How to Build a Sales Pipeline: Stages, Exit Criteria and Next Steps

Most sales pipelines are built the way most filing cabinets are: someone sat down, invented some labels that sounded sensible, and then spent two years working around them. The stages read “Contacted”, “Follow-up”, “Negotiation”, “Hot”. Half the deals sit in one column, nobody can say why a deal is in the column it is in, and the forecast is a mood rather than a number.

A pipeline is not a to-do list and it is not a record of how busy you have been. It is a map of where each prospective customer is on their way to a decision. Get that one idea right and the rest — how many stages, what to call them, when a deal may move — mostly follows. This guide walks through building a pipeline that a small team can keep honest, with exit criteria for every stage and a weekly routine that keeps it clean.

Key takeaways

How many stages, and what to call them

For most small and mid-sized businesses the right number is five to seven, plus the two closing outcomes of won and lost. The reasoning is practical rather than mathematical.

A useful test: could two people on your team, looking at the same deal, independently pick the same stage? If not, either the stages overlap or the exit criteria are missing.

Name stages by buyer outcome

Read each stage name as a sentence beginning “The buyer has…” or “The buyer is…”. If it fits, the name describes the buyer. If you find yourself writing “We have…”, it is describing you.

A deal in “Reviewing a written proposal” tells you what to do next; a deal in “Proposal sent” only tells you a document left the building.

Exit criteria: what must be true to move forward

The single most useful thing you can add to a pipeline is a written exit criterion for each stage. It is the answer to the question “what has to be true before this deal is allowed into the next column?”

Good exit criteria have three properties:

• They are observable. Something happened that you could point to: an email reply, a meeting held, a document received. Not “they seemed keen”.
• They involve the buyer. Something the buyer said, agreed to or did. Your own actions alone rarely prove the deal has advanced.
• They are few. One to three conditions per stage. A checklist of ten items will be ignored.

Here is a worked example for a service business, say a small agency, an installer or a consultancy selling projects worth a few thousand to a few tens of thousands. Adapt the wording to your own work; the structure is what matters.

A lead cannot reach “Solution agreed” just because you ran a demo; the buyer has to say it fits.

Stages versus tasks: where “follow up” belongs

The most common design mistake is turning tasks into stages. It happens because tasks are concrete and stages are abstract. “Send proposal” feels like a milestone, so it becomes a column.

The distinction is simple. A stage is a condition that stays true for a while: the buyer is evaluating, the buyer is deciding. A task is an action with an owner and a due date that is finished and then gone: send the proposal, call on Thursday, chase the reference.

“Follow-up” is the worst offender, because it is not a position in the journey at all. A deal in a “Follow-up” column has simply been put somewhere to be forgotten. If the buyer is not responding, the stage they were last in is still the truthful answer; the chase is a dated task attached to the deal.

Stage probabilities: useful, but only as estimates

Most CRMs let you attach a win probability to each stage, and the pipeline total then becomes a weighted forecast: a deal worth 10,000 in a stage marked 40 percent counts as 4,000. It is a handy device, and it is easy to misuse.

The numbers you type in on day one are guesses. A stage called “Proposal out” does not have a 50 percent chance of winning because the template said 50; it has whatever chance your business historically converts at that point. Those figures differ enormously between businesses, between lead sources and between a referral and a cold enquiry.

Start with a gentle ladder that rises as the buyer commits, and never let a probability fall as a deal moves forward. After a few months, calibrate: for each stage, count the closed deals that ever reached it and how many of those were won. That ratio is your real figure. With few closed deals the numbers are noisy, so adjust gently and revisit each quarter.

Treat the weighted total as a rough planning figure with a wide margin. It is good for asking “is next month looking thin?” and poor for promising a number to a bank.

The deal fields that actually matter

Every field you add to a deal is a field somebody has to fill in. Most small teams need only a few, and they should be mandatory in spirit if not in software.

