Pipeline Design

A pipeline nobody updates is worse than no pipeline, because it produces numbers that look real.

Forecasting from a stale pipeline is how businesses plan around revenue that was never going to arrive.

The fix isn't discipline. It's designing stages people can actually move records between.

See the CRM Blueprint · Request Your Free Systems Audit

The one rule

A stage is a state the record is in, not a task someone performs.

The difference matters because a state can be verified by looking. A task can't — "send proposal" is either not done, in progress, or done, and the pipeline can't tell you which.

When stages are tasks, records stop moving, because nobody knows whether a half-done task counts.

How many stages

Five to seven. Below five you lose useful information; above seven the distinctions stop being meaningful and people stop moving records.

If you're tempted by nine stages, the usual cause is trying to encode a process inside the pipeline. The process belongs in a procedure. The pipeline records where things are.

Define each stage properly

For each stage, write three things down:

Entry condition — what makes a record arrive here. Specific enough that two people would agree.

Exit condition — what makes it leave, and where it goes next.

Maximum age — how long a record can sit here before it's stale.

The third is what stops a pipeline becoming a graveyard. Without it, records sit in "Negotiating" for eight months and your pipeline total is fiction.

A workable default

Adjust the numbers to your sales cycle. Have numbers.

Define "Lost" precisely

The most avoided decision in pipeline design: how long without a response before a record is lost?

Pick a number. Three weeks, six weeks — the specific figure matters less than the existence of one.

Without it, records never leave the pipeline. They accumulate, the total inflates, and the forecast becomes meaningless. Nobody wants to be the person who marked something lost, so nothing ever is.

Lost is not permanent. A lost record can be revived. Treating it as a failure is why pipelines bloat.

Always capture a lost reason

A required dropdown, not free text. Six to eight options:

Price · timing · went with a competitor · did it in-house · no budget · no response · not a fit · project cancelled

This is the single most valuable field in a CRM and the one most often left out. Without it you cannot answer the most useful question about your sales process: why do people say no?

Six months of lost reasons will tell you more about what to change than any amount of analysis of the deals you won.

Required fields at each transition

A small number of fields should be mandatory before a record advances:

Two or three fields per transition. More than that and people start working around the system — which is worse than having no requirement, because now the data is both incomplete and wrong.

Next action, always

Every record in an open stage needs a next action and a date.

This single discipline does more for pipeline hygiene than any other. A record without a next action is not being worked on, whatever stage it's sitting in.

Make it visible: a list of records with no next action, or a next action date in the past, reviewed weekly. That list is your actual to-do list.

Don't build separate pipelines for everything

The common mistake is a pipeline per service, per team, or per location. You end up with six pipelines, no overall view, and records in the wrong one.

Use one pipeline and a field for the distinction. Filter when you need to look at a segment.

The exception is a genuinely different process — new business versus renewals, for example, where the stages themselves differ. Different stages justify a different pipeline. Different categories of the same process don't.

What to automate, and what not to

Worth automating:

• Alerting when a record exceeds its stage maximum age
• Flagging records with no next action date
• Creating a follow-up task when a stage changes
• Notifying the owner when a record enters a stage

Not worth automating:

• Moving records between stages automatically. The movement is the human judgement; automating it produces a pipeline that advances on its own and reflects nothing.
• Emailing the customer on every stage change. Internal states are not customer-facing news.

Keeping it honest

Weekly: records past their stage maximum, records with no next action.

Monthly: does the stage distribution look plausible? A pipeline with forty records in "Proposal sent" and three in "Negotiating" means proposals aren't being followed up.

Quarterly: are all stages being used? A stage nobody uses should be removed. Are lost reasons being recorded? If a third of lost records have no reason, the field isn't required.

Signs your pipeline isn't working

• Records that haven't moved in over a month and nobody has noticed
• Anyone maintaining a parallel spreadsheet
• The forecast being adjusted by instinct before it's used
• "Lost" being almost empty while "Negotiating" keeps growing
• Needing to ask someone for the status of a deal

The last one is the clearest. If the pipeline can't answer it, it isn't a pipeline.

A stage is where something is, not what someone needs to do. Design it that way and people will keep it current.

Pipeline not reflecting reality?

A free systems audit covers your pipeline structure, your data quality, and what to fix first.

Request Your Free Systems Audit

More in this category

• Lead Capture Best Practices
• Automation Planning Checklist