Your pipeline is real when every open deal has an amount, a close date still in the future, a dated next step, recent activity, and a stage it reached by meeting written exit criteria. Forecasts miss when they rest on default stage probabilities and gut feel instead of your own closed history. Fix the pipeline data first. The forecast follows.
What this covers
- How do I know if my sales pipeline is real?
- What should our pipeline stages be?
- How much pipeline do we need?
- How do I run a pipeline review that is useful?
- Why do deals stall, and what do we do about them?
- Why is our sales forecast always wrong?
- How do I build a forecast I can defend?
- What are the red flags in a forecast?
- Who owns pipeline and forecasting?
How do I know if my sales pipeline is real?
Run the pipeline reality check. Pull every open opportunity and test each one against the eight checks below. Most are filters you can build in HubSpot or Salesforce yourself; the two that look at history, pushed close dates and days in stage, need field history tracking switched on. A deal that fails one check is a question for its owner. A deal that fails three is not pipeline, whatever stage it sits in.
- Reported $1.55M
- Passes every check $1.55M
- A question for the owner $0K
- Not pipeline $0K
- Northwind Labs $120K
- Fernhill Systems $250KClose date in the pastNo activity in 30 daysNo next stepSingle-threaded
- Rivet Analytics $80K
- Quarry Health $150KClose date pushedStuck in stageQuarter-end close date
- Lumen Freight $100KPlaceholder amountSingle-threaded
- Ostrava Bank $300KClose date in the pastNo activity in 30 daysNo next step
- Pinecrest Retail $60K
- Halden Logistics $180KQuarter-end close date
- Marlow Software $90K
- Tessel AI $220KStuck in stageClose date pushedNo next step
Run every check and $350K of the $1.55M reported passes all of them. $920K fails three or more and is not pipeline.
An illustration with made-up deals, not a client's.
The eight checks, and what a failure usually means:
| Check | The filter | What a failure usually means |
|---|---|---|
| Close date in the past | Open, close date before today | Nobody is managing the deal. It is lost or forgotten |
| No recent activity | No logged email, call or meeting in 30 days | The buyer has gone quiet, or the rep is not logging |
| No next step | Next step empty, or with no date | The deal has no plan, so it has no momentum |
| No real amount | Amount empty, or a round placeholder repeated across deals | The deal was never scoped with the buyer |
| Stuck in stage | Days in current stage above twice the median for won deals | Stalled. See stalled deals |
| Pushed close date | Close date moved two or more times | Optimism, not a buying process |
| Single-threaded late | Proposal stage or later with one contact on the deal | One person leaving or going quiet kills it |
| Quarter-end cluster | Close dates bunched in the last week of the quarter | Close dates set by habit, not by the buyer |
Add up the value of the deals that pass every check. That is your clean pipeline. Compare it to the number on the dashboard. The gap between the two is the most useful figure in this guide, because it tells you how much of your reported pipeline is hope.
Two cautions. First, the "no recent activity" check depends on activity being logged automatically. If reps log by hand, a quiet deal and a lazy rep look identical. Second, "twice the median" only works once you have enough won deals for the median to mean something. With fewer than about thirty, use your own judgement and say so.
If you are the founder: do this yourself once, before you ask anyone else to. It takes an hour, and you will learn more about your sales process than from any dashboard.
What should our pipeline stages be?
Stages should describe what the buyer has done, not what the seller has done, and each one should end with an exit criterion someone else can check. "Interested" is a feeling. "Confirmed budget and named the person who signs" is a stage. Five to seven stages is enough for most B2B sales processes.
Here is a starting point to adapt, not copy. The right stages depend on how your buyers buy. Step through each stage to see what moves a deal on, what the CRM should require, and what is not enough on its own.
- Moves on when
- The buyer has described a problem in their own words and agreed to a second conversation
- Required to move
- Problem, source, primary contact
- Not enough on its own
- A friendly first call
- Moves on when
- Problem, timing and budget range confirmed; the person who signs is named
- Required to move
- Amount estimate, close date, economic buyer
- Not enough on its own
- "They have budget," with no number
- Moves on when
- The buyer has seen how you solve their problem and agreed on what a proposal should contain
- Required to move
- Decision process, competitors, a second contact
- Not enough on its own
- A demo that went well
- Moves on when
- A proposal is sent and the buyer has confirmed the steps and dates to a decision
- Required to move
- Proposal date, mutual plan dates
- Not enough on its own
- A proposal sent with no reply
- Moves on when
- Verbal yes; legal, security or procurement steps are under way
- Required to move
- Contract sent date, signatory
- Not enough on its own
- "Looks good" from your champion
- Moves on when
- Signed, or the buyer has said no or stopped responding
- Required to move
- Loss reason from a picklist
- Not enough on its own
- Leaving it open just in case
Three rules make stages work in practice:
- Make the fields required, not requested. Use stage validation in the CRM so a deal cannot move to Qualified without an economic buyer. Asking nicely does not survive the end of a quarter.
