The B2B Sales Process: Stages, and Where Deals Actually Leak

Every B2B SaaS company has a sales process. Most of them have a diagram of one on a wall and a different one running in the CRM, and a third, unwritten one that the reps actually follow. The gap between those three is where the forecast goes wrong.

The B2B sales process is the defined sequence of stages a deal moves through from first contact to closed revenue, with a clear entry and exit criterion at each stage. That second clause is the whole game. A process is not a list of activities a rep performs. It is a set of gates a deal must pass through, and each gate is defined by something the buyer does, not something the seller hopes. I have built revenue and forecast models for B2B SaaS companies for two decades, and the single most common defect I find is a process where the stages are named after seller activity ("sent proposal," "had demo") instead of buyer commitment. Those processes forecast terribly, because seller activity is cheap and buyer commitment is not.

Let me lay out the stages that actually matter, then show you where the deals are leaking, which is almost never where you are looking.

The five working stages

Strip a B2B SaaS pipeline down and you get five stages that earn their place. Marketing owns everything before stage one. Lead, MQL, SQL, and the rest are handoff machinery, not the sales process proper, and folding them into your funnel math is how teams end up with a "conversion problem" that is really a lead-quality problem two departments upstream.

Stage	Entry criterion (buyer behavior)	Exit criterion	What it is really testing
1. Qualified opportunity	Buyer confirms a real problem and agrees to a working session	Discovery booked with the right people in the room	Is there a problem worth solving, and access to solve it?
2. Discovery and validation	Buyer shares context, current state, and decision process	Problem, impact, and decision path are documented and confirmed	Do we understand the problem well enough to win?
3. Solution and proposal	Buyer agrees the proposed solution fits and asks for terms	Proposal delivered against a confirmed need	Does our solution map to their actual problem?
4. Negotiation	Buyer engages on price, terms, or procurement	Verbal or written commitment, paper in motion	Is this deal real, or are we a column in a spreadsheet?
5. Closed	Signature and funding	Won or lost, with a reason code	What actually happened, and why?

Notice that every entry criterion is a buyer behavior. A deal does not enter negotiation because the rep emailed a quote. It enters when the buyer engages on terms. This distinction sounds pedantic until you watch what it does to forecast accuracy. When stages are defined by buyer behavior, a deal in stage four is genuinely more likely to close than a deal in stage two, and your stage-weighted forecast means something. When stages are defined by seller activity, a deal can be in stage four because the rep was busy, and the forecast is noise dressed as a number.

The count is not sacred. Some motions need a separate technical-validation or security-review stage. The rule is that you add a stage only when it has its own objective entry criterion and its own distinct conversion rate. If two adjacent stages convert at nearly the same rate and gate on the same buyer behavior, they are one stage wearing two name tags, and you should merge them. More on tightening the whole system in the sales process optimization guide.

The contrarian part: the leak is at the top, not the bottom

Here is the claim I will defend for the rest of this piece. When a B2B SaaS team has a conversion problem, the instinct is to look at the bottom of the funnel, at closing. That instinct is almost always wrong. The largest, most expensive leak in a B2B sales process sits at the entry gate, where weakly qualified deals are admitted, and it does not show up there. It shows up two and three stages later as a stalled deal, a slipped close date, a "we went dark" in the CRM.

Think about what a badly qualified deal looks like as it moves. It enters stage one because the rep needed pipeline and the buyer was polite. It survives discovery because nobody disqualified it. It reaches proposal, and then it stops, because there was never a real problem, a real budget, or a real decision path underneath it. The energy to advance a deal comes from the buyer's pain, and a deal admitted without pain runs out of road right around proposal. So the conversion report shows a leak at stage three. The team responds by coaching proposal skills and closing technique. They are treating the place the deal died instead of the place it was doomed.

This matters more right now than it did three years ago, because the market got harder. Median B2B win rates have fallen to 19% (First Page Sage, 2025), and sales cycles have lengthened 22% since 2022 (Digital Bloom, 2025). When win rates fall and cycles stretch, the cost of carrying a junk deal goes up. It occupies a rep, inflates your pipeline coverage with revenue that will never arrive, and stretches your average cycle as it drifts for months before anyone calls it. A loose entry gate in a 19% win-rate market is not a minor inefficiency. It is the difference between a forecast you can trust and one you cannot.

The fix is not at the bottom. It is to make stage one hard to enter. If the buyer cannot articulate a problem worth solving and give you access to the people who own it, the deal does not get a stage. It stays a lead. That single discipline does more for mid-funnel conversion than any amount of closing training, because it stops the deals that were going to stall from ever entering the count.

A worked example: the leak map at Solvorg

Numbers below are illustrative, not a benchmark. They exist to show the mechanism.

