Field notes
Why your B2B sales forecast is always wrong
Your reps aren’t the problem. The forecast is built on what they hope will close, and that doesn’t hold up under a quota. Three causes, and how to fix each one.
The forecast is built on hope, which is why it misses. Tighten your deal definitions and base the number on evidence rather than optimism, and it turns into something you can plan against.
On this page
Your forecast misses because it's built on what reps hope will close. The reps can sell fine; hope just doesn't hold up as a number you can bank on.
I've reviewed a lot of pipelines. The forecasts that miss tend to break in the same few places, and the forecasting tool is never one of them.
Part of it is that the data was never good to begin with. Salesforce's own State of Sales research finds reps spend less than 30% of their time selling - the rest is admin, much of it feeding a CRM they don't trust enough to feed carefully.
The stages mean nothing
Ask two reps what "stage three" means and you'll get two answers. If a stage has no exit criterion, a specific, checkable thing that must be true to advance, then "stage three" is just a feeling with a number attached, and that feeling is what you're forecasting.
The close dates are fiction
Close dates get set once and never moved. A deal slips a month; the date doesn't. By quarter-end you've got a stack of deals "closing this week" that have been closing this week since March. Every pipeline has a graveyard, and most people forecast straight off it.
Nobody's paid to be honest
A rep who sandbags looks like a hero when they beat the number. Play it straight and you look like you're missing. Optimism gets rewarded, so optimism is what lands in the CRM. The number that travels up the chain is one nobody really trusts.
What actually fixes it
Define exit criteria per stage. Not a description, a checklist: "Economic buyer identified. Next step booked. Pain confirmed in writing." A deal advances when those boxes are ticked, and a rep feeling good about it counts for nothing. Now a stage means something, and stage-based forecasting starts to work.
Make slippage visible. Track how many times a close date has moved. A deal that's slipped three times is not closing this quarter, whatever the CRM says. Surface it and the graveyard empties itself.
Forecast from the system rather than the story. When the stages are real and the dates are honest, the pipeline forecasts itself, with no spreadsheet and no Friday call where everyone reads out numbers they invented that morning.
This was never about a forecasting product. The problem sits upstream, in data you can't trust, and the forecast is just the first place it surfaces. Fixing that is a RevOps job, and a GTM diagnostic is the fastest way to see how far off your pipeline really is. Sort out the inputs and the number stops being dramatic, which is the whole point.