Custom software vs off-the-shelf: how to choose
A simple test for whether to buy a ready-made tool or build something that fits your process exactly — and the three situations where the answer is obvious.
Every software decision gets framed as a debate about quality. It is not. Ready-made software is usually excellent — it has been refined by thousands of companies you will never meet. The question is narrower and more useful:
Which parts of your business are ordinary, and which parts are the reason customers choose you?
Buy the ordinary parts. Build the rest. Almost every good decision follows from that one sentence.
Where ready-made genuinely wins
For anything standardised, a ready-made product will beat a custom build on cost, on time and on the thousand details that only surface after a thousand companies have hit them.
Payroll. Accounting. Email. Document storage. Video calls. Basic CRM. These are solved problems, and rebuilding them is not craftsmanship — it is spending your budget on someone else's already-solved problem.
If a category has five mature competitors, that is evidence the problem is standard. Buy.
Where it stops fitting
Ready-made software encodes assumptions about how you work. Trouble starts where your reality and those assumptions diverge:
- Your process is genuinely different, and the difference is why customers
choose you.
- You run several systems that must agree, and none of them talk.
- The workaround has become the process. Someone maintains a spreadsheet that
the "real" system cannot replace.
- You are paying per seat for something most seats barely use.
None of these alone mean build. They mean measure.
The three questions
1. Is this how we compete, or just how we operate?
A logistics company's routing logic is how it competes. Its payroll is how it operates. Build the first. Buy the second.
If you cannot decide which a system is, it is almost certainly operations — the things that genuinely differentiate a business are rarely ambiguous to the people running it.
2. What is the workaround costing us?
Measure it, do not estimate it. For one week, record every time someone re-types data into a second system, fixes something the software got wrong, or keeps a side spreadsheet.
Under 30 minutes a day, the workaround is cheaper than a build. Over two hours a day, that is roughly a full salary per year, and a custom system usually pays for itself within the first.
3. What does it cost over three years, not one?
| Off-the-shelf | Custom | |
|---|---|---|
| Up-front | Low or none | High |
| Per user, per month | Yes, forever | No |
| Fits your process | You adapt | It adapts |
| Integration with your other systems | Whatever they support | Whatever you need |
| Your data | Their format | Yours |
| If the vendor changes direction | You migrate | Nothing happens |
| Maintenance | Included | Yours to budget |
Per-seat pricing is fine at ten people and a serious number at eighty. Custom has no per-seat cost but does have maintenance. Run both over three years. One year flatters ready-made; five flatters custom; three is honest.
The most expensive software is not the one with the highest price. It is the one nobody can change when the business changes.
The answer most people miss
It is rarely all or nothing. The arrangement that works most often is ready-made for the standard parts, a custom layer for the part that is actually yours, and a real integration between them.
This is only possible if the ready-made products you choose have proper APIs. Which is why "does it have a documented API, and can I see the documentation now?" belongs in your first vendor conversation, long before pricing. A product that cannot be connected to is a product you will eventually replace.
Three situations where it is obvious
Buy, without hesitation: you are under ten people, the process is standard, and you need it working this month.
Build, without hesitation: you have tried two ready-made products, both needed the same workaround, and that workaround is now someone's job.
Do neither yet: you cannot describe the process you want the software to support. No system fixes an undecided process — it just makes the indecision more expensive.
Common questions
Can we start ready-made and build later?
Yes, and it is often the sensible order — on one condition: you must be able to export your data in full, without asking permission. Confirm that before you sign, not when you leave.
How long does custom take?
Eight to sixteen weeks for a focused first version. Anyone quoting four weeks has not asked enough questions yet. See what a web app actually costs to build for how scope moves that.
We are in a specific industry — does that change things?
It moves the line rather than removing it. Restaurants are a good example, because the standard products are mature but multi-site stock is where they stop: ready-made or custom POS works through that case in detail.
Keep reading
Newsletter
Occasional notes on software, automation and running a better business. No spam.