It took meteorologists 200 years to get from folklore wisdom to reliable 10-day forecasts. The breakthrough wasn't a better thermometer. It was the realization that no single model can predict the weather. Only by combining multiple independent data sources—ensemble forecasting—did predictions become reliable.

Most agencies and IT service companies are still stuck in the folklore phase of revenue forecasting. They glance at the CRM, add up open deals, and declare: "Looks good." Or they check project utilization and hope the numbers work out.

They usually don't. Traditional B2B sales forecasts are accurate roughly 50% of the time. Coin-flip territory. On top of that, 44% of deals slip out of the pipeline unexpectedly. And according to Forrester, 86% of all B2B purchases stall at some point—usually right when you thought the close was certain.

Yet half of all agencies still forecast next quarter's revenue using a spreadsheet and the sales lead's gut feeling.

Why Single-Source Forecasts Fail

Meteorology has the butterfly effect. A small shift in initial conditions—a low-pressure system forming two degrees further south—can completely change the weather a week later.

Agency sales work the same way. A missed follow-up call. A personnel change at the client. A sudden budget freeze in November because the CFO got nervous. Small events that make a to-the-dollar pipeline forecast worthless overnight.

That's why no single forecasting method works reliably. If you only track open proposals, you miss that ongoing projects already consume all available capacity. If you only measure utilization, you miss the new retainer starting next month—or the one that just got cancelled.

Meteorologists solved this decades ago: combine multiple imperfect models. Each one is inaccurate on its own. Together, they produce forecasts that are surprisingly close to reality.

For agencies, this means something specific: stop trying to perfect one pipeline view. Instead, combine three independent revenue streams into a single forecast.

Three Revenue Streams, One Ensemble

A reliable sales pipeline has three layers. Each one is incomplete on its own. Together, they form the ensemble.

Open Proposals: Weighted Probability Instead of Wishful Thinking

Open proposals represent potential revenue. Emphasis on potential. The skill isn't having many proposals out there. It's being honest about which ones will actually close.

A digital agency in Hamburg—20 people, specialized in e-commerce—had a classic problem in 2023. The sales team rated every open deal at "90% probability." Sounds confident. In reality, it was self-deception. Actual forecast accuracy sat at 62%. Twice a year, management had to explain why the quarter came in 30% under plan.

After introducing weighted pipeline forecasting—based on historical close rates rather than optimism—accuracy rose to 84%. No magic involved. Just honesty in numerical form.

A €100,000 proposal at 50% probability enters the forecast at €50,000. Not at €100,000 and a good feeling.

In Leadtime, the Estimates Overview in the finance section shows exactly this: all open proposals with value, probability, and expected close date. Three proposals at €40,000, €50,000, and €30,000 with 60%, 70%, and 40% probability produce a weighted forecast of €71,000—not a guarantee, but a working foundation.

Ongoing Projects: Your In-Flight Revenue

Running projects determine what actually shows up on invoices in the coming weeks. For most agencies, this is where the bulk of short-term revenue lives. And yet surprisingly few know exactly when each project triggers which invoice.

An IT systems house in Munich—45 employees, focused on software integration—had exactly this transparency problem in 2024. Project plans existed. But they lived in one system, accounting in another. The CFO discovered every month—to his surprise—that either 40% more or 30% fewer invoices went out than expected. Cash flow planning? More like cash flow guessing.

The Turnover Insights Dashboard makes this connection visible: when revenue from current projects gets invoiced, broken down by billing model. Hourly, fixed-price packages, milestones.

A three-month website project with milestone billing—€15,000 after design, €20,000 after development, €10,000 after go-live—spreads across the forecast timeline. Instead of cramming everything into one month and being surprised when payment arrives only after sign-off.

Retainers & Subscriptions: The Predictable Foundation

The third layer is the most stable. Recurring revenue—retainers, maintenance contracts, monthly subscriptions—is the part of your forecast you already know at the start of each month. Regardless of whether a single new deal closes.

A consulting firm in Cologne with 30 consultants deliberately shifted their strategy in 2023. Instead of relying solely on one-off projects, they started actively selling monthly retainers. The result was less dramatic than expected—and still transformative: 60% of monthly revenue became predictable. Leadership could finally plan ahead realistically. Hiring. Investments. Without operating on a knife's edge every month.

Five clients at €5,000 per month each equals €25,000 in baseline revenue. Every month. That's the floor everything else builds on.

From Ensemble to Decision

A forecast that lives in a spreadsheet and plays no role in actual decisions is a waste of everyone's time. The real value emerges when the three layers together make options visible.

If the forecast shows two large projects likely starting the same month, you can act now—not when chaos has already arrived. Book freelancers. Shift timelines. Or tell a client honestly that their project can start two weeks later, before they notice themselves.

Meteorologists call this concept "confidence intervals"—ranges instead of point predictions. Your forecast isn't "Q2 will be €280,000." It's: "Q2 will be between €240,000 and €310,000, depending on three open deals and whether one retainer gets renewed." Less impressive on a management slide. Infinitely more useful for real decisions.

Why Three Data Sources Beat One

The reason ensemble forecasting works—in weather and in agency revenue—is almost obvious: every single data source has systematic blind spots. Open proposals overestimate. Project plans underestimate delays. Retainer forecasts ignore churn risk.

Combined, the biases partially cancel out. Not perfectly—as far as I can tell, no service company will ever match the accuracy of a 10-day weather forecast. But the jump from 50% to 80%+ accuracy is real when you stop relying on a single source.

Leadtime connects sales, project execution, and financial planning in one system. That means the three layers of your forecast draw from the same, current data. No manual consolidation across three tools. No stale spreadsheets that someone will "update next week."

We built this because at our previous company, we spent too many years staring out the window and hoping it wouldn't rain. Rarely worked.