria where reps update deals, not in a wiki nobody opens. In Heed, each stage has a "done when" checklist that admins can edit, and the deal page shows what is still missing to reach the next stage.
4. Review the stages after 30 days. Merge any stage that deals skip through, and split any stage where deals sit for weeks without anyone knowing why.

The first re-staging is uncomfortable, because the pipeline shrinks on paper. The deals were never really there. Now you know which ones are.