The email arrived at 2200: "New guidance from DoD. All construction projects suspended pending review. Effective immediately."

We were four months into the fiscal year at Malmstrom Air Force Base. I was managing a $125 million budget, and construction made up $40 million of it. Projects were already underway. Contracts were signed. Vendors had mobilized.

And now? Frozen.

By 0800 the next morning, I needed a new forecast. Not a guess—a defendable projection of what would actually execute, what could be redirected, and what impact this would have on the entire base operation.

Welcome to forecasting under fire.

The Illusion of Certainty

Most business forecasts are fiction.

Not lies—fiction. Carefully constructed narratives built on assumptions that might have been true when you made them but are already outdated.

You forecast Q2 revenue at $X because that's what happened last Q2, adjusted for growth. Except three things have changed: a competitor launched a new product, your top sales rep quit, and there's now a supply chain delay affecting your main product line.

Your forecast is already wrong, and the quarter hasn't even started.

The military taught me something critical: the purpose of a forecast is not to be right. It's to be useful.

Let me explain.

Useful Beats Accurate

In Iraq, managing $33 billion in financial flows, I never once built a forecast that was "accurate."

How could I? The operational environment changed daily:

  • Missions got accelerated or delayed based on intelligence
  • Equipment broke and needed emergency replacement
  • Coalition partners changed their funding commitments
  • Iraqi ministries struggled with their own budget execution
  • Congressional guidance shifted mid-year

If I'd tried to build an "accurate" forecast, I would have spent all my time updating it and no time actually managing the money.

Instead, we built useful forecasts—projections that helped us make better decisions even when they were wrong.

The difference is fundamental.

The Three-Forecast Model

At Peterson Air Force Base, overseeing $7 billion in space programs, I learned that single-point forecasts are worthless.

Reality doesn't land on a single number. It lands in a range.

We started building three forecasts:

The Optimistic Case

What happens if things go better than expected?

  • Sales cycles close faster
  • Costs come in under budget
  • Efficiency gains materialize
  • New opportunities emerge

This isn't fantasy—it's "what if the winds are at our back?"

Probability: ~20%

The Realistic Case

What happens if things track roughly to plan?

  • Most projects deliver close to schedule
  • Costs run near budget
  • Revenue comes in within 10% of target
  • Normal operational friction

This is your baseline—the most likely path based on current information.

Probability: ~60%

The Defensive Case

What happens if things go south?

  • Key projects get delayed
  • Major customer churns
  • Costs overrun
  • Market conditions deteriorate

This isn't doomsday—it's "what if we hit significant headwinds?"

Probability: ~20%

Now you've got a range. And ranges let you plan.

In the optimistic case, you're funding growth initiatives and building capacity.

In the realistic case, you're executing the plan and building reserves.

In the defensive case, you're cutting costs and preserving cash.

You're not trying to predict which one will happen. You're prepared for any of them.

Rolling Forecasts Beat Annual Budgets

Here's a dirty secret about traditional budgets: they're obsolete the day you finish them.

You spend October through December building next year's budget. By February, market conditions have changed. By April, you've launched new products. By July, half your assumptions are wrong.

But you're still managing to a budget that was built on October's information.

At Malmstrom, we shifted to rolling forecasts. Instead of one annual budget, we maintained a continuous 12-month outlook that we updated every month.

Here's how it worked:

On the 5th of every month (after the prior month closed), we'd:

  1. Drop the month we just completed
  2. Add a new month 12 months out
  3. Update the intervening months based on current information
  4. Identify variances from the prior forecast
  5. Adjust resourcing and priorities accordingly

This gave us something the annual budget never could: current information driving current decisions.

When that DoD construction freeze hit, we didn't need to rebuild the forecast from scratch. We updated the rolling forecast with new assumptions, and immediately saw the ripple effects across the next 12 months.

Decision made in hours, not weeks.

Leading Indicators Over Lagging Metrics

Most business forecasts are built on lagging indicators—things that already happened.

Last month's revenue. Last quarter's pipeline. Last year's growth rate.

That's like driving by looking in the rearview mirror.

In combat finance, we tracked leading indicators—signals of what was coming:

  • Commitment rate: How fast were we obligating funds? (Predicted year-end execution)
  • Requisition pipeline: What was in the approval chain? (Signaled next month's spending)
  • Vendor performance: Were deliveries on time? (Indicated schedule risk)
  • Leadership priorities: What was the commander focused on? (Showed where resources would flow)

These gave us 30-60 days of warning before issues hit the financials.

In business, leading indicators might be:

  • Pipeline velocity: How fast are deals moving through stages?
  • Customer engagement: Are users more or less active?
  • Employee sentiment: Is the team confident or concerned?
  • Market signals: What are competitors doing? What are customers saying?

Track these, and you see problems before they hit your P&L.

