What sales pipeline stages are — and what they aren’t
Sales pipeline stages are the checkpoints a deal passes through between first contact and a signature, and their entire job is to make two things true: any rep can tell you where a deal really stands, and the team can forecast from the distribution of deals across stages. That is it. A stage model is a shared language, not a methodology — it should describe the buyer’s progress so plainly that two reps looking at the same deal file it in the same stage.
What stages are not: a to-do list, an activity log, or a forecasting hack. The moment a stage describes seller behavior (“contacted”, “followed up”) instead of buyer state (“qualified”, “evaluating”), the pipeline stops measuring reality and starts measuring effort — and effort does not close.
The standard model: five stages with entry and exit criteria
Most working B2B pipelines converge on the same skeleton, whatever the labels. Here it is with the part most teams skip — explicit entry and exit criteria for every stage:
| Stage | Entry criteria | Exit criteria (buyer-verifiable) | Typical action |
|---|---|---|---|
| 1. Lead | A real person at a real company has shown interest or fits the ICP | Two-way contact established; basic fit confirmed | Fast first response; research the account |
| 2. Qualified | Fit + need confirmed in a real conversation | Buyer confirms a problem worth solving, a budget range, and a decision process | Discovery call; identify the decision-makers |
| 3. Evaluation | Buyer is actively assessing the solution | Success criteria agreed; demo/trial/scoping completed with the decision-makers | Demo, trial support, technical validation |
| 4. Proposal | Buyer has requested pricing or a formal proposal | Proposal delivered and verbally accepted; legal/procurement engaged | Negotiate; drive to a signature date |
| 5. Closed | Contract signed — or a definitive no | Won: payment/contract. Lost: reason logged | Handoff to onboarding, or a logged loss reason |
Two design notes. First, the criteria are about the buyer, not the rep — “buyer confirms budget” can be checked; “rep feels good about budget” cannot. Second, closed-lost is a stage, not a shameful deletion: loss reasons are the cheapest sales research you will ever collect, and a pipeline that quietly drops dead deals loses them.
Sales stages examples: SaaS vs. services
The skeleton holds across business models; only the middle changes. In SaaS, stage 3 is usually a trial or product evaluation — the buyer proves value to themselves inside the product, and the strongest stage-3 exit signal is product usage (a team invited, an integration connected), not a meeting. In services and agencies, stage 3 is scoping — the evaluation happens in conversations, and the exit artifact is an agreed statement of work.
| Stage | SaaS example | Services example |
|---|---|---|
| Lead | Signed up for the newsletter; identified company visited pricing | Referral intro; inbound inquiry from the website |
| Qualified | Discovery call: right size, real pain, budget range named | Fit call: problem in your wheelhouse, budget realistic |
| Evaluation | Active trial; success criteria = team invited + core feature adopted | Scoping sessions; draft statement of work reviewed |
| Proposal | Order form sent; procurement/security review running | Proposal + SOW delivered; terms in negotiation |
| Closed | Subscription started (payment event) | Contract signed; kickoff scheduled |
If you run product-led motion alongside sales, resist the urge to bolt extra stages on for it — a trial signup is a Lead or Qualified entry event, and product-qualified behavior is an evaluation exit signal. Use lead scoring to rank who inside those stages deserves attention today; that is scoring’s job, not the pipeline’s.
The common mistakes that break pipelines
- Too many stages. Nine stages feels precise and measures nothing — reps cannot place deals consistently, so stage data becomes noise with extra admin. Four to six is the working range; add a stage only when two deals in the same stage routinely need different next actions.
- Activity-based stages. “Emailed”, “Called twice”, “Demo booked” describe the seller. A deal can absorb unlimited seller activity while the buyer stands perfectly still — and the pipeline should say so.
- No exit criteria. Without a written, buyer-verifiable condition per stage, deals advance on rep optimism, and your stage-weighted forecast inherits that optimism at scale.
- Stages as forecast categories. “Commit” and “Best case” are forecast judgments layered on top of stages, not stages themselves. Mixing them forces reps to choose between describing the deal and describing their confidence.
- The zombie middle. No time-in-stage limit means stage 3 silently accumulates dead deals. Set a staleness threshold per stage (say, 30 days without buyer-side movement) and force a decision: advance, recycle, or close-lost.
How this maps to a Lead → Qualified → Customer pipeline
You may notice modern CRMs — ours included — ship fewer columns than the five-stage model above. That is deliberate compression, not disagreement. BusinessMCP’s default board is Lead → Qualified → Customer → Churned, with a Prospects column upstream for contacts sourced by the AI SDR before they respond. The mapping:
Stages 3–4 of the classic model collapse into Qualified: for most SMB deal volumes, deal value + next-step notes carry that nuance better than extra columns reps must maintain.
The important design choice is which transitions are event-driven. A contact identifying on your site enters as a Lead automatically — company identification means that happens before any form fill for a meaningful share of B2B traffic. A payment advances a contact to Customer with no human involved. The only transitions left to judgment are the ones that genuinely require it: Lead → Qualified, and the decision to call a deal dead.
That split — machines advance what events can prove, humans advance what conversations reveal — is also your best defense against stage drift, the quiet rot we cover in the companion guide on CRM data hygiene. A stage that updates itself is a stage that is never two quarters stale.
Frequently asked questions
How many stages should a sales pipeline have?
Four to six for almost every B2B team. Fewer than four and you lose forecasting resolution; more than six and reps can no longer place deals consistently, which turns stage data into noise. Add a stage only when deals in an existing stage routinely require different next actions — that is the only signal that a stage is genuinely missing.
What is the difference between a sales pipeline and a sales funnel?
Same underlying data, two views. The pipeline is deal-level: which opportunities sit in which stage, worth how much, owned by whom. The funnel is aggregate: what fraction converts from each stage to the next over time. You manage deals in the pipeline and diagnose the process in the funnel.
What are entry and exit criteria in pipeline stages?
Written, checkable conditions for when a deal enters and leaves each stage — ideally verifiable from the buyer’s side, like “buyer confirmed budget range and decision process,” not “rep is confident.” They are what make stage data mean the same thing across reps, which is the precondition for trusting any conversion or forecast number.
Where do MQL and SQL fit into pipeline stages?
MQL/SQL are qualification labels from the marketing handoff world, and they map onto the front of the pipeline: an MQL is roughly a Lead (engaged, fit unconfirmed) and an SQL is roughly Qualified (fit and need confirmed by sales). Keeping them as lifecycle labels rather than extra pipeline columns avoids double bookkeeping.
Should trials and demos be their own pipeline stage?
A demo or trial is usually the content of the Evaluation stage, not a stage itself. Track it as the current next step, and treat completion of an agreed evaluation — success criteria met, decision-makers present — as the exit criterion. Making every activity a stage is the most common way pipelines bloat past usefulness.
Sources
BusinessMCP Team
Every guide is written from running BusinessMCP on its own platform — the match rates, reply rates, and deliverability lessons are from our own data, not recycled blog folklore. About BusinessMCP
Turn your business into one AI-ready MCP server
Connect your tools, install one tracking script, and expose your unified data to any AI agent through a single secure endpoint.
Get started free