- Mark losses, every time, with a reason. A deal that is neither won nor lost nor moving inflates your pipeline and your win rate at the same time.
- Delete the default probabilities. CRMs ship with stage probabilities that describe nobody's business. Replace them with your own historical stage-to-close rates once you have the data, and leave them out of forecasts until then.
How much pipeline do we need?
Enough that your own history says you will hit the number. The common "three times quota" rule is a guess that happens to suit some teams. Work out yours from one ratio: of the pipeline open at the start of a past quarter that was due to close in it, what share of the value closed won in that quarter?
Coverage needed = 1 ÷ in-quarter conversion rate
An illustration with made-up numbers: if, over the last four quarters, 25% of start-of-quarter pipeline value closed won within the quarter, you need 4 times the remaining target in pipeline. If it was 40%, you need 2.5 times. Measure it by segment, because a team selling small deals and a team selling large ones will have very different ratios.
You are $600K short of the pipeline your history says you need. The 3x rule would tell you that you are fine with $1.8M, which is $600K short of what your history says.
Your in-quarter rate is the share of pipeline value open at the start of past quarters, and due in them, that closed won within the quarter.
Coverage only means something if the pipeline is clean. Four times a pipeline full of deals that fail the reality check is not four times anything. Run the check first, then measure coverage on what survives.
Want a second pair of eyes on your pipeline?
Take the free RevOps Health Check for a scored read on your data, pipeline and forecast, or book a free 30-minute review and we will run the reality check with you.
Take the RevOps Health Check Book a free 30-minute reviewHow do I run a pipeline review that is useful?
Review by exception, not deal by deal. A weekly pipeline review should spend its time on what changed and what looks wrong, not on reps reading their deals aloud. Keep it separate from the forecast call: the review is about the health of deals, the forecast call is about the number.
1. What entered and what left (5 minutes)
New deals created, deals that reached Qualified, and deals closed won or lost since last week. Ask about every loss reason that says "other".
2. What moved the wrong way (10 minutes)
Deals that moved back a stage, had their close date pushed, or had their amount cut. Each of these is the buyer telling you something.
3. What fails the reality check (10 to 15 minutes)
The deals from the check above, sorted by value. For each, one question: what happens next, and when?
4. Agree actions, by name and date (5 minutes)
Every deal discussed leaves with an owner, a next step and a date in the CRM, or it is closed lost.
Three questions do most of the work in a deal conversation: who signs, what happens if they do nothing, and what has the buyer done, not said, since last week. A rep who cannot answer the second question does not yet have a deal with a reason to close.
Why do deals stall, and what do we do about them?
Deals stall for a short list of reasons, and almost all of them leave a trace in the CRM. Find the trace, then fix the cause, not the stage.
- Economic buyerEmpty, 64 days after Qualified
- Close datePushed 3 times, now the last day of the quarter
- ContactsOne, the same since the first call
- Last buyer action"Send the deck to my boss", 41 days ago
Three times the median in Solution, with no economic buyer: this deal needs a dated plan agreed with the buyer at the next review, or it is closed lost.
An illustration with made-up deal, not a client's.
| Cause | Signal in the CRM | What to do |
|---|---|---|
| No economic buyer | Economic buyer field empty past Qualified | Ask your contact to introduce the person who signs, with a reason for them to care |
| No reason to act now | No compelling event recorded; close date pushed | Agree the cost of doing nothing with the buyer, or move the deal out of this quarter |
| Single-threaded | One contact on a late-stage deal | Add the people who use, approve and pay before sending a proposal |
| Proposal before process | Proposal sent, no decision steps or dates recorded | Agree the decision process and dates, then revise the proposal |
| Champion left or changed role | Primary contact bounced or job title changed | Rebuild the relationship from another contact; treat it as a new deal |
Set a close-or-revive rule: any deal that stays past twice the median time in its stage gets a decision at the next review. Either it gets a dated plan agreed with the buyer, or it is closed lost with a reason. Closed lost is not failure. It is accurate.
Why is our sales forecast always wrong?
Usually for one or more of seven reasons, and most of them are data problems, not forecasting problems:
- Default stage probabilities. Weighting pipeline by 10, 25, 50 and 75 percent describes a generic company, not yours.
- Close dates set by habit. End-of-quarter dates entered because the field is required, not because the buyer agreed them.
- Unmarked losses. Dead deals left open keep counting.
- Optimism and sandbagging. Some reps call everything, some call nothing, and the forecast averages two kinds of error.
- A few large deals. When three deals make up half the quarter, a weighted total hides a binary outcome. Name those deals individually.
- New and renewal business mixed. They close at different rates and need separate forecasts.
- No record of past forecasts. If nobody writes down what was forecast, nobody can measure how wrong it was, and nothing improves.
How do I build a forecast I can defend?
