Most pipeline advice is written as though you are designing a system for a hundred people. You are not. You are a small B2B team who needs something workable by Monday.
This is how to build one in an afternoon, and how to avoid the four mistakes that make a pipeline stop being useful within a quarter.
Step one: decide what a deal is
Before stages, settle what actually enters the pipeline.
A deal should be a specific opportunity with a named buyer, a rough value and a plausible reason it might close. Not “we should talk to them at some point”. Not an enquiry nobody has spoken to yet.
This matters more than any other decision here, because a pipeline full of unqualified interest produces a terrible win rate and a forecast nobody can use. Enquiries need somewhere else to live, whether that is a lead inbox, a separate stage before the pipeline, or a list.
Rule of thumb: if you would not be surprised to lose it, it belongs in the pipeline. If you would not be surprised to never hear from them again, it does not.
Step two: name your stages after the buyer
This is where most pipelines go wrong.
Stages usually get named after what the seller did. Proposal Sent. Demo Given. Following Up. Each of those describes an action your team took, which tells you nothing about whether the deal is progressing, because you can send a proposal into silence.
Name them after evidence from the buyer instead.
| Instead of | Use |
|---|---|
| Contacted | Discovery call held |
| Demo given | Requirements confirmed |
| Proposal sent | Proposal acknowledged, review booked |
| Following up | Terms under review |
| Negotiation | Verbal agreement, contract issued |
The difference is that every stage on the right requires the customer to have done something. A rep cannot advance a deal through wishful thinking, and a deal sitting in one stage for six weeks is unambiguously stuck rather than possibly progressing.
Step three: use four to six stages
Fewer than four and the pipeline does not tell you where deals die. More than six and reps stop updating it, because the difference between adjacent stages becomes a judgement call nobody wants to make twice a week.
A workable default for B2B services:
- Qualified — discovery held, need confirmed, budget plausible
- Proposal — scope agreed and quoted, proposal acknowledged
- Negotiation — terms or pricing under active discussion
- Closed Won
- Closed Lost
That is it. If you sell something with a genuine procurement or legal step, add one for it. Otherwise resist.
Step four: set probabilities from your own history
Every stage carries a win probability, and this is what turns a pipeline into a forecast. Most teams inherit whatever their CRM shipped with and never revisit it.
Do this instead. Take every deal from the last twelve months that reached each stage, and work out what proportion of them eventually closed. That percentage is your probability. It takes about fifteen minutes with an export and a spreadsheet.
If you have no history, start with something conservative and honest: 20 per cent at Qualified, 50 at Proposal, 75 at Negotiation. Then recalculate after two quarters, because your real numbers will differ and they are the only ones that matter.
Two things to watch. Probabilities that only ever go up are a sign your stages are describing seller activity rather than buyer commitment. And a stage where the real close rate is far below the assigned probability is where your forecast is leaking.
Step five: define what “won” and “lost” require
Both need a rule, and lost needs a reason.
Winning should trigger something: a client record, a handover, an onboarding step. If closing a deal just moves a card and nothing else happens, information gets lost between sales and delivery.
Losing should require a reason from a fixed list. Price, competitor, timing, no decision, unqualified, deliverables. Not a free text box, because free text cannot be counted, and the whole value of loss reasons is in the counting.
Six months of that data will tell you more about your positioning than any amount of market research.
Step six: put a next action on every deal
The last step takes five minutes and does more than the previous five combined.
Every deal in the pipeline needs a next action with a date and an owner. No exceptions. When you first apply the rule you will find deals nobody can think of a next action for, and those deals are already lost. Mark them so and your pipeline immediately gets more honest.
Maintaining the rule is what stops a pipeline decaying. A deal without a next step is not progressing, whatever stage it is sitting in.
The four mistakes to avoid
Too many stages. Every stage you add is a decision a rep has to make every week. Six is a ceiling, not a target.
Stages that describe your process, not the buyer’s. If a rep can advance a deal without the customer doing anything, the stage is measuring effort rather than progress.
Probabilities nobody checked. Inherited defaults produce a forecast that is confidently wrong. Fifteen minutes of arithmetic fixes it.
Free text loss reasons. They feel more informative and they are entirely uncountable. Use a fixed list and add to it only when the same reason appears three times.
What to do after the afternoon
Run it for a quarter without changing anything. You need one full cycle before you can tell what is working, and pipelines that get redesigned monthly never accumulate enough history to be worth analysing.
At the end of the quarter, look at three things: which stage loses the most deals, whether your probabilities matched reality, and how many deals sat with no next action. Each of those points at a specific fix, and none of them require new software.
Setting up a pipeline in PragmaCRM takes about ten minutes: define your stages, set your win probabilities, load your product catalogue so deal values calculate themselves. Try it free for 14 days.
