Start with one process, not a platform
Big software projects fail because they try to change everything at once. A first phase that fixes one process, and is useful on its own, beats a year-long platform. Here is how to pick that process.
An owner tells us their business needs a new system: a CRM, a better ERP, a customer portal, everything connected. Sometimes that is true. Almost always, the fastest way to get there is to fix one process first and let the rest follow.
This is not caution for its own sake. It is the approach most likely to end with software people actually use.
Why big projects stall
A platform project needs every department's requirements before it can start, so it starts with months of meetings. It needs everything to be ready before anything goes live, so nobody sees a benefit for a year. And by the time it launches, the business has changed and the requirements are out of date.
Worse, the people who were supposed to use it have spent a year doing their jobs the old way, and the old way is now even more entrenched.
Why one process works
One process is small enough to understand completely. It is quick enough to ship in weeks. It is visible enough that everyone can see whether it worked. And it teaches both sides, us and you, how the other works before anyone commits to something bigger.
Most importantly, it is useful on its own. If you stop after the first phase, you still have something that works and saves time every day.
If you stop after the first phase, you still have something that works.
Four tests for choosing it
- It happens often Daily or weekly. A process that runs once a quarter will not show its value for a year.
- It is mostly retyping or chasing Moving information from one screen to another, or asking people where something is. The clearest sign software can help.
- It touches money or customers Orders, quotes, invoices, refunds, delivery promises. Improvements here show up in cash and in complaints.
- Someone owns it A person who knows how it works today and will tell us, plainly, when the new version is wrong.
What good first processes look like
In manufacturing, the strongest candidates are usually at the edges of the ERP: purchase orders arriving as PDFs and being typed in, requests for quotes waiting in a shared inbox, delay reasons kept in a spreadsheet nobody trusts, maintenance requests written on a whiteboard.
In retail, they are where channels meet: stock that disagrees between the shop and the website, returns waiting days to be refunded, customer service answering where is my order by looking it up in three places.
In each case the process is narrow, frequent and painful, and the systems on either side of it are fine.
What a poor first process looks like
Anything that needs three departments to agree on a new way of working before it can start. Anything that happens a few times a year. Anything where nobody can say what better would look like. And anything that is really a symptom of a bigger decision, such as replacing an ERP, that has not been made yet.
Then the next one
Each phase builds on the last. The order intake from phase one gives you clean order data, which makes the delivery promise in phase two possible, which makes the customer portal in phase three straightforward. The system grows into the one you would have specified at the start, minus the parts that turned out not to matter, and you have been using it the whole time.
Questions people ask
- How small should the first process be?
Small enough to ship in weeks and to measure before and after. If it needs three departments to agree on a new way of working, it is too big.
- What if the first phase is the only phase?
Then you still have something that works. That is the test of a good first process.
Not sure which process to start with? Write down the three tasks your team complains about most. One of them is almost always the right first phase.
Canonical URL: https://ugibo.com/blog/start-with-one-process