There’s a particular kind of dread that sets in when a number you trust starts sliding the wrong way and you can’t explain it.
A solar company in Colorado knows that dread well. Their margins on Colorado work had held steady for a long time, healthy enough that nobody gave them a second look. Then, over a couple of months, profitability in the state fell by half. No warning. No obvious cause. Just a number that used to be fine, and suddenly wasn’t.
When the ground shifts like that, the instinct is almost always the same. You look for the thing that changed most recently, and you point at it. The company had just installed a new general manager in the region, so the story practically wrote itself. New leader, new problems. Maybe he was in over his head. Maybe the team hadn’t taken to him. There were meetings about it, hard conversations, the slow and uncomfortable work of wondering whether they’d made a hiring mistake.
They hadn’t. The new manager had nothing to do with it. But it took a while, and a different way of looking at the data, to see that.
The clue was hiding in the average
A company-wide margin number is a comforting thing to watch, right up until it betrays you. The trouble is that it’s an average, and averages survive by smoothing. They take the good and the bad and blend them into one tidy figure, then quietly bury the real story underneath.
“Colorado is down fifty percent” sounds like a problem with Colorado. So the company did something simple, and a little tedious. Instead of looking at the state as one block, they broke it apart by AHJ, the authority having jurisdiction, and looked city by city, county by county. The moment they did, the fog lifted.
It wasn’t all of Colorado. It was Denver. One city, dragging the whole state’s number down, while the healthy work everywhere else propped up the average just enough to keep the real problem out of sight. Every Denver job had quietly turned into a loser, and the blended number had been politely hiding it for weeks.
The real culprit was a decision nobody thought to suspect
Here’s where the story doubles back on itself.
Months earlier, well before the margins moved, the company had made a decision that looked entirely reasonable at the time. At the procurement level, they chose to standardize on a single type of racking across the entire country. One product instead of a dozen. Better pricing, simpler ordering, fewer parts to manage. In the meeting where it was decided, it was the obvious call, the kind that earns nods around the table and gets filed under “smart cost savings.”
But decisions don’t live in meetings. They live out in the field, where the rules are local and unforgiving. Denver, as it happens, requires two inspections on a job like this: one of the roof with the racking installed, and a second once the modules go up. With the newly standardized racking in the mix, every Denver project now demanded two separate truck rolls where one used to do.
A crew. A truck. The fuel. The better part of a day, on every install, in one city. The procurement team shaved a few dollars off each panel and, without ever seeing it happen, handed the field a standing bill that arrived month after month. The margin number wasn’t lying when it dropped. It was simply reporting, late, on a choice made a quarter earlier.
The 60-day trap
This is the pattern worth slowing down for, because it shows up everywhere, not just in Denver.
When you make a decision, you rarely see its effect right away. The call gets made, everyone moves on, and the consequence surfaces in the numbers thirty, forty-five, sixty days later. By then the decision feels like ancient history, settled and shelved, and your attention has drifted to whatever changed most recently. So when a bad number finally shows up, your eye goes straight to the newest thing in view. The new manager. The recent hire. Last week’s process tweak. Almost never the procurement choice from two months back, because in your mind that one is already done and filed away.
That isn’t a failure of intelligence or attention. It’s just how memory works. We weigh the recent and the visible far more heavily than the distant and the quiet, and a company that runs on gut feel and a pile of spreadsheets has no real way to trace a number backward to the decision that caused it. The lag does the damage. Recency bias picks the wrong suspect.
What actually springs the trap
You can’t make the lag go away. Decisions take time to ripple through a business and into the numbers, and no software on earth changes that. What you can change is whether the trail is still warm when you go looking.
When your data is scattered across spreadsheets and locked inside people’s heads, the investigation dead-ends at “Colorado is down.” There’s nowhere left to dig. But when every project carries its own context, the AHJ, the racking type, the crew, the equipment, the lead source, you can keep slicing the number finer and finer until the culprit has nowhere left to hide.
That’s exactly how this company found its answer. Not a hunch. Not a scapegoat. Not a reorganization. They asked “down where, precisely?” and kept asking until the answer narrowed to a single city and a single decision they’d half forgotten making.
The takeaway isn’t “keep an eye on Denver.” It’s bigger than that, and more useful. The most recent change is hardly ever the real cause, and the only thing that reliably tells you the difference is data you can cut by the dimension that matters. That kind of clarity is the whole reason a business runs on a system like SolarSuccess instead of a spreadsheet and a good memory.
Blame is fast, and it’s free, and it’s usually wrong. The truth takes a little more work to find. It’s worth every minute.
