Predictability or Opportunity?
Updated: Aug 10
Why a more predictable pipeline may be holding you back.
A friend recently asked me about forecast accuracy:
Is it better to have more deals in play—or fewer deals with the same expected number of wins?
Working through the answer took me down some nuanced paths. It is a great question because the answer exposes a danger of pursuing forecast accuracy.
Here's the TL;DR.
If the only objective is to make the forecast more accurate, the better pipeline is the one with fewer, higher-probability deals. But that may not be the pipeline with the greatest potential for sales.
A Commit list illustrates the tension. The more you limit the forecast to high-probability deals, the more predictable it becomes. But that same narrowing can also pull management attention toward the safest part of the pipeline.
Of course, sales teams also work deals in other forecast categories. The issue is whether a forecasting process organized around confidence-to-close systematically pulls attention toward the deals that are easiest to predict rather than the deals where additional effort can create the most incremental sales. Our evidence suggests that this bias is real and better focus can yield 60% more sales productivity.
What follows is probably not how you think about your forecast. The example below looks at the question above through the distribution of possible outcomes. It asks what range of outcomes different pipeline compositions can produce—not just the expected result.
That distinction matters because the deals that make the forecast most reliable are not necessarily the ones where management has the best opportunity to change the outcome.
Fewer deals make more accurate forecasts
Consider two pipelines. To isolate the effect of probability, assume all deals are the same size, require the same selling effort, and are independent.
Pipeline A: 50 deals, each with a 90% probability of being won
Pipeline B: 450 deals, each with a 10% probability of being won
For now, treat them as alternative pipelines. We will later consider what happens when both sets of opportunities are available to management.
Both pipelines have the same 45 expected won deals. But their distributions are very different.
Pipeline A produces a much tighter range of possible outcomes. Pipeline B produces a wider range. So, under a no-feedback forecast—one in which the forecast does not influence what happens—Pipeline A is more predictable.
In fact, from a conventional reward-and-risk perspective, Pipeline A is better in this toy example. Both pipelines have the same expected payoff—45 wins—but Pipeline A has much less uncertainty around that outcome. Pipeline B offers more upside, but only by accepting correspondingly more downside risk. And if you were constrained to working 50 deals, you would choose the 50 with a 90% chance of winning over 50 drawn from the 10% pool.
But the example is artificial. Every deal is identical in size. And it assumes management cannot change the result. The team is just taking orders. Under those assumptions, Pipeline B offers no prioritization advantage. There is no reason to prefer one 10% deal over another, and no intervention can improve the outcome.
Now suppose all 500 opportunities are available to the same company. It likely has the capacity to work more than the 50 safest deals—but not enough capacity to give all 500 everything they need.
Real pipelines are also not composed of interchangeable deals. Win probabilities differ, and management actions affect some opportunities more than others. Sales leaders must decide where to invest limited resources.
That changes the problem from choosing between two pipelines to choosing where within the broader pipeline to focus.
The bigger pipeline offers more opportunities to intervene
Some deals may respond to executive involvement. Others may benefit from technical resources, pricing support, a different stakeholder strategy, or more intensive follow-up. Some may be unlikely to close regardless of effort.
The value of the broader pipeline is not simply that it contains more deals. More deals alone can create more noise. The potential value comes from the differences among those deals—and from management’s ability to identify where intervention creates the greatest impact.
That violates the no-feedback assumption, which is exactly the point. Sales leaders do this every day.
A sales forecast is not a weather forecast. Sales leaders can change what happens.
And the deals that make your forecast most reliable may be precisely the deals where management intervention has the least leverage to improve the outcome.
More opportunities create a harder focus problem
The larger pipeline also creates a management challenge:
Which deals should the team focus on?
As the number of opportunities grows, the number of possible ways to allocate attention explodes.
Say you have the resources to pursue 10 deals (assume all take the same effort). If you have 12 deals in your pipeline, there are 66 possible 10-deal portfolios. With 100 deals to choose from there are more than 17 trillion 10-deal portfolios.
Without a way to optimize focus, more opportunity can become a distraction.
But focus can be optimized with sales math.
Is it really about accuracy?
In the real world, there is a tension between focusing on the most predictable part of the pipeline and pursuing the broader opportunity set that may offer more room to grow.
That matters because one of the most common problems we hear is:
We need to improve forecast accuracy.
This is usually driven by the pain of missed forecasts and erratic sales. But often, the business problem is not the forecast. It's that the company is not selling enough.
Improving forecast accuracy does not necessarily solve that problem. In fact, a company can improve apparent forecast accuracy by issuing a conservative forecast, such as a Commit list that resembles Pipeline A, and managing toward it. The forecast becomes easier to hit, but the team may ease up once the number looks secure.
Accuracy improves while sales remain below their potential.
What about those 95% forecast accuracy claims?
SalesTech vendors routinely claim 90%, 95%, 97%, or even 98% forecast accuracy, often coupled with “early in the quarter” qualifiers.
Sounds great. But what do those claims mean? These products are forecasting sales bookings, so for purposes of this test, we interpret 95% forecast accuracy to mean actual bookings (not GAAP revenue) landing within ±5% of the forecast.
Many sales organizations divide the pipeline into categories such as Commit, Best Case, and Pipeline. Those labels imply very different odds of closing. But even if those odds were known perfectly, the resulting forecast would still have a range of possible outcomes.
Our interactive Forecast Accuracy Explorer models pipelines with three Forecast Categories. You choose how many deals are in each category and the win probability associated with each.
It then calculates the exact distribution of possible outcomes and asks a simple question:
What is the probability that the 95% accuracy claim will fail—that actual bookings will fall outside ±5% of the forecast?
The explorer also shows the implied deal coverage produced by the mix of categories—useful because a forecast can become more predictable simply by concentrating the pipeline in higher-probability deals.
Adjust the number of deals and win probabilities in each Forecast Category to reflect your business. Under many realistic combinations, repeatable ±5% accuracy is mathematically unlikely—even when those probabilities are known exactly.
But that's only half the point.
Selling more is the objective
So if consistently hyper-accurate forecasting is not realistic, what’s the point of a forecast?
Its greater value is in helping management decide what to do next.
Most forecast processes don't do a good job at this. A Commit-centered management process can pull attention toward the Pipeline A portion of the opportunity set.
But sales math can help you focus better. It can answer: Which deals deserve attention? Where can intervention improve the odds? Where should scarce resources be allocated? Which parts of the pipeline need more demand generation? How can focus shift the distribution toward a better result?
This is forecast feedback, and it is the sales superpower.
When teams focused on the deals prioritized by Funnelcast, they produced 60% more in-quarter sales on average than when they followed same-sized lists based on their existing priorities.
Same pipelines, same team, better focus, better result.
Ask a better question
Before trying to improve forecast accuracy, ask whether the target is mathematically realistic—and whether achieving it would actually help the company sell more.
The pipeline that produces the most accurate forecast is not necessarily the best pipeline to focus on—particularly when you have the capacity to work beyond the safest deals and a way to optimize resource allocation.
A forecast should not merely predict the result. It should help improve it.
Comments