• Value. What the deal is worth if it closes. Use a consistent basis: first-year revenue, total project price or monthly recurring amount, but pick one and stick to it, or totals become meaningless.
• Expected close date. When you realistically expect a decision. This should be a date the buyer has helped you set, not one that makes the quarter look good.
• Owner. One named person. Deals owned by “the team” are owned by nobody.
• Next step. The single next action, with a date. This is the most important field of all and gets its own section below.

Stale deals: the discipline that keeps a pipeline honest

Pipelines rot in a predictable way. Deals go quiet, nobody wants to admit the buyer has vanished, and the card sits in the same column for months. The total looks healthy; the reality is that a third of it is fiction. This is the single biggest reason forecasts are wrong.

The cure is one rule, applied without exceptions: every open deal has a next step with a date.

• A next step is a specific action, not an intention. “Call Dana Thursday to ask about the board meeting” is a next step. “Keep in touch” is not.
• The date is a real date, not “soon”. When the date passes without the step happening, the deal is flagged as needing attention.
• If you cannot name a next step, the deal is not alive. Either create one by contacting the buyer, or close the deal as lost.

Closing a dead deal is not a failure: it cleans your forecast and, with a reason recorded, teaches you something.

One pipeline or several?

The temptation is to create a separate pipeline for every product, region or team. Resist it. Every extra pipeline fragments your reporting, multiplies the stage definitions you must maintain and makes it harder to see the whole business at once.

Create a separate pipeline only when the sales process itself is truly different — different stages, different buyers, different exit criteria. Good reasons look like this:

• New business versus renewals or upsells. An existing customer does not need qualifying from scratch, and the steps are different.
• Very different deal types. A two-week fixed-price job and a six-month enterprise engagement do not share a sensible set of stages.
• Different currencies or markets, where you want the totals kept apart.

Poor reasons include a different product, salesperson or lead source; use a field or the owner instead.

Won and lost: closing the loop properly

Won and lost are not decorations. They are the two outcomes every deal must eventually reach, and treating them as proper closing stages gives you the data to improve everything else.

Record the actual close date and value

When a deal is won, store the date it closed and the final value, which may differ from what you first expected. Comparing expected with actual close dates shows how optimistic your estimates are.

Always record a lost reason

This is the habit that separates teams who learn from teams who repeat themselves. A short, consistent list of reasons is better than free text nobody can count. Typical entries:

• Chose a competitor
• No budget or budget cut
• Decided to do it in-house
• No decision; project shelved
• Price
• Poor fit; we should not have pursued it
• No response after agreed follow-ups

Review the counts every quarter. If poor fit leads, qualification is too loose; if a competitor leads, find out what they offered.

Moving backwards is fine

Deals do not always travel in a straight line. A buyer in “Terms and timing” may bring in a new decision maker who has not seen the solution, which means the honest stage is earlier. Move the deal back. Teams often resist this because it looks like a setback, but a deal in the wrong stage corrupts your forecast and your conversion numbers.

The 30-minute weekly pipeline review

A pipeline stays accurate only if somebody looks at it regularly. A short weekly review, held at the same time each week, does more than any amount of cleaning up in a panic at quarter end. Thirty minutes is enough for a small team.

• First 10 minutes: deals that entered or changed stage, checked against exit criteria.
• Next 10: deals closing soon. What is the buyer waiting for, and what is the next step?
• Next 5: stale and overdue deals. Assign a dated next step or close them.
• Last 5: wins and losses have a date, a value and, for losses, a reason.

Two ground rules help. First, the review is about the pipeline, not about people; the goal is accuracy, not blame. Second, every deal discussed leaves the meeting with a dated next step. If the meeting ends and nothing in the system has changed, it was a conversation, not a review.

Common mistakes

• Too many stages. If people disagree about where a deal sits, merge stages.
• Stages that are really tasks. Replace verbs like send contract with the buyer state that results.
• Optimistic close dates. Base them on something the buyer told you, not on when you need the number.
• Deals that never die. A card with no movement for months and no next step should be closed as lost.
• Skipping the lost reason. It takes ten seconds and is the cheapest market research available.

