A stage is an opinion a rep typed. A sent proposal is a number somebody asked you to put in writing.
Most forecasting arguments are about weighting. What should stage four be worth, is sixty per cent too generous, should the number flex by rep.
The arguments are long because the input is weak. A pipeline stage is a judgement, entered at a point in time, by a person with an incentive to be optimistic, and nothing in the system forces it to change when the deal does. Two deals sitting in the same column can be a week apart and a quarter apart, and the forecast has no way to tell.
Meanwhile there is a far better signal in the same CRM, usually unused for this purpose entirely.
Why a sent proposal is different in kind
Three things are true of a priced proposal that are not true of a stage.
It has a number. Not an estimate of deal size entered at creation and never revisited. The actual figure, for the actual scope, that you would honour if they said yes this afternoon.
It was requested. Nobody asks for a priced proposal for something they have no intention of considering. The request is a small act of commitment, and it is made by the buyer rather than recorded by the seller. That is the whole difference.
It has a date. You know exactly when it was sent, which means you know exactly how long it has been sitting, which is the most predictive thing about it.
A stage has none of these properties. It is a label describing how a rep feels about a conversation, and it degrades quietly the moment attention moves elsewhere.
The forecast this gives you
Split the number in two and stop trying to make one figure do both jobs.
The committed number is proposals sent and not yet answered, each at its real value, aged. This is your next few weeks, and it is accurate enough to plan on.
Coverage is everything earlier than that, and it answers a different question: whether there will be enough proposals to send in two months. Coverage is a volume question, not a value question, and weighting it to a dollar figure is where most forecasts go wrong.
Keeping them separate makes both more useful. When the committed number is light, you have a closing problem this month. When coverage is light, you have a pipeline problem next quarter. Blended into one weighted figure, both failures look the same and neither gets fixed.
Ageing is the part people skip
A proposal's age is more informative than its value.
Work out your own average time from proposal to decision, which takes ten minutes against last year. Then anything meaningfully past that average is not pending. It is a no that nobody has been willing to say, and treating it as pipeline is how a forecast stays comfortable and wrong.
The useful consequence is that this is actionable in a way a stage never is. "This has been with you three weeks and our usual answer comes back in eight days, so I would rather hear a no than keep guessing" is a real conversation with a real prompt behind it. "I notice you are still in stage four" is not a sentence anyone can say out loud.
What engagement adds, and what it does not
If your proposals go out as tracked links, you can see whether they were opened, how often, and by how many people.
What that reliably tells you is whether it reached the right room. A proposal opened once and never again has been read by your contact and nobody else. A proposal opened six times across three days by more viewers than you sent it to has been circulated internally, which is the strongest buying signal available short of a signature. It means somebody is building a case on your behalf.
What it does not tell you is intent. Plenty of deals close after a single open, and plenty of enthusiastically opened proposals were being used to benchmark the incumbent. Treat engagement as evidence about distribution, not about enthusiasm, and it will stop misleading you.
Why this does not happen by default
Two reasons, both ordinary.
The proposal is usually built somewhere else. A document tool, a spreadsheet, a template a rep keeps locally. So the number that the buyer actually received never arrives back in the system that produces the forecast, and the CRM holds an estimate from three weeks earlier.
And the proposal is treated as an output rather than an event. It gets sent, the deal gets dragged a column to the right, and the artefact itself is filed. All the information in it is discarded at exactly the moment it becomes true.
In Signal, proposals are built from the deal and the product catalogue, so the total is worked out rather than retyped and the proposal and the deal cannot disagree. They go out as a tracked link, they carry draft, sent and accepted states, and the Proposal Hub shows the value in play with engagement per proposal. That last screen is the forecast, and it needs no weighting argument to be useful.
The test for any forecast
Ask what would have to change for the number to move.
If the answer is "a rep updates a field", the forecast measures attention rather than demand, and attention is lowest in exactly the weeks you most need the number.
If the answer is "a customer answers a proposal", you are measuring the thing you actually want to know.
Nobody needs a better weighting model. Most teams need a forecast that is built from what buyers did rather than from what sellers typed.