Scenario Planning for Black Swans

In 2006, nobody was forecasting a global financial crisis.

In 2019, nobody was forecasting a pandemic shutdown.

In 2022, nobody was forecasting Silicon Valley Bank would collapse.

Black swans happen. The question is: are you ready?

At Malmstrom, we ran quarterly scenario exercises. We'd take 90 minutes and war-game through "what if" scenarios:

Scenario 1: "Congress cuts our budget by 20% mid-year. What gets cut? What's protected? How fast can we react?"

Scenario 2: "A major disaster hits and we need to deploy 50% of our people on 72 hours' notice. Who covers what? How do critical functions continue?"

Scenario 3: "Our largest vendor goes bankrupt mid-contract. What's our backup plan? How much does it cost? What's the timeline?"

We didn't predict which scenario would happen. We built muscle memory for rapid response.

When that construction freeze hit, we didn't panic. We'd war-gamed a similar scenario six months earlier. We knew our options, our decision tree, and our fall-back positions.

Execution was clean because planning was thorough.

The Weekly Forecast Refresh

Here's what killed most forecasts I've seen: they're static documents that get updated quarterly (maybe).

We made forecasting a weekly discipline:

Monday morning (first thing, 30 minutes):

  1. Review actuals vs. forecast from last week. Not to assign blame—to understand variance.
  2. Update current month projection based on what we now know. Week 1? Still pretty uncertain. Week 3? High confidence.
  3. Flag risks to out-months. What changed this week that affects Q2? Q3?
  4. Adjust resource allocation. If the forecast shifted, do we need to reallocate people or money?

This wasn't a formal meeting. It was a discipline. The finance lead spent 30 minutes every Monday refreshing the outlook.

Every week.

That weekly cadence meant our forecast was never more than 7 days stale. Compare that to most businesses, where the forecast is 90+ days old before anyone updates it.

The Confidence Meter

Not all forecasts are equally reliable.

In the military, we used confidence levels:

  • High confidence (80%+): Based on signed contracts, committed funding, projects in execution
  • Medium confidence (50-80%): Based on strong pipeline, historical trends, normal assumptions
  • Low confidence (<50%): Based on early indicators, market signals, informed speculation

When we briefed the commander, we'd color-code the forecast:

  • Green = high confidence
  • Yellow = medium confidence
  • Red = low confidence

"Sir, we're forecasting $8.2M in execution this month. $6.1M is green—already committed. $1.5M is yellow—in final approval. $600K is red—dependent on vendor delivery that's been inconsistent."

Now the commander knows where the risk is. He's not assuming the whole $8.2M is certain.

In business, this looks like:

  • Green revenue: Signed contracts, recurring subscriptions, booked orders
  • Yellow revenue: Verbal commits, late-stage pipeline, renewal assumptions
  • Red revenue: Early pipeline, new markets, optimistic assumptions

Hit your green number? You're stable. Hit green + yellow? You're on plan. Need red to materialize? You're taking risk.

Transparency matters.

Reforecasting Without Drama

The worst thing you can do is make forecast updates feel like failure.

I've seen businesses where revising the forecast is treated like admitting defeat. So people cling to outdated projections rather than acknowledge reality.

That's insane.

In Iraq, we reforecast constantly. Sometimes daily. Not because we were wrong—because conditions changed.

A mission accelerated? Reforecast.

A contract got delayed? Reforecast.

New guidance from DoD? Reforecast.

There was zero shame in it. The question wasn't "why was your forecast wrong?" It was "given what we now know, what's the new projection?"

Make reforecasting normal. Make it fast. Make it blameless.

Building Your Forecast Discipline

If you want forecasting that survives contact with reality, here's where to start:

This month:

  1. Build three versions of your forecast: optimistic, realistic, defensive
  2. Identify your top 5 leading indicators and start tracking them weekly
  3. Add confidence levels to your current forecast (green/yellow/red)

This quarter:

  1. Shift to a rolling 12-month forecast, updated monthly
  2. Run your first scenario planning session (pick 3 "what if" scenarios)
  3. Implement weekly forecast refresh discipline

This year:

  1. Train your leadership team to forecast in ranges, not single points
  2. Build leading indicator dashboards that predict 30-60 days out
  3. Make reforecasting a normal, blameless part of operations

The Forecast Is a Tool, Not a Promise

After two decades in finance—from managing small budgets as a junior enlisted troop to overseeing $33 billion in combat operations—here's what I know:

The forecast will be wrong. That's not the point.

The point is to have a framework that helps you make better decisions in uncertainty. To see risks before they become crises. To allocate resources based on current information, not outdated assumptions.

A good forecast doesn't predict the future. It prepares you for multiple futures.

That's how you plan under fire.

—Gabe

Next in the series: Military-Grade Financial Discipline - the habits that separate good from great.