Sales Pipeline Stages for Agencies: Exit Criteria and Rot

A stage should describe the buyer’s state, not your activity, and every stage needs an exit criterion someone else could check.

Designing sales pipeline stages for agencies that survive contact with a real quarter

Most agency pipelines are built by listing the things the agency does — call, propose, follow up, close — and turning each into a stage. It looks tidy and it produces a board that cannot answer the only question a pipeline exists to answer: which of these deals is actually going to happen, and what is standing in the way.

The fix is a change in perspective rather than in tooling. A stage should describe the buyer’s state, not your activity. “Proposal sent” tells you what you did. “Proposal under review with the decision maker” tells you where the deal is. The first is a to-do list wearing a board; the second is a forecast.

Every stage needs an exit criterion someone else could check

This is the discipline that makes everything else work. For each stage, write one sentence describing the observable fact that must be true before a deal leaves it. Not a feeling, not “seems keen” — something a colleague could verify from the record.

• Qualified — the need, the timeframe and who decides are all recorded on the deal.
• Scoped — the work has been described specifically enough to price, and the client has seen that description.
• Proposal with the decision maker — the person who can say yes has confirmed receipt.
• Verbal agreement — they have said yes and named what still has to happen before signature.

Write those criteria down next to the pipeline, not in someone’s head. The test of a good set is that two people looking at the same deal put it in the same stage. If they do not, the criteria are opinions and the board’s numbers are decoration.

How many stages

Fewer than feels natural. Each stage has to earn its existence by being a point where deals genuinely and visibly get stuck, because the reason to separate two states is that they have different failure modes and different next actions. Five or six is comfortable for most agency sales; beyond that, stages start splitting on activity rather than state and the board becomes admin.

A useful pruning test: if no deal has ever sat in a stage for more than a day, it is not a stage, it is a task. Fold it into the neighbouring one and put the task on the deal instead.

What belongs on the deal rather than in a stage

The lost reason deserves emphasis. A pipeline that records only that a deal was lost teaches you nothing; one that records whether it was price, timing, scope, a competitor or no decision at all will, within a quarter, tell you which part of the offer to change.

Probability weighting is a report, not a stage property

Attaching a fixed percentage to each stage is convenient and quietly misleading, because it assumes every deal in a stage is equally likely, which is exactly what an experienced seller knows to be false. Two deals sitting in the same stage, one moving weekly and one untouched for a month, are not the same deal.

If you use weighting, treat it as a rough planning aid rather than a forecast, and pair it with age. Time in stage is a more honest signal than a percentage, because it is measured rather than assumed.

Rot: the failure mode nobody notices

The most common pipeline problem is not deals moving backwards. It is deals not moving at all, while everyone looks at a board that still shows them as live. A pipeline full of stale opportunities inflates the forecast, hides the real volume of active work, and makes the whole thing feel too big to review.

Three things fix it, and none of them require discipline from the person who is already avoiding the deal:

• Define stale per stage. An early conversation may go quiet for weeks legitimately; a deal at verbal agreement should not.
• Automate the flag rather than the nag. A stale opportunity trigger can surface deals that have not moved within their stage’s window, which turns a review meeting into a short list rather than a scroll.
• Make closing a deal as lost a normal act. If losing feels like an admission, nobody does it, and the board fills with fiction. A required lost reason with no blame attached is what makes it routine.

One pipeline is rarely enough — and four is too many

An agency runs at least two different processes with different states, and forcing them into one board is why so many pipelines look strange. New business moves through interest, scope, proposal and close. Delivery moves through kickoff, build, review and launch. Renewal and expansion move through usage, review and decision. Those are three sets of states with almost nothing in common.

Separate them, and keep each one short. The connection between them matters more than the boards themselves: a deal that closes should create the delivery record, and a delivery that ends should feed the renewal view. Retention is where most agency revenue actually comes from, which is why the renewal pipeline deserves the same design attention as the new-business one — the mechanics are covered in reducing churn at an agency.

Where automation should touch the pipeline, and where it should not

Automation is good at the mechanical consequences of a stage change: creating the tasks the new stage requires, notifying the person who now owns it, starting or stopping a follow-up sequence, and recording the transition with a timestamp. Which sequence a deal belongs in, and which it should be suppressed from, is the routing and suppression question, and it should read the stage rather than duplicate the logic.

