Fix the brakes before you fit the new engine

Walk into almost any organisation that is struggling and you will find the same two corners of the same building.

In one corner, people are copying figures between spreadsheets because two systems that should talk to each other never have. A report takes twenty minutes to run, so nobody runs it more than once a week. Someone performs a manual reconciliation every Thursday that exists because of a decision made in 2011 by a person who left in 2014. Everybody complains about all of this. Nobody is authorised to fix any of it.

In the other corner, a recently arrived executive is presenting a slide deck about a two million pound programme that will change everything, and technical debt will finally become a thing of the past.

I have watched this pairing so many times it has stopped being funny, though it took a while. It is fitting an electric motor to a 1960s classic while the brakes do not work, the wiring is held together with tape and optimism, and nobody has looked at the suspension since the Callaghan government.

Transformation has to be transformational, apparently#

Somewhere along the line we decided that improvement only counts if it is theatrical. Capital T, press release, launch event, enough acronyms to fill a white paper. A programme that fixes eleven irritating things and saves four hundred hours a month does not get a launch event, because how would you photograph it?

So the small fixes stay in the backlog, where they have the peculiar status of being universally acknowledged and permanently unfunded. Meanwhile eighteen months and several million pounds go into a platform replacement which, after the inevitable overrun, delivers a system that still does not talk to finance properly and leaves the original operational headaches exactly where they were. Different login screen, same Thursday reconciliation.

I want to be fair about why this happens, because "executives are vain" is a lazy answer and mostly wrong. A big programme is legible. It has a name, a budget line, a steering committee and a date, all of which make it possible to be seen to be doing something. Forty small improvements are administratively invisible, even when they add up to more. The incentive is not stupidity, it is the need to point at something.

Follow the money and it makes sense#

There is no margin in automating your invoice processing. There is enormous margin in a two year programme that requires the seller to remain present for all of it, scoped against requirements everyone privately knows are incomplete, with the gaps harvested later as billable change.

I am not damning the individuals here. They have targets and shareholders, they were not running a charity, and most of them believe in what they are selling. But the model rewards scale and duration rather than resolution, and a model that rewards duration will produce projects that last. It could hardly do otherwise.

Which means the buyer has to supply the discipline the market will not. Nobody is coming to suggest a smaller engagement.

The economics nobody puts on a slide#

Set the two approaches side by side and the numbers are not close.

Picture a manufacturer whose production team transcribes data between three systems by hand because the ERP replacement, now eighteen months into planning, will make it all unnecessary. Automating that transcription is not a hard problem. It is a few weeks of work, it costs less than a month of the programme's consultancy fees, and the people doing that transcription get their Tuesdays back immediately. Meanwhile the ERP programme continues, undisturbed, towards whatever it eventually becomes.

The point is not that the big programme was wrong. Sometimes the engine really does need replacing. The point is that eighteen months of avoidable misery was treated as an acceptable cost of waiting, and nobody put that on a slide, because the people bearing it were not in the room where the slides were made.

Incremental work has a property that big-bang work structurally cannot have: the benefit arrives before the money runs out. You find out within weeks whether you were right. If you were wrong, you were wrong cheaply, and you learned something you can use on the next one. Every improvement pays for the next and builds the confidence to attempt something slightly harder.

Big programmes invert this entirely. You find out at the end, having spent everything, at the precise moment when admitting failure is most expensive and least likely.

What this asks of you#

The awkward part is that this approach requires knowing your own operation properly, and that is genuinely harder than buying a platform. You have to sit with the people doing the work and find out what actually eats their week, which means hearing things about decisions your predecessors made. You have to sequence the improvements so they compound rather than collide. You have to resist the urge to make it into a programme, because the moment it becomes a programme it acquires a steering committee and starts behaving like the thing you were avoiding.

And you have to accept that this will never look as impressive as the alternative. There is no launch. There is a Thursday when someone notices the reconciliation has quietly stopped being necessary.

I find that deeply satisfying, and I accept that I may need to get out more.

The most beautiful car in the world is useless if the engine turns over and the brakes do not bite. Fix the brakes, drive it while you plan the rest, and let somebody else hold the launch event.