JM178: patterns in the heat

The terrace was nearly empty by 9:30.

A lazy breeze pushed the Cluj summer heat sideways. The plates had been cleared twenty minutes ago. The ash in Paul’s cigarette bent but didn’t fall.

Gabor was quiet.

Then Paul said:

“We lost that insurance deal. The big one.”

Gabor nodded. No need to ask which.

“We worked three months on it. Six calls, strategy workshop, the budget was confirmed. We had good signs all the way. Then: ‘you’re too expensive.’ That’s it.”

“Too expensive, full stop?”

“Said both our price level and pricing model weren’t a fit. But we’ve used this exact setup before. Margin structure, blended rate bands, even threw in platform integration at cost.”

He paused and flicked the ash.

“So I’ve spent the last four days trying to reverse-engineer what went wrong. Did we over-signal? Did they anchor us too high? Was the ‘partner vibe’ off?”

“Or maybe they were never serious,” Gabor offered.

Paul gave him a sharp look.

“You’re not helping.”

Gabor smiled.

“You ever heard of apophenia?”

Paul frowned.

“The tendency to see patterns where there aren’t any,” Gabor continued. “It’s why people see faces in trees. Or think the stock market has moods.”

“And that’s what you think I’m doing?”

“I think it’s what we all do. Especially after we lose something we thought we understood.”

He poured the last of his wine into Paul’s glass.

“There’s a sub-type called clustering illusion. You see two or three similar client behaviors, and assume you’ve discovered a trend. ‘They all hate outcome-based pricing.’ ‘They all want local teams.’ It feels true. But it’s not statistically sound.”

Paul didn’t answer. He was listening now.

“Another one’s called illusory correlation. You hear feedback once or twice, let’s say, a founder tells you hourly billing makes them nervous, and suddenly your whole deck is reworked to preempt that. You believe there’s a link. But there isn’t.”

“So we’re just guessing?”

“No. But we’re telling ourselves stories. And then calling those stories strategy.”

“Sounds depressing.”

“It’s worse if you believe the data. Like Monte Carlo fallacy. Thinking just because you lost two in a row, the next one will surely land. Or vice versa. We start reacting to streaks. But each deal’s a different coin toss.”

Paul exhaled hard.

“But what’s the alternative? Not learning anything?”

“No. You still learn. But not from noise. You learn from variance. From tagging context. From logging why a proposal was shaped a certain way. Not just whether it closed.”

“We don’t have enough data for that.”

“Exactly. Which is why most pricing decisions in IT services are vulnerable to emotional overfitting. We think we’re optimizing, but we’re rationalizing.”

Paul took a sip of the wine. Then another.

“So what do we do instead?”

“We get more disciplined. We separate feedback from narrative. We stop believing every ‘too expensive’ and start logging who said it, when, in what format, and whether the deal had real intent. And we stop trying to force symmetry. Not every loss needs a reason.”

The cicadas were loud now. Streetlights flickered across the square.

Paul stubbed out the cigarette.

“You’re right. I wanted the loss to mean something.”

“It does. Just not what you think.”

They didn’t talk much after that. The bill came. No argument.

They split it, exactly in half.

PS.

I write this newsletter for fun. 

Fictional characters, imagined dialogues, but grounded in real conversations.

All based on what I see every day in my work on pricing for IT services:

> Building Agentic AI workflows for pricing in IT & software services companies (1,000+ employees)

> Ongoing fractional pricing & revenue strategy work with several tech companies

> Delivering pricing workshops and sprints

> Writing a book about the impact of AI on the business of selling and delivering IT services

> and pricing webinars