An owner asks me whether they should buy a platform or have something built. It's the wrong question, but it's the wrong question in a useful way, so let's take it seriously.
Here's how the decision actually works.
Buy anything that isn't yours
If a hundred thousand other businesses need the exact same thing you need, buy it. Don't think about it, don't get creative, just buy the boring option and move on.
That covers: accounting, payroll, email, calendars, payments, phones, file storage, e-signatures. These are solved. Somebody has spent a decade and a hundred million dollars making the software you'd build in a weekend, and theirs is better, cheaper, and won't break when you're on vacation.
The instinct to build here is almost always ego or false economy. You are not going to out-engineer a payments company. Give them the two percent.
Build the part where you're different
Every business has one or two things it does in a way nobody else does. Your quoting logic. Your dispatch rules. The specific way you qualify a lead, or the sequence you use to onboard a customer that took you eight years to get right.
That's the part you can't buy, and the part where off-the-shelf software makes you worse, because it will quietly bend you into doing things the way it wants. That's fine for accounting. It's not fine for the thing that makes you money.
So the rule is:
Buy the commodity. Build the differentiator. And be honest with yourself about which is which.
Most owners get this exactly backwards. They buy a giant platform that promises to run their whole business, then spend six months in the settings menu trying to make it do their thing, and end up with a worse version of their thing plus a subscription.
The third option nobody talks about
Here's what actually happens in most successful small businesses, and nobody calls it anything, so I'll name it: buy the parts, build the seams.
You buy the phone system. You buy the CRM. You buy the accounting software. And then you build a thin layer of custom logic that connects them, encodes your rules, and does the specific things your business needs that no vendor will ever build for you.
That layer is small. It's often a few hundred lines of logic. It does things like:
- When a call goes unanswered, create a lead, text the caller, and put a task on the right person's list.
- When a quote's been out five days with no response, send a follow-up, unless they've replied, unless it's over a certain amount in which case tell me to call.
- When a job closes, wait three hours, then request a review through whichever channel we have for them.
- Every Monday, tell me the five customers who used to order and haven't recently.
None of that is a product. All of it is your business. And it's cheap to build precisely because it sits on top of things you already bought.
This is the option people miss. They think it's "buy a platform" or "build a whole system from scratch," and both of those are bad. The right answer is usually five percent custom and ninety-five percent stuff you already pay for.
When "build" is genuinely wrong
Don't build if:
Nobody owns it. Custom software with no owner is a time bomb. If you build it and the person who understands it leaves, you now have a system nobody can change and everybody depends on. That's worse than a mediocre vendor.
It changes constantly. Tax rules, compliance, payment regulations. Let a vendor eat that maintenance burden. That's what you're paying them for and it's the best deal in software.
The requirement is fuzzy. If you can't write down what it should do, building it will cost three times the estimate and you'll hate the result. Go use a cheap tool badly for two months first. You'll learn what you actually need, and that knowledge is worth more than the two months.
You're building it to avoid a subscription. The build costs more than five years of the subscription, every time. Build for capability, never for cost.
When "buy" is genuinely wrong
Don't buy if:
It forces you to change how you make money. Software that says "we recommend restructuring your process to fit our model" is fine for a workflow you don't care about and disastrous for the one that's your edge.
You'll only use eight percent of it. Big platforms are priced for people who use all of it. If you need one feature, you're subsidizing forty you'll never touch, and you're carrying the complexity of all of them.
The data is trapped. Ask, before you sign: can I export everything, in a usable format, whenever I want? If the answer is fuzzy, walk. Your customer list is the only asset in the business that compounds. Never put it somewhere you can't get it out of.
The arithmetic, honestly
People want a formula. Here's the closest thing.
Take the process. Estimate the hours per month it currently costs you in labor, and the money it currently loses in leaks. Multiply by twelve. That's the annual bleed.
Now price the two options.
Buy: subscription, times twelve, plus the setup time, plus a realistic guess at how much of your process you'll have to bend.
Build: the one-time cost, plus maybe fifteen percent a year in maintenance, because things change and something always needs adjusting.
If the annual bleed is less than either number, do neither. Leave it alone and go find a bigger problem. This is a legitimate outcome and it's the one nobody picks, because a spreadsheet is less fun than shopping.
If the bleed is large and the process is generic, buy.
If the bleed is large and the process is yours, build the seam. It's usually the cheapest of the three and nobody offers it to you, because there's no recurring revenue in it for a vendor.
The question to ask yourself first
Before any of this: is the process even right?
Automating a bad process gets you a bad process at scale. Buying software for a bad process gets you a bad process with a monthly bill. I've watched people spend real money digitizing a workflow that should simply have been deleted.
So the honest order is: fix the process, then decide whether to buy, build, or leave it alone. Most people skip step one, because software feels like progress and thinking doesn't.
If you're staring at a decision like this and want a straight opinion from someone with no product to sell you, that's most of my job. Tell me what you're trying to fix and I'll tell you honestly whether it needs software at all. Reach out here.