Drawn by Hand

I keep coming back to a passage in James Barr's A Line in the Sand: two diplomats in 1916, a map of the Ottoman Empire, and a line drawn with a pencil that created nations. The line was arbitrary. Everything that followed from it was not. I grew up on one side of that line, and what stays with me isn't the history. It's the pattern. How something drawn by hand, by people making reasonable judgments with incomplete information, becomes the thing that determines identity.

Finance has the same pattern. The boundary between investment grade and junk (BBB- on one side, BB+ on the other) is a single notch that determines which funds can hold your debt, what your cost of capital is, who you are. And Tech has one too, though nobody talks about it that way: the COGS line on the P&L. Above it, you're a software company. Below it, you're a services business that happens to bill monthly. That distinction is a 3-5x difference in revenue multiple at exit. And that line, like Sykes-Picot, was drawn by hand.

The interesting question isn't that favorable interpretation exists. Ambiguous rules plus human judgment guarantee it. The interesting question is why favorable interpretation is the only kind that occurs.

Each individual classification, examined on its own terms, holds up. The VP of Engineering reclassifies one engineer to R&D: reasonable, and in France, where social charges add 45% on top of base salary, that single move saves 120,000 euros in fully loaded cost from the margin calculation. Better still, under R&D the costs become eligible for the CIR, a 30% tax credit. The VP of Customer Success keeps a borderline CSM below the line: reasonable, because the role is primarily adoption. The CFO capitalizes development costs per IFRS guidance: reasonable, and it's what the auditors approved. The board doesn't question a margin that matches the benchmark: reasonable, because why would they question a number that confirms the thesis they funded?

Fifteen individually reasonable choices, each made by a different person for different reasons at different times. Every one defensible. The aggregate is a margin that is fifteen steps from reality, and every step was correct.

There's no meeting where someone says "let's inflate the margin." There's no memo. There's no conspiracy. There are just fifteen people, each doing their job, each choosing the defensible answer that happens to point in the same direction. The system doesn't produce liars. It produces optimists with good arguments.

Every other metric in the Tech stack inherits whatever distortion lives in gross margin. If the margin is off by ten points, a company moves from "efficient growth" to "barely viable" without a single customer changing their behavior. Nothing happened in the business. Something happened on the line.

And the benchmarks themselves are built on the same inflated data. Companies aim for the median, the median is built on favorable classifications, so the actual economics required to reach the median are worse than the median implies. A collective fiction resting on individual fictions, all of them defensible.

The due diligence moment is where accumulated reasonableness finally meets an external perspective. The CSM team goes fully into COGS. The DevOps split gets recalculated. The capitalized R&D gets unwound. The margin drops from 82% to 68%. At a 10x revenue multiple on 5 million euros of ARR, that's the difference between a 50 million and a 34 million euro valuation. Sixteen million euros of value that never existed.

But the investor's reclassification is also a judgment call. The investor wants the conservative number, because their job is to not overpay. Their reclassification reflects their incentives just as the company's original classification reflected the company's. The truth is probably somewhere between 82% and 68%, and the fact that there is no methodology that can pin it down is not a flaw in the analysis. It's the whole point.

The margins that get scrutinized are the ones that look bad. The ones that should get scrutinized are the ones that look perfect: 82%, right at benchmark, clean and uncomplicated. Because a perfect gross margin is almost always the product of fifteen reasonable choices that all happened to point in the same direction. Sykes and Picot thought they were being reasonable too. Accumulated reasonableness is the most durable form of distortion, because no single decision can be reversed without admitting it was wrong. And none of them were wrong.