Most software does not fail because it was bad. It fails because it was bought to solve a problem nobody had written down.
Somebody sees a demo. It looks like the future. Six months later the company is paying for it, three people use it, and the old spreadsheet is still open in another tab.
Seven rules. They are boring on purpose. Boring is what survives contact with a real business.
01
Write down the job before you look at any product
One page. What has to happen, who does it today, where it goes wrong. Do this before a single demo. Once you have seen a slick product, your requirements quietly reshape themselves around what it happens to do well.
02
Never buy from a demo
A demo is a rehearsed run on clean data by someone who built it. Your business is a messy run on messy data by someone who is busy. Ask to do the demo yourself, with your own records, or treat what you saw as an advert.
03
Test the ugliest job you have, not the tidiest
Everything handles the simple case. Pick the awkward one: the customer who pays in instalments, the deal with two owners, the job that gets cancelled halfway. That is where the real cost of a tool hides.
04
Ask what happens when you leave
Can you export everything, including notes, files and history, in a format something else can read? If the answer is vague, the price you were quoted is not the price. We have archived 3,709 call recordings for a client because a supplier was about to switch the links off. That is not a rare event.
05
Count the work, not the licence
The subscription is the small number. The real cost is moving the data in, teaching people, rebuilding reports and the weeks where two systems run side by side. Budget for that or the project will stall at eighty per cent.
06
Buy the smallest thing that does the job
Extra features are not free. Every module you switch on is another thing to configure, another place for data to go, another thing that breaks when the supplier changes it. Unused capability is a liability, not a spare tyre.
07
Run one team for one month before you commit
A real pilot, with real work, and a written note of what went wrong. Most suppliers will agree to this. The ones who will not have told you something useful about how the rest of the relationship will feel.
The cheapest tool is often the one you already own
Before adding anything, check what you are already paying for and not using. Most businesses we open up are running two tools that do the same job, and paying for both.
The same goes for data. On one job, flood, land height, soil and water information all turned out to be available free from government sources. The client had been paying for it. When a figure is not available, the report now leaves it blank rather than guessing.
What implementation really means
Implementation is not installation. It is the boring middle: deciding what each field means, agreeing who owns it, moving the old records, rebuilding the reports and turning off the thing it replaced.
That last step matters more than people expect. If the old system stays switched on, both stay half-used forever, and you now maintain two.
Set a date to turn it off before you start. Write it down. A project without an ending is a subscription.