Solvorg is a mid-market B2B SaaS company. Their pipeline looks healthy: coverage is above target, activity is up, reps are busy. But closed revenue keeps landing short of the stage-weighted forecast, and nobody can say why. So we built what I call a Leak Map: stage conversion measured per stage, by segment, over a rolling window, instead of one blended funnel number.

Here is what the blended view said versus what the Leak Map found.

Stage transition	Blended conversion	Enterprise conversion	Mid-market conversion
Stage 1 to 2	78%	81%	76%
Stage 2 to 3	64%	41%	79%
Stage 3 to 4	55%	38%	68%
Stage 4 to 5	70%	66%	72%

The blended column looks like a stage-three problem. Proposal-to-negotiation sits at 55%, the lowest number in the funnel, so the obvious read is "our proposals are weak, coach the close." That read is wrong, and the segmentation proves it. Mid-market converts cleanly all the way down. The entire drag is in enterprise, and it does not start at proposal. It starts at stage two, where enterprise deals convert at 41% against mid-market's 79%.

That stage-two cliff is the tell. Enterprise deals are entering stage one, surviving the first transition on momentum, and then falling apart in discovery, because they were never qualified for the enterprise motion in the first place. When we pulled the deal records, the pattern was exactly the entry-gate leak: enterprise opportunities were being created off a single interested contact with no confirmed problem, no budget visibility, and no access to the buying committee. They cleared stage one because the bar was a conversation, not a commitment.

So the real diagnosis at Solvorg was not "fix proposals." It was "the enterprise entry gate is letting in deals that cannot survive discovery." The proposal-stage number was the place the deals visibly died, three stages downstream from where they were admitted. Coaching the close would have done nothing. Tightening the stage-one criterion for the enterprise segment, requiring a confirmed problem, a budget signal, and committee access before a deal could be created, is what moved the number. A single blended funnel chart would have sent Solvorg to coach the wrong stage for two quarters.

How to instrument the process so the leak shows itself

A Leak Map is only as good as the data feeding it, and most CRMs are configured to hide exactly the signal you need. Four instrumentation habits make the process readable.

Define every stage by an entry criterion, and enforce it. This is the foundation, and most teams skip straight past it. If a stage can be entered by seller activity, your conversion data is contaminated at the source. Write the buyer behavior that gates each stage, put it in the CRM as a required field, and make reps confirm it to advance a deal. A stage with no enforced entry criterion is a folder, not a gate. This is the same discipline that underpins honest sales pipeline metrics: the metrics are only as trustworthy as the stage definitions beneath them. Measure conversion per stage, by segment, never blended. The Solvorg leak was invisible in aggregate and obvious by segment. Enterprise, commercial, and SMB run different motions with different conversion shapes, and the blended average is a weighted lie that hides the one stage that is leaking. Cut every stage transition by segment as a default, not as a special investigation you run after you have already missed the quarter. Track time-in-stage alongside conversion. Conversion tells you whether deals advance. Time-in-stage tells you whether they are advancing or drifting. A deal that sits in discovery for ninety days has not converted and has not died, so it inflates your open pipeline while contributing nothing, and it quietly stretches your average cycle. Watching time-in-stage by stage is how you catch the drift that a conversion snapshot misses, and it ties directly to your overall sales cycle length, which lengthens silently when deals are allowed to camp in the middle. Require a reason code on every loss. A closed-lost with no reason is a lesson you paid for and threw away. When losses carry a structured reason, the Leak Map gains a second dimension: not just where deals leak, but why. If half your stage-two enterprise losses code as "no confirmed budget," you are not looking at a discovery problem, you are looking at the entry-gate leak again, confirmed from the exit side.

Run those four together and pipeline velocity becomes a number you can actually decompose. Velocity rolls deal count, deal value, win rate, and cycle into one rate, and when it drops, an instrumented process tells you which of the four moved and at which stage. Model it with a pipeline velocity calculator before you commit to a plan, then watch the instrumented stages tell you whether your fix is working.

Read the process from the top down

If you take one thing from this, make it the direction you look. When conversion sags, the gravity of the dashboard pulls you to the bottom of the funnel, to the proposal and the close, because that is where the deal visibly stopped. Resist it. The stage where a deal dies is rarely the stage that killed it. Walk the Leak Map upward from the leak, by segment, until you find the gate that should have held and did not, and far more often than not it is the entry gate, letting in deals that were never going to make it.

A tight entry criterion is the cheapest forecast accuracy you will ever buy. It costs you some pipeline that was going to evaporate anyway, and it returns a funnel where a deal in stage four genuinely belongs there. The catch is that an entry-gate leak never shows up at the entry gate. It surfaces three stages downstream, weeks later, disguised as a closing problem, which is precisely why it survives quarter after quarter in a same-period funnel report. You see it only when stage conversion is pinned to a live forecast and read by segment, which is the kind of model ORM builds, so the leak surfaces while there are still deals left to save.