Every acquisition I have ever seen begins with a beautiful set of projections.

They are usually clean. They reconcile. They grow at a sensible rate. Margins expand just enough to feel earned, but not so much as to feel greedy. If projections were houses, these would be freshly painted with working door handles and no visible cracks.

And to be clear, there is nothing dishonest about them. Even distressed acquisitions have beautiful projections; the buyer is betting that he/she will turn it around or sell the parts at a greater price.

The problem however is that projections do not break loudly. They break quietly. One assumption stops being true, and the rest of the spreadsheet politely follows it off a cliff.

Let me start with a story.

A buyer acquired a mid-sized energy services business. The company had been around for over a decade, operated in a familiar basin, and showed steady growth. The projections assumed modest revenue expansion, flat pricing, and “conservative” cost adjustments. Nothing heroic.

The buyer was disciplined. He haircut growth, normalized margins, and even ran sensitivities.

Six months after closing, the business was under stress. Not because revenue collapsed. Not because oil prices crashed. Not because management lied.

One assumption quietly stopped being true.

A large customer changed how it scheduled work. (We will talk more about concentration next month). Work still existed, but it came in bursts instead of evenly. That small operational change rippled outward. Accounts receivable stretched. Working capital expanded. The revolver filled faster than planned. Decisions that were “temporary” became permanent.

The projections did not fail all at once. They simply lost their footing.

I know projections are guesses. You know this too. Yet we both anchor to them anyway. I have watched disciplined buyers - myself included - adjust around the CIM instead of questioning whether the CIM's world still exists.

We tell ourselves we are being conservative because the spreadsheet feels sober. We forget that exceptional years tend to regress to the mean, and average years have a bad habit of showing up unannounced.

Worse, we confuse volume with resilience. Ten customers feel safer than two, until they all stop calling the same week. Then you realize they all behave the same way under stress, so you really had customer concentration. The spreadsheet does not warn you of this.

A projection is saying what will happen if a set of things remain true. A projection is a 'what-if' story told through numbers. The mistake buyers make is focusing on how sensitive the numbers are (sensitivity analysis: tinkering with variables to see if the math still works), instead of how fragile the assumptions are (fragility analysis: questioning if the world that allows those variables to exist still stands).

A 10% downside case is comforting. It gives the illusion of preparation for a floor. But it avoids the harder question: What has to remain true for this business to perform at all?"

Will the drilling superintendent still call you when his vendor screws up? Will the three guys who actually know how to run the equipment stay after the sale? Will the bank keep the revolver open when receivables stretch from 45 days to 75?

That is a decision-making question, not a modeling one, which is why buyers have to be conscious about improving their decision-making skills.

When I look at projections, I mentally sort assumptions into four buckets. If any one of these quietly fails, the math will follow.

I have seen so many analysts obsess about creating robust projections that capture “every” possible path. This is extremely valuable but sometimes misses the point. Analysts dream of the opportunity to say, “my projections were right”. Whereas for buyers, a good projection is not the one that comes true, it is the one that clearly tells you what decision you are actually making.

You are not betting on a growth rate, that’s the outcome. And as I said earlier we need to stop treating projections as outcomes. What you are really betting on that customers behave, people stay, capital flows, and control is real.

The marbles are great until they break.

You will not predict which one cracks first. No one does.

But you can know which ones, if they break, take the whole game with them.

That is not forecasting. That is decision-making.

Next in this two-part series on Decision-Making Under Uncertaintyseries: Improving Decision-Making Skills as a Searcher: A Framework for Capital Allocation Under Incomplete Information.