How to Build a Sales Pipeline That Actually Forecasts

A sales pipeline forecasts when its stages are defined by what the buyer has done, not by what the rep hopes happens next. Most pipelines fail as forecasting instruments because they measure seller effort. A rep sends a deck and drags the deal to "Proposal." Nobody on the buying side has agreed to anything, but the stage now says progress and the forecast built on top of it inherits the fiction. Building a pipeline that forecasts is mostly about removing that fiction. You define each stage by a buyer action, you write down what has to be true to leave it, you size how much coverage you actually need, and then you keep the whole thing clean enough to trust. We build forecast models on top of these pipelines at ORM, and the ones that predict well are the ones that were built honestly in the first place.

Why should you define pipeline stages by buyer action, not seller activity?

Because a seller's activity tells you what your team did, and a buyer's action tells you whether you are going to win. "Sent proposal" is an activity. "Buyer confirmed budget and looped in their VP of Finance" is a commitment. Only the second one moves a deal closer to closed, and only the second one belongs in a stage definition.

The test is simple. For every stage, ask what the buyer must have done for a deal to sit there. If the answer is something your rep controls alone, the stage is measuring the wrong side of the table. Stages anchored to buyer actions are hard to inflate, because a rep cannot fake a signed order form or a scheduled technical evaluation. That is exactly why they forecast better. The pipeline stops reflecting optimism and starts reflecting evidence.

What are the stages of a sales pipeline?

A workable B2B SaaS pipeline has five stages, each defined by a buyer action and each with a written condition for moving forward. The table below is the version we recommend teams start from and adapt to their motion.

Five stages is usually enough. When teams run ten, most of the extra ones describe internal steps no buyer would recognize, and the pipeline gets harder to keep clean without forecasting any better for the trouble.

How do you set exit criteria that hold?

Exit criteria hold when every one of them is a fact you could verify from outside the deal. "Rep feels good about it" is not verifiable. "Buyer returned a redlined contract" is. Write the criteria as observable events, require all of them before a deal advances, and the stage becomes a real gate instead of a label.

This is where win rate discipline is won or lost. A stage with soft exit criteria fills up with deals that were never really qualified, which drags conversion down at every later stage and makes your historical stage-to-stage rates useless for prediction. Tight criteria keep the junk out early. The deals that survive each gate start to look alike, so the rates you measure between stages actually mean something, and a model can learn from them.

How much pipeline coverage do you need?

Enough that the deals likely to close cover the target, which for most teams lands between 3x and 5x, though the honest answer depends on your win rate. Pipeline coverage is the ratio of open qualified pipeline to your goal for the period. Across ORM customers the median sits around 3.5x, with a spread from roughly 1.4x on the tight end to 5x on the loose end. A predictable, high win rate needs less coverage. A volatile one needs more.

The number to distrust is a single coverage figure quoted with no context. A team can carry 4x and still miss badly if the pipeline is concentrated in one stage, dependent on a couple of large deals, or stuffed with opportunities nobody has touched in months. Coverage is an input, never the conclusion. Before you trust yours, work out how to calculate pipeline coverage properly, and understand why the 3x rule on its own is wrong.

Why is pipeline hygiene the real forecasting problem?

Because a pipeline is only as good as its hygiene, and most pipelines are dirtier than the people running them believe. The pattern shows up in our data over and over. At least 10% of the average pipeline is stale, meaning nobody has touched it in twelve months, yet it still sits there inflating coverage. Worse, on the first day of the quarter only about 20% of the pipeline already dated to close in that quarter actually closes inside it. Eighty percent of the value you can see on day one will not land the way the dates claim. If your forecast trusts those dates, it is wrong before the quarter starts.

Cleaning it comes down to defining what counts as real. We treat meaningful activity as a change in stage, close date, or amount. A logged call or a fresh note is fine, but it is not movement. When a deal has gone months with no change to any of those three fields, it is not pipeline, it is decoration, and it should come out of the forecast until something real happens.

The single most useful signal to watch is a rep moving a close date. When a seller pushes a date out, the deal is telling you it is slipping, and a deal that slips from one quarter to the next is less likely to close at all, even when it is sitting in commit. That one field change is more predictive than most of the optimism stacked around it. Watch it early, because catching deal slippage before it wrecks the quarter is the difference between adjusting in week two and apologizing in week twelve.

What turns a clean pipeline into a forecast?

A clean pipeline becomes a forecast when the structure you built lets a model see the quarter before it happens. Buyer-action stages feed it honest inputs. Because the exit criteria are verifiable, the stage-to-stage conversion rates stay stable enough to trust. Coverage read in context tells you whether the raw volume even sits in the right place, and ruthless hygiene strips out the stale and slipping deals so the model reads signal instead of noise. Get those four right and you can split the quarter into what will close from existing pipeline and what still has to be created, which is the whole point of learning how to create a sales forecast that holds up past week one. The pipeline is the instrument. Keep it honest and it will tell you the truth early enough to do something about it.