Automation is bad at moving deals forward. A stage change is an assertion that something is true about the buyer, and a rule that advances a deal because an email was opened is manufacturing that assertion. The one honest exception is moving deals into a pipeline: an inbound lead that has answered qualifying questions, for instance through a scored quiz, can be created directly at the qualified stage, because the criterion has genuinely been met.

Reviewing the pipeline without it taking an hour

• Look at stale deals first, not the biggest ones. The board’s honesty depends on it.
• For each stale deal, take one of three decisions: a next action with a date, a stage change, or closed-lost with a reason. Nothing else counts as a decision.
• Check that every open deal has a next action with a date on it. A deal without one is not being worked.
• Read the lost reasons from the period since the last review as a set, rather than one at a time. The pattern is the point.

The mistakes that show up in almost every audit

• Stages named after seller activity. The board reports effort rather than progress.
• No written exit criteria. Two people stage the same deal differently, so the numbers mean nothing.
• Too many stages. Updating the board becomes the work.
• No lost reason. The most useful data in the whole system, discarded at the moment it is available.
• Nothing watching for age. The forecast quietly fills with deals that ended weeks ago.
• Delivery living in the sales pipeline. Two processes with different states sharing one board, so neither reads correctly.

Choosing the CRM that holds the pipeline: GoHighLevel vs Salesforce compares a per-user sales CRM with a flat-priced all-in-one, using each vendor’s published prices.

Frequently Asked Questions

How many stages should an agency sales pipeline have?

Usually five or six. Each stage has to earn its place by being a point where deals genuinely get stuck, because the reason to separate two states is that they have different failure modes and different next actions. A useful pruning test is whether any deal has ever sat in the stage for more than a day. If not, it is a task rather than a stage, and it should be folded into its neighbour.

Should stages be named after what the agency does?

No. A stage should describe the buyer’s state, not your activity. ‘Proposal sent’ records what you did; ‘proposal under review with the decision maker’ records where the deal actually is. The first turns the board into a to-do list, the second makes it a forecast. This single change fixes more pipelines than any amount of automation.

What is an exit criterion and why does each stage need one?

It is one sentence describing the observable fact that must be true before a deal leaves the stage, written so that a colleague could verify it from the record rather than from a conversation. The test of a good set is that two people looking at the same deal put it in the same stage. If they do not, the criteria are opinions and every number the board produces is decoration.

Should each stage have a win probability percentage?

Treat weighting as a rough planning aid rather than a forecast. A fixed percentage per stage assumes every deal in that stage is equally likely, which is exactly what an experienced seller knows to be false: two deals in the same stage, one moving weekly and one untouched for a month, are not the same deal. Pair any weighting with time in stage, which is measured rather than assumed.

What should I do about deals that never move?

Define what stale means per stage, since an early conversation can legitimately go quiet for weeks while a deal at verbal agreement cannot. Automate the flag so review starts from a short list rather than a scroll. And make closing a deal as lost a normal act with a required reason and no blame attached, because if losing feels like an admission nobody does it and the board fills with fiction.

Should sales and delivery share one pipeline?

No. They are different processes with different states. New business moves through interest, scope, proposal and close; delivery moves through kickoff, build, review and launch; renewal moves through usage, review and decision. Forcing them onto one board means none of them reads correctly. Keep them separate and concentrate on the handoffs, so a closed deal creates the delivery record and a finished delivery feeds the renewal view.

Should automation move deals between stages?

Generally not. A stage change asserts something is true about the buyer, and a rule that advances a deal because an email was opened is manufacturing that assertion. Automation should handle the consequences of a stage change instead: creating tasks, notifying the new owner, starting or stopping sequences, recording the transition. The honest exception is creating a deal directly at a qualified stage when the qualifying answers have genuinely been captured.

We build your GoHighLevel account for you

CRM, pipelines, funnels, SMS and automation — configured for your business, not left for you to figure out.

• Setup (one-time)$850
• Ongoing (per month)$38.80

Get my free setup quote

No obligation quote · reply within 1 business day

This site is independently operated and is not officially affiliated with, endorsed by, or sponsored by HighLevel LLC. We do not receive referral commissions — this is our own flat-fee implementation service.