Build vs buy software: how to choose, and when not to build

Custom software vs off the shelf

Neither column wins. Buying is faster and safer for common jobs. Building is better for the rare job that sets your business apart.

Five questions to ask before you build

1. Have you tried the best tool for this? If you haven't used one for a month, you don't know what's missing yet.
2. Is the missing part how you earn? A slightly awkward invoice screen is annoying. A quote process that loses deals costs real money.
3. Are you working around the tool? Spreadsheets next to the system, or data typed in twice, are the clearest sign the tool doesn't fit.
4. Is the process stable? If your steps changed twice this year, wait. Software locks in the way you work today.
5. Who looks after it? Someone has to fix it when a connected tool changes. Have that answer before you start.

Three or more clear answers pointing to build means it's worth getting a price. Otherwise, buy and check again in a year.

When you should not build

Don't build in these cases, even if someone offers you a good price.

• Email, calendars, chat and file storage. These are solved problems, and they're cheap.
• Accounting and payroll. Tax rules change, and a good product keeps up. We'd connect to yours and leave it in place.
• A standard CRM for a small sales team. If HubSpot or Salesforce fits how you sell, use it.
• A simple online store. Start on a store builder, and see what customers do first.
• A process you've never run by hand. Do it manually until you know the steps.
• A tool one person wants and nobody else will open. Name the daily users first.
• When the saving is smaller than the cost. If it saves two hours a month, a $3,000 build takes years to pay back.
• When you're copying a competitor's feature without knowing if customers use it.

Saying no to a build costs nothing. Fixing or abandoning a system nobody uses costs far more than the tool would have.

What custom software costs

Custom software development cost depends on size. Clutch puts the average software project on its site at about $132,480, which includes very large builds. Ours are fixed prices: one tool or website from $1,500, a business system from $3,000 to $10,000, a full platform from $10,000. Care is from $1,000 a year once it's live, and new features start from $500, quoted first. For more on what moves those numbers, see our app development cost guide.

Per-seat pricing is where the comparison gets interesting. Here's what a 10-person sales team would pay over three years, using prices checked on 3 November 2026.

Read that table carefully. Salesforce and HubSpot do far more than a small custom system, and you get the updates for free. If your team uses those features, they're worth the money. The custom number wins only if you need a narrower thing that fits your steps exactly. A $25 tool may well be your best answer.

The middle path: buy the core, build the edge

It's rarely all or nothing. Most good setups buy the common parts and build the one piece that's different. Keep your accounting product, and add a tool that feeds it. Keep your CRM, and build the quote step it can't do. For freight forwarders, we build tools around CargoWise and leave it in place.

Automation sits here too. Joining two products often takes a small piece of work, and sometimes only a no-code tool. Our automation examples show what that looks like.

When building paid off: four real cases

• ATCG India's sales team lived in spreadsheets and Word templates, and quotes went cold. A custom sales system now blocks new quotes until old ones are followed up. The client asked to keep that lock exactly as it is.
• Gayatri Industries sells steel pipes by weight and by the piece, by grade, size and gauge, with Hindi labels for staff. Its stock and money owed sat in a ledger book and a calculator. ₹2.4 crore (about $280,000) of sales went through the system from April to October 2026.
• Pluzo's loyalty data sat in an old checkout system, an old vendor's CRM and a monthly spreadsheet. 4,400+ guests moved into one system, and 600+ members kept every point.
• Digidukan's shopkeepers ask for parts the way they talk at the counter. So we built voice search in Hindi and English into its own shop app. Monthly sales are up 8x since January.

All four had the same shape: the business depended on a rule, a language or a way of selling that it couldn't give up.

If you do build

Protect yourself the same way you would on any large purchase.

• Get a fixed price and a go-live date in writing before work starts.
• See a working part on your own data before you pay for the whole.
• Agree who owns it. With us, code, hosting and domain are in your name.
• Ask what the yearly running cost is, and what it covers.