How to Build a Sales Pipeline You Can Actually Forecast

Most small-business pipelines are a list of people who once seemed interested, sorted by how optimistic the owner feels about each one. That is not a forecast. It is a mood, formatted as a table.

A pipeline becomes useful when two things are true: the stages are defined by buyer behaviour, and you know your own conversion ratios. Neither requires software.

Define stages by what the buyer did

The single most common failure is stages named after feelings — "interested," "warm," "promising." Those cannot be verified, so deals sit in them indefinitely and the forecast inherits the optimism.

Each stage should be triggered by an observable action:

| Stage | Entry condition — something the buyer did | |---|---| | 1. Contacted | You reached out. No response yet. | | 2. Engaged | They replied and agreed to a conversation. | | 3. Qualified | Call happened. Real problem, rough budget, decision path confirmed. | | 4. Proposed | Proposal sent, and a walkthrough is booked or has happened. | | 5. Closed | Won or lost, with a reason recorded. |

Notice stage 3 requires all three qualification conditions from the discovery call. A deal where you had a nice conversation but never established budget or decision path is stage 2, however encouraging it felt.

That rule alone fixes most forecasts, because the deals that inflate a pipeline are almost always the ones sitting in a stage they never earned.

Track seven columns

A spreadsheet is enough for a long time:

COMPANY   STAGE   VALUE   ENTERED    NEXT ACTION        DUE      SOURCE
Acme      3       £2,160  12 Aug     Send proposal      29 Aug   Referral
Belrose   4       £4,800  19 Aug     Walkthrough call   28 Aug   Trade forum
Corven    2       £1,800  22 Aug     Book discovery     27 Aug   Referral

Two columns matter more than the rest.

Next action and due date. A deal with no dated next step is not in your pipeline — it is in your memory, and it will be forgotten during a busy delivery week. This is the mechanism that makes follow-up happen at all.

Source. After twenty deals this tells you which channel actually produces revenue rather than activity, which is frequently not the one that produces the most leads.

The four ratios

Once around twenty deals have gone through, calculate:

1. Outreach → conversation — what share of approaches produce a real conversation
2. Conversation → proposal — what share of conversations are worth proposing to
3. Proposal → close — your close rate
4. Average cycle length — first contact to signature, in days

Now the pipeline forecasts. Working backwards from a target of ten new customers:

| | | |---|---:| | Customers wanted | 10 | | Proposal → close, 40% | 25 proposals | | Conversation → proposal, 50% | 50 conversations | | Outreach → conversation, 30% | 165 approaches |

Those percentages are placeholders until you have your own. That is the entire argument for tracking from deal one — twenty untracked conversations teach you very little; twenty tracked ones give you a planning model.

And the cycle length tells you when. If your average is 45 days, work started today lands in roughly six weeks — which means a thin month two months out is a prospecting problem you can still fix today.

Read the ratios as diagnosis

Each one points at a different problem:

• Low outreach → conversation. Targeting or message. You're reaching people who don't have the problem, or describing it in words they don't use.
• Low conversation → proposal. Either you're not qualifying (proposing to everyone) or the offer doesn't fit what you're hearing.
• Low proposal → close. Pricing, or the proposal itself. Check whether you're sending into silence rather than walking it through.
• High close rate, above roughly 70%. You're priced below the market. That's a pricing signal, not a triumph.
• Long cycle. Usually a decision-maker problem — you're talking to someone who has to persuade someone else.

Weight the forecast with your own numbers

The familiar 10/25/50/75/90 probability pattern is a guess presented as arithmetic. Once you have twenty closed deals, calculate what actually happened:

Of deals that reached stage 4, how many closed?

That percentage is yours, and it will differ from the template. Multiply the value of each open deal by its stage's historical rate and sum — that is a forecast with evidence behind it.

Before you have twenty deals, do not weight at all. Just count deals per stage and watch how they move. A fabricated probability is worse than no probability, because it produces confidence.

Close dead deals deliberately

The most common way a pipeline lies is by retaining deals that will never close. They inflate the total, produce a comfortable forecast, and consume follow-up attention.

Two rules:

Stage-age limit. Any deal sitting in one stage longer than your average cycle gets a decision — advance it, or close it out with the permission-to-close message.

Record a reason on every loss. Price, timing, scope, competitor, went silent, wrong fit. After twenty losses the pattern tells you what to fix, and it is rarely what you assumed.

The mistakes

1. Stages named after feelings. Unverifiable, so nothing ever leaves them.
2. Weighting before you have data. Confidence with no basis.
3. Deals with no dated next action. They aren't tracked, they're remembered.
4. Never closing anything out. An inflated pipeline forecasts a month that doesn't arrive.
5. Buying a CRM first. A tidy empty system is not progress.
6. Not recording loss reasons. You lose the only free feedback in sales.

What to do next

Build the seven-column sheet today and put every open deal in it with a dated next action. Then keep it for twenty deals before calculating anything — the ratios are worth waiting for, and they turn a list of hopeful names into something you can plan a business around.