Sales pipeline stages: designing a B2B pipeline that tells the truth
Define B2B sales pipeline stages with exit criteria a buyer can confirm, adapt them to your sales motion, measure them honestly and run them on LinkedIn.
By Swimlane Updated September 14, 2026 6 min read
A sales pipeline stage is a claim about the buyer. “Qualified” should mean something the prospect has confirmed, not something the rep hopes. When stages describe what the seller did (“email sent”, “followed up”), the board fills up with activity and stops telling you where revenue will come from. This guide covers how to design sales pipeline stages that stay honest: definitions, exit criteria, variants by sales motion, the metrics worth tracking, and how to run the early stages on LinkedIn.
What makes a good sales pipeline stage
Five properties separate a useful stage from a label:
- It is defined by a buyer action or a checkable fact. “Meeting held”, “proposal received”, “legal review started”. Not “working on it”.
- Its exit criterion is binary. Anyone on the team can look at the evidence and agree the card belongs in the next column.
- There are few of them. Most B2B cycles read well with five to seven open stages. Beyond that, reps stop agreeing on which one applies.
- Every open card has a next action and an owner. A card without a next action is not in a stage, it is parked.
- Closed stages exist from day one. Won, and lost or abandoned, with a reason. Without them, dead deals stay open and inflate everything.
A reference set of B2B pipeline stages
Use this as a starting point and rename freely. The exit criteria matter more than the names.
- Target. Fits your ideal customer profile; no two-way exchange yet. Exit: they reply, or accept a meeting request.
- Conversation. A real exchange is under way. Exit: they have described a problem or a goal in their own words.
- Qualified. Need, their role in the decision and a rough timing are confirmed, using whatever framework you already trust (BANT, MEDDIC or your own). Exit: a discovery meeting is booked.
- Discovery done. The meeting happened; you know the requirements and how they buy. Exit: they agree to receive a proposal or a tailored demo.
- Proposal. Scope and price are in their hands. Exit: they negotiate terms or say yes.
- Commit. Verbal yes; legal or procurement is involved. Exit: signature.
- Won and Lost. Closed. Lost carries a reason you can group later: timing, budget, competitor, no decision.
Should stages carry a win probability? Only if you forecast with them, and only if you calibrate the numbers from your own closed deals. A probability copied from a template is a guess with a percent sign.
Adapting the stages to your sales motion
Short, transactional cycles
Collapse the middle: Conversation → Qualified → Proposal → Won / Lost. If a deal routinely closes in a couple of calls, a separate Discovery stage only adds clicks.
Enterprise cycles
Add a stage only where deals actually wait: security review, pilot, procurement. If a step never blocks anything, it is a checklist item, not a stage.
LinkedIn-first and founder-led selling
The first stages are conversation stages: invited, connected, conversation open, replied, meeting booked. They change daily and deserve their own board. The companion article on running LinkedIn conversations on a kanban board covers that day-to-day board; this guide is about the stage design behind it.
One pipeline per motion
Partners, renewals and new business move through different steps. Mixing them on one board makes every metric meaningless. Create a pipeline per motion.
Stage metrics worth tracking
- Stage-to-stage conversion. Compute it from your own history and look for the biggest drop. Someone else’s benchmark tells you nothing about your market, price or deal size.
- Time in stage. Take your median per stage and treat anything past twice that as stalled. Stalled cards get a decision: advance, close, or a dated next step.
- Entries at the top each week. A healthy late pipeline with an empty top is next quarter’s shortfall.
- Loss reasons by stage. Losing at Proposal on price is a different problem from losing at Qualified on timing.
- Coverage from your own win rate. If you win one proposal in four, you need four times your target sitting in Proposal. That is arithmetic, not a rule of thumb.
For the LinkedIn side of the funnel, the article on LinkedIn prospecting KPIs goes through acceptance, reply and meeting rates.
Common pipeline design mistakes
- Activity stages. “Email sent” measures effort. Replace it with the buyer event you were hoping for.
- No exit criteria. Reps push deals forward optimistically, and the late stages look full right until the quarter ends.
- A “Nurture” column that never empties. Close the card as lost or abandoned with a reason, and set a reminder for when to come back.
- Too many stages. If two stages always move together, merge them.
- No closed stages. A board where nothing ever leaves becomes a list of names.
- Probabilities nobody recalibrates. Check them against real outcomes every quarter or remove them.
A weekly pipeline review
Keep it short and mechanical:
- Open every card past its time limit and make a decision on it.
- Check that every open card has a next action with a date.
- Close what is dead. Honest closing is what keeps conversion rates meaningful.
- Once a quarter, reread the stage definitions with the team and fix the one people argue about.
Running your pipeline stages in Swimlane
Swimlane is where the conversation stages live, next to the LinkedIn inbox. Here is what its pipelines do today:
- Boards and stages. Pipelines are kanban boards. You create as many as you need, name the stages, reorder them by dragging and give them a colour.
- Closed stages are a column type. A stage is Regular, Done or Abandoned. Contacts in a Done or Abandoned column are treated as closed: they stop being suggested in the dashboard’s “People to interact with”.
- Putting a contact in a stage. Drag cards on the board, pick the stage from the contact page next to the conversation, or set it from the conversation list in the inbox. A contact holds one stage per pipeline.
- Acting on a whole column. A stage’s menu can send a connection request to every contact in it, or send one message to all of them with variables such as
{firstName}. Sends are queued and capped to what is left of your daily limits. - A suggested stage. The AI analysis on a contact reads the conversation and can suggest one of your existing stages, with a Move button to accept it. Clear stage names help it pick well.
- Cards you can read. Each card shows the photo, name, headline, company, tags and the contact’s ICP level, one of five levels from very low to excellent.
- Counts per stage. A dashboard widget shows how many contacts sit in each stage of a pipeline.
What it deliberately does not have: deal amounts, probabilities, required fields per stage or time-in-stage reports. Those belong to the deal stages in your CRM. When a conversation becomes an opportunity, hand the contact over through webhooks, the public API, n8n or Make; the webhook CRM integration guide explains what each option carries. The LinkedIn CRM page shows the boards and the contact record.
The pipeline encodes your strategy
The stages you choose are your theory of how deals progress, and the exit criteria are how you test it. Write them down, hold the team to them, and change them when your closed deals say you were wrong.