When Small and Fast Is Wrong

There is a version of this approach that I would not trust. It is the version that hears “small and fast” and translates it as “cheap and reckless”.

That is not what I mean.

Some controls exist because somebody, somewhere, learned an expensive lesson. Security, backups, access control, audit trails and resilience are not corporate theatre when the consequence of getting them wrong is serious. The trick is learning to distinguish ceremony from control.

Ceremony is activity that looks reassuring without reducing meaningful risk. Control is protection proportionate to something that could genuinely hurt people or the business. This approach should challenge the first and respect the second.

A useful way to think about an experiment is through two questions, what happens if this goes wrong, and how easy is it to reverse? Low consequence and easy to reverse? Experiment freely. High consequence or difficult to reverse? Slow down, engineer properly and bring in the people who understand the risk better than you do.

A global point-of-sale, kitchen and inventory platform is a good example. I have spent much of my career around restaurant technology. I know enough about tills to know I do not want to invent an enterprise POS in my spare room. A till sits in the middle of payments, tax, menu control, kitchen production, reporting and the restaurant’s ability to trade. The right answer was a proper tender, mature products, serious due diligence, integration work, support and a budget proportionate to the consequence of failure. We challenged suppliers and spent proper money because the problem had earned it.

That is not the approach losing. That is the approach working.

If a small investigation gives you enough confidence to spend serious money on the right mature platform, the investigation may have saved you far more than a home-grown replacement ever could.

The phrase I now use is simple. Make the uncertainty cheap, never the safety.

Prototype cheaply. Productionise responsibly.

The little experiment has to earn the right to become important, and once it does, the responsibilities around it have to grow up as well.

The same principle applies below that scale. If an experiment begins touching personal data, influencing consequential decisions or becoming something people depend upon every day, the standard changes. You do not need to smother the first sketch in governance, but you do need to notice when the sketch has become part of the building.

That is what I mean by separating ceremony from control. I have little affection for a process that exists mainly to demonstrate that a process was followed. I have enormous respect for a control that prevents a real and understood failure. This way of working should make us more demanding about that distinction, not less responsible.