Use at least two independent methods, reconcile the difference, and keep a record of every forecast you make. A board, a lender or an acquirer does not expect you to be right. They expect you to know how wrong you usually are and why.
| Method | How it works | Good for | Weak when |
|---|---|---|---|
| Rep commit | Each rep calls deals as commit, best case or pipeline; managers adjust | Knowledge the CRM does not have | Reps are new, or incentives reward sandbagging |
| Weighted pipeline, your own rates | Open pipeline weighted by your historical stage-to-close rates, by segment | A steady, data-driven baseline | Stages are not followed, or history is short |
| Historical conversion | Start-of-quarter pipeline times your in-quarter conversion rate, plus deals typically created and closed in-quarter | Checking the other two | The business changed: new segment, pricing or team |
Then measure accuracy, every quarter, at fixed points:
Forecast accuracy = 1 − |actual − forecast| ÷ actual
Record the forecast at week 1, week 4 and week 8 of each quarter. Accuracy at week 1 tells you about the pipeline. Accuracy at week 8 tells you about deal management. Over four quarters, the history itself becomes the strongest part of your board pack.
In this example the week 1 forecast was 40% too high in Q1. By Q4, after three quarters of cleaning the pipeline, it was within 9%. Without the recorded forecasts, nobody could show that.
An illustration with made-up numbers, not a client's.
If you are the CFO: forecast bookings, not revenue, from the CRM, then convert to revenue with your own recognition rules. Mixing the two is the most common reason the sales forecast and the finance plan disagree. The definitions are in our GTM metrics guide.
What are the red flags in a forecast?
Whether you are an investor in diligence or a board member reading a deck, these patterns mean the forecast needs questions before it gets trust:
- No record of past forecasts, so accuracy cannot be shown
- A large share of the quarter's pipeline created in the last 30 days of the quarter
- Close dates pushed repeatedly on the deals that make up the commit
- Late-stage deals with no logged activity in weeks
- A win rate well above what is normal for the deal size, which usually means losses are not being marked
- Close dates bunched in the last week of every quarter
- Pipeline stage definitions changed in the last two quarters, which breaks comparisons with the past
- Booked revenue in the CRM that does not reconcile to billing
We run these checks as part of GTM systems diligence, rated red, amber or green, before close.
Who owns pipeline and forecasting?
Three different people, and problems start when nobody owns the middle one:
- The truth of each deal belongs to the rep and their manager.
- The system, meaning stages, required fields, validation, reports and the forecast model, belongs to sales operations or revenue operations. In a small company with neither, it usually falls to the founder or head of sales, alongside a full-time job.
- The plan number belongs to the CEO and CFO.
It is bigger than a quick fix if three or more of these are true:
- More than a quarter of open pipeline value fails the reality check
- Stages have no written exit criteria, or reps skip them
- The forecast has missed by more than you can explain for two quarters running
- Nobody can say who owns the CRM setup
- Pipeline numbers are rebuilt by hand for every board meeting
That is the work behind our forecast and pipeline rebuild (from $15,000), the sales operations work that keeps stages followed, and the AI ops monitor that checks every open deal against these rules each morning.
Not sure where your pipeline stands?
Book a free 30-minute review. You leave with the three fixes worth doing first, whether or not you work with us.
Take the RevOps Health Check Book a free 30-minute reviewFrequently asked questions
How do I know if my sales pipeline is real?
Test every open deal against a short list of checks: an amount, a close date still in the future, a dated next step, activity in the last 30 days, more than one contact on late-stage deals, and a close date that has not been pushed repeatedly. Add up the value of the deals that pass every check. That is your real pipeline, and the gap to the reported number is how much of it is hope.
How much pipeline coverage do I need?
Work it out from your own history instead of a 3x rule. Find what share of the pipeline value open at the start of past quarters closed won within the quarter. Coverage needed is one divided by that rate: if 25% closes in-quarter, you need 4 times the remaining target. Measure it by segment and only on pipeline that passes basic hygiene checks.
What should B2B sales pipeline stages be?
Five to seven stages that describe what the buyer has done, each ending in an exit criterion someone else can check, such as 'budget confirmed and the person who signs is named'. A common set is Discovery, Qualified, Solution, Proposal, Commit and Closed won or lost, with required fields enforced by CRM validation and a loss reason on every lost deal.
Why is my sales forecast always wrong?
Usually because it relies on default CRM stage probabilities, close dates set by habit, dead deals that were never marked lost, and rep judgement that mixes optimism with sandbagging. A few large deals and mixing new with renewal business make it worse. Most of these are data problems, so fixing pipeline hygiene improves the forecast more than a new forecasting tool.
How do you measure forecast accuracy?
Record the forecast at fixed points in each quarter, for example weeks 1, 4 and 8, and compare it with what actually closed: forecast accuracy equals one minus the absolute difference between actual and forecast, divided by actual. Keep the history. After four quarters it shows a board how reliable the forecast is and where it goes wrong.
Keep reading: the GTM metrics and KPI guide, why your CRM data is costing you revenue, and the CRM data hygiene playbook.