How Scalry handles this

Scalry’s CRM is built around exactly this structure. You can create one or more pipelines, each with a name and a currency and one marked as the default. Each pipeline is made of ordered stages. A stage has a colour, a win probability from 0 to 100 percent, and can be marked as a won stage or a lost stage, so closing a deal in either state is part of the pipeline rather than a separate step.

A deal carries a name, the linked contact and company, its pipeline and stage, a value and currency, an expected close date, an actual close date, an owner, a status, a lost reason and room for custom data for anything particular to your business. Creating a deal records a “deal created” activity, and moving it between stages records a “stage changed” activity, so you can see when a deal moved and build the history of how long it spent in each place.

The contact’s timeline brings together notes, emails, calls, meetings, tasks, SMS, WhatsApp messages, funnel visits and social interactions, which makes it easier to check a deal against its exit criteria: did the buyer actually say yes to the proposal, or did you only send it? The exit criteria themselves, the rule that every deal has a dated next step and the weekly review are practices you bring; the software keeps the data in one place so the practices are easy to follow.

Questions to ask before you commit

Whichever way you are leaning, these are the questions that change the answer. Work through them against your own numbers rather than anyone's feature matrix — including this one.

• Could two people place the same deal in the same stage? If not, the criteria are missing or the stages overlap.
• Are any stages really tasks? Names that begin with a verb are the giveaway.
• Does every open deal have a next step with a date? Count the ones that do not.
• Where did our stage probabilities come from? If the answer is “a template”, calibrate them.
• What is the oldest open deal, and why is it still open?
• What are our top three lost reasons? If you cannot say, you are not recording them.
• How far off are our expected close dates? Compare them with actual dates on the last ten wins.

If you cannot answer the first three, you are not ready to choose between Scalry and your current pipeline yet — you are still deciding what you need, which is a cheaper problem to solve first.

A plan for the next two weeks

Days 1 to 3: list the steps your last five won deals went through, group them into five to seven buyer-based stages and write one to three exit criteria for each. Days 4 to 7: rebuild the pipeline, set rough rising probabilities, move every open deal to the stage its evidence supports and give each a dated next step or close it with a reason. Days 8 to 14: hold your first thirty-minute review and adjust any criteria that felt unclear.

Frequently asked questions

How many stages should a sales pipeline have?

Five to seven working stages, plus won and lost, is right for most small and mid-sized businesses. Fewer than that hides where deals stall; more than that makes it hard for people to agree on which stage a deal belongs in. If two people could not independently place the same deal in the same stage, merge stages.

What are exit criteria in a sales pipeline?

Exit criteria are short, checkable statements of what must be true before a deal moves to the next stage, for example that the buyer has confirmed the problem and a budget range. They should be observable, involve something the buyer said or did, and be limited to one to three conditions per stage.

What is the difference between a pipeline stage and a task?

A stage is a condition of the buyer that stays true for a while, such as reviewing a proposal. A task is an action with an owner and a due date that is completed and then gone, such as sending the proposal. If a stage name starts with a verb, it is probably a task that should live on a task list.

How should I set stage probabilities?

Start with a rising ladder of rough estimates, then replace them with your own numbers once you have closed deals: for each stage, divide the deals that were eventually won by the deals that ever reached it. Treat the weighted pipeline total as a planning estimate with a wide margin, not a promise.

When should I create a separate pipeline?

Only when the sales process is genuinely different, with different stages, buyers or exit criteria. New business versus renewals is the classic example. A different product, salesperson or lead source is not a reason; use a field or the owner instead and keep one pipeline.

What should I do with deals that have gone quiet?

Every open deal needs a next step with a date. If a deal has none, contact the buyer to create one. If the agreed follow-ups get no response, close the deal as lost with an honest reason such as no response. Dead deals left open inflate the forecast and hide real problems.

Why does recording a lost reason matter?

A lost reason turns a failure into data. Counting reasons over a quarter shows whether you lose on price, to a particular competitor, through poor-fit leads or because buyers never decide, and each of those points to a different fix. Use a short, consistent list so the answers can be counted.