Ask a service business where its data is and it will point at the CRM. That is the data everyone has, the clean structured records that every competitor also keeps. The data that actually matters is somewhere else. It is in the quote that was won and the three that were lost, and why. It is in the exception the senior tech handled without thinking. It is in the call where the client said the real reason they nearly walked. That data is the exhaust of the work, and almost nobody is capturing it.

Why the exhaust is the moat

In a world of cheap models, the scarce input is data that cannot be copied. Anyone can buy the same model you can. Nobody else has the record of how your business actually wins, loses, and decides. That is genuinely proprietary, and in an AI world it becomes more valuable, not less, because it is the one thing a competitor’s model cannot be trained on. The moat was never the app. It is the data underneath it that only you have.

Why it goes uncaptured

It goes uncaptured because capturing it is work, and the work lives at the edges. It happens on the ute, at the point of care, on the phone, in the moment a judgement call gets made. No form catches it. The only way to capture it is to run the work on a layer that records as it goes, so the data is a by-product of doing the job rather than a second job laid on top.

Everyone keeps the records. Almost no one keeps the reasoning. The reasoning is the moat.

From exhaust to weapon

Captured, that data stops being a slide in a deck and becomes something you can aim. Pointed at the funnel, it tells the sales team which leads look like the ones that closed. Pointed at the margin, it flags the job types that quietly lose money. Pointed at the stack, it shows which tools are doing the work of one. The data is not an asset you store. It is a weapon you fire, daily, at the numbers that move the business.

Where to start

You do not start by buying a data platform. You start by picking the one workflow where the most valuable exhaust is escaping, and you run it on a layer that keeps it. The data compounds from there. Every job makes the next one sharper. That is the part that cannot be bought, only built.