Monte Carlo vs Weighted Pipeline Forecasting

What is the difference between Monte Carlo and weighted pipeline forecasting?

Weighted pipeline multiplies each deal by its close probability and returns one number. Monte Carlo simulates the quarter thousands of times and returns the full range of outcomes those same probabilities imply. Same inputs, very different output.

Weighted pipeline tells you the expected value. Monte Carlo tells you the expected value plus how much it can move and how often. If a CFO asks what the quarter looks like, weighted pipeline answers with one figure. On a hypothetical pipeline, that figure is $4.1M. Simulation on the same inputs answers that the median is $4.1M, that there is a 62 percent chance of clearing the $3.9M goal, and that the worst decile lands at $2.8M because three deals carry a third of the number.

The second answer is the one an operator can act on.

How does a Monte Carlo revenue forecast work?

Every open deal is flipped as a weighted coin, the wins are added up, and the whole exercise repeats until the shape of the outcome stabilizes.

The mechanics are simple:

1. Assign every open deal a close probability and an amount. 2. For one simulated quarter, draw a random number per deal and mark it won if the draw falls under its probability. 3. Sum the won amounts into a single quarter total. 4. Repeat ten thousand times and sort the totals.

The sorted list is the forecast. The 50th percentile is your median case, the 10th and 90th percentiles bound the realistic range, and the share of runs above quota is your probability of hitting the number.

Nothing in that process is exotic. A spreadsheet with a random function does it, and so does forty lines of Python. The difficulty was never the simulation, it was the inputs.

Which method should you use for which decision?

Use weighted pipeline when you need one comparable figure across teams. Use simulation when the decision depends on downside risk. The split is about the decision, not the sophistication. Weighted pipeline is fine for a coverage conversation or a rep-to-rep comparison. When the question is whether to pull spend forward, hire ahead of plan, or tell the board a number with a straight face, the range carries information the point estimate destroys.

Why does concentration risk break weighted pipeline?

Because an average hides the fact that one deal can decide the quarter. Take a $4M weighted pipeline built from four hundred small deals and a $4M weighted pipeline where three deals at 60 percent make up half the total. The expected values match. The risk profiles have nothing in common.

Simulation makes that visible immediately. The four hundred deal pipeline produces a narrow bell curve, since the law of large numbers does the work. The concentrated pipeline produces a lumpy, multi-peaked distribution where the outcome is basically a question of which of the three big deals land. No amount of staring at $4M reveals that.

This is the same critique that applies to pipeline coverage as a stand-alone metric. A team can hold 4x coverage and still miss badly when the pipeline is concentrated in a few deals, aged, or sitting in the wrong segment. Coverage and weighted value both compress the composition of the quarter into one figure, and composition is where the risk lives.

What do both methods miss?

Both only model the pipeline that already exists in the CRM. That is the gap that matters more than the choice between them.

A real quarter has three revenue sources:

- Carry-over deals already in the pipeline on day one that are expected to close this period. - In-quarter deals that do not exist yet and will be created, qualified, and closed inside the same period. - Pull-forward deals from future periods that close early, usually with a discount and a hole left behind.

Weighted pipeline models the first source. A basic simulation models the first source with better error bars. Neither models the second or third unless you add them on purpose.

The size of that omission is easy to underestimate. In ORM data, of the pipeline value carrying a close date inside the current quarter as of day one, roughly 20 percent actually closes in that quarter. The other 80 percent moves, shrinks, or dies. If four fifths of your dated pipeline does not land as dated, then a large share of every quarter has to come from deals nobody has created yet, and a simulation over existing records cannot see any of it.

How should you build a forecast that uses the best of both?

Keep the distribution, fix the inputs, and add the invisible pipeline as its own modeled term.

Start with probabilities grounded in historical conversion by segment and stage rather than in CRM defaults. Then correct the amounts, because deals close for less than they carry in the CRM more often than they close for more. Then add an in-quarter creation estimate built from your own history of pipeline created and closed inside the same period, and simulate that term too.

Once those three fixes are in, the simulation stops being a math exercise and starts describing the operating mechanics of the quarter. That is the same standard any method should be held to. Hand-built approaches usually reach around 90 percent accuracy on new and expansion revenue and cost real analyst time to hold there, and they go stale the moment conditions shift. ORM targets 95 percent without manual adjustments and holds it from day 1 through day 90, which is the window where a forecast can still change what you do.

Getting the number right in the last week is not forecasting. For the ordering of the build, see how to forecast revenue and treat both weighted pipeline and simulation as views on top of it rather than replacements for it.