Most of the time you should buy the product. It is cheaper, it can be running by Monday, somebody else stays up patching it, and if it does what you need then a software company telling you otherwise is selling to you rather than advising you.
The awkward part is that plenty of businesses carry on buying products long after that stopped being true for them, and nothing ever tells them. There is no alert. What tends to happen instead is that the cost quietly moves somewhere nobody is looking.
Where the line usually sits
Have a think about who spends Thursday afternoon exporting something out of one system so they can paste it into another, or rebuilding the same report every month, or ringing round for numbers that should have been in one place already. Most businesses past about 10 people can name that person immediately, and have been able to for years, and have never once put a figure on those afternoons. A day a week of somebody on £30k works out at about £6,000 a year spent having a person do the job of a database, and you will spend it again next year without anyone raising it.
Licences do something similar, only quieter, because a direct debit never feels like a decision the way an invoice does. 12 people on a £40 seat is under £500 a month and nobody is going to escalate that. The number that would actually change your mind is the 5 year one, with the 2 hires you are already planning and the price rise they will put through in year 3, and hardly anybody works it out because nothing in the calendar ever asks them to.
Then there is the case where the way you work is the reason people buy from you rather than from someone else. Generic tools are built for the average of everybody who bought them, which is exactly what you want from payroll and exactly what you do not want from the process you compete on, since over a couple of years the software will keep nudging you back towards that average and you will mostly let it, because arguing with software is tiring.
And occasionally nothing on the market fits at all, which sounds like a rare edge case until you meet a few businesses doing something a bit specific.
When to leave well alone
Accounting, payroll, email, a normal sales pipeline. Thousands of businesses want the identical thing, somebody has already built it properly, and a bespoke version would cost more and be worse. Buy it and stop thinking about it.
The genuinely difficult one is the tool that nearly fits, where the gap is a daily irritation. Irritation is a poor reason to spend £40,000, and yet it sits behind a fair amount of custom software that never earned its money back, so put hours or lost sales against the gap before anyone quotes you. If it comes out at 2 hours a month, you have just saved yourself a project.
Why the quotes come back so high
Worth knowing, if you have ever had a number back that made your eyes water.
Most of what goes into a business application has nothing to do with your business. Logins, password resets, user accounts, who is allowed to see what, an audit trail so you can prove who changed what and when, invoicing, a reporting screen somebody will export to Excel anyway. It all takes weeks, and it is near enough identical from one project to the next, so when you are quoted from a blank page a serious chunk of your money goes on rebuilding the login screen the same developer built last year for a client in an unrelated industry.
Ask what share of the estimate is groundwork and what share is the part that is genuinely yours. Anyone who has thought about it will give you a number, and anyone who has not will give you a paragraph.
The bit people forget to ask
The build is the short part, though nobody mentions that at the pitch stage, and the software will need attention for as long as you use it while your business keeps changing underneath it.
The way it goes wrong is boringly familiar. Somebody builds it, it works, they move on, and 2 years later you want one extra field on one form and there is nobody left who will safely touch the code. So ask early who hosts it, who picks up when it falls over at 3AM, and whether those are the people who wrote it. Where 3 separate companies point at each other, the job of holding it together has landed on you.
How we do it
Software development is our main service rather than a sideline, and we stay with what we build.
All that groundwork above is already written in BadgerFramework, our own codebase, where the logins, permissions, billing and reporting are tested and running in production for other clients. You start somewhere near the middle instead of at a blank page, which puts your budget on the part that is actually about your business and is why custom work with us tends to land lower than people are braced for.
We pick the tools to suit the job rather than because they are the only ones we know, and we host and look after the result, so the company that wrote your software is also the one keeping it online, patched and watched around the clock. A dedicated team stays with the system afterwards, so when you do want that extra field next year it is a conversation with people who already know how the whole thing fits together rather than an archaeology dig through somebody else's code.
The commercial side is deliberately boring. Price agreed before anything starts, changes priced before they are built, no long tie-in, and nothing to pay until you are happy with what we have built.
If you are not sure which side of the line you sit on, that is what a first conversation is for. Both founders join it, one technical and one commercial, so you find out what is possible and roughly what it takes in the same call. Now and again we tell people to go and buy something off the shelf.