Sales Pipeline Stages: A Five-Day Clean-Up With Real Exit Criteria

Open your CRM and sort by last activity. The deals at the bottom of that list are still counted in your coverage, still shaped like pipeline, and still feeding a forecast. Five focused days can strip them out and leave a number you can plan against.

The routine is for a sales leader or an AE who owns a book of open opportunities. Allow ninety minutes a day. You need a spreadsheet export and a willingness to change some stage labels, not a new tool.

The Week Laid Out

Day	Job	Question it answers
Day 1	Pull the raw list	How big is the mess?
Day 2	Define each stage by buyer evidence	What must be true to move on?
Day 3	Score every open deal	Which deals fail their stage?
Day 4	Decide each failing deal's fate	Is it early, quiet or dead?
Day 5	Rebuild coverage	What do we really hold?

Day 1: Pull the Raw List and Resist the Urge to Fix

Export every open opportunity with owner, stage, amount, created date, last logged activity and next step. Add two calculated columns: days in current stage and days since last activity. Sort by the second.

Touch nothing today. A first look at this data is a measurement, and editing it changes the answer. It will not say which deals are dead. It tells you how long the week will be.

Day 2: Write the Test Each Stage Must Pass

Close's guide to sales pipeline stages names seven, from lead generation through retention, but sets no explicit entry or exit conditions for any of them. That gap is where optimistic stage creep lives, so fill it yourself.

For every stage, write one sentence of evidence that a stranger could verify from the record. Use this layout: stage, proof of buyer action, where the proof lives, and the maximum age before you question it.

• Discovery to evaluation: the buyer stated a cost of the problem and a date by which it matters. Proof lives in call notes. Review after 30 days.
• Evaluation to decision: someone other than your first contact has seen the solution. Proof lives in a meeting invite or a reply.
• Decision to signature: procurement or legal steps carry names and dates in a shared plan.

Write every line as something the buyer did. "Proposal sent" describes you. "Buyer shared the proposal with finance" describes them. Writing the lines will not reveal which deals fail. It makes the test fair before you apply it.

Set the maximum age from your own history, not from a blog. Take the deals you won last year, find how long each spent in every stage, and use the slowest healthy cases as the ceiling. A deal that has sat far beyond that ceiling is not necessarily lost, but it owes the team an explanation. Ten minutes in a spreadsheet gives you these ceilings, and they make Day 3 far less arguable, because the rule comes from your own wins and not from an outsider's opinion.

Day 3: Score Every Deal Against Its Own Stage

Add an evidence column and mark each opportunity pass, fail or unknown. A rep who disagrees should add the missing proof to the record instead of arguing about it in a meeting.

Do not set a quota of failures. The count is a finding. If almost everything fails, your criteria may be too tight for your cycle, so run them on three deals you know well before blaming the pipeline. The scoring marks deals but never explains them; that needs a conversation with the rep.

Day 4: Sort Failing Deals Into Three Piles

1. Wrong stage. The deal is real but early. Move it back and let the forecast absorb the change.
2. Quiet. The buyer was engaged once and then went silent. Send one specific message and put a seven-day check in the calendar.
3. Dead. No contact, no next step, no reason to act. Close it as lost with an honest reason code.

Never delete a record. Closed-lost with a reason is data you can use next quarter, and the six pipeline clean-up practices cover how to keep the list tidy afterwards. Antoine Buteau, writing on sales management in May 2026, proposes a blunt dormancy test: no conversation with the buying committee for 14 days means the deal is flagged, not counted. That is one practitioner's rule, so tune the number to your own cycle length.

Day 5: Recompute Coverage on the Survivors

Salesforce argues the familiar 3x rule works only when win rate is exactly 33% and the cycle is exactly one year, and offers a formula: quota, divided by win rate, divided by the number of sales cycles that fit in a year. Run it with your own win rate and cycle length, then compare the answer with the surviving list.

Take a rep with a quarterly quota of 150,000 and a win rate of 25%. The pipeline needed is 600,000. If the clean list holds 450,000, the gap is 150,000, and that gap is the prospecting target for the next weeks. The full method is in the 3x coverage piece.

Finish with one line for your team: what is real, what the target is, what is missing. The exercise will not tell you what closes. It gives you an honest starting point.

Let Buyer Activity Overrule the Rep's Stage

Most teams police stages by asking reps whether a deal meets the criteria. That fails for a simple reason: the person who benefits from a larger pipeline is the person scoring it. It also pushes the same conversation into every weekly review.

Prefer evidence the CRM captures without the rep typing it: email replies, meeting attendance, shared-document opens. Gong's 2022 survey of 928 revenue professionals found that only 24% of sales leaders trusted the forecast commitment from their reps. The same release says organisations using insight from customer interactions were about twice as likely to land within 5% of their monthly forecast: 22%, against 11% for those relying on seller-entered data. That is a vendor survey from a conversation-recording company, and the absolute numbers are low either way, so read it as a direction.

Without a way to capture buyer activity automatically, for instance when buyers work on channels you cannot log, the call does not hold. Then keep a required evidence field with a link to the proof, and spot-check five deals a week yourself. A stage without proof is a guess in a nicer font.

Moves That Waste the Week

• Buying software first. A tool presents confused data in a cleaner chart.
• Cleaning alone. Every rep owns their list and agrees their own failures.
• Counting "nurture" as pipeline. A holding stage has no buyer milestone, so keep it outside the coverage number.

Where the Routine Breaks

• Reps fear a smaller number. Present it as protecting the forecast. Penalising honest closures ends the practice immediately.
• Criteria too heavy for the cycle. If nearly all deals fail, loosen the maximum ages first.
• The mess returns. Give each rep a ten-minute weekly pass, and read why pipeline reviews measure the wrong week. The CRM hygiene checklist covers the daily habits.

The Next Step and How to Judge It

Run the routine once, record the before and after counts, and share both with the team so they see clarity, not blame. Judge success one quarter later: if forecast calls move closer to outcomes, the stages are doing their job. The routine promises no better quarter, only a truer picture of the one you are in.

FAQs about sales pipeline stages

How many stages should a sales pipeline have?

Close's guide lists seven. Your count should match how buyers actually decide, with each stage tied to evidence the buyer provides.

What does a stage exit criterion look like?

It is a verifiable fact about the buyer, such as a named signer, a stated deadline or a dated mutual plan, that must exist before a deal moves on.

How often should I audit the pipeline?

Do a full audit each quarter and a brief one weekly. The five-day routine above is built for the first full pass.

Should dead deals be deleted?

No. Close them as lost with a reason, or park them in nurture outside the coverage count, so history stays and the forecast is clean.