Build vs buy: 6 metrics for a 10-50 person team

Most 10 to 50 person companies track too many numbers and decide too few things. The founder's dashboard has fourteen charts, yet the question that actually costs money — should we build this internal tool or buy it — gets settled by whoever is loudest in the room. It hits ops and finance leads hardest, because they end up paying for the subscription and the abandoned build at the same time.

Which metrics actually matter at this size?

Five numbers cover almost every build-versus-buy call a 10 to 50 person company will make in a year. Skip the rest until headcount doubles.

• Cost per seat per month, including implementation, the admin's time, and any integration fee — not just the sticker price on the vendor's page.
• Hours per month lost to workarounds, measured honestly. The spreadsheet-plus-WhatsApp layer is a cost even when nobody invoices it.
• Payback period on a build, in months, priced using your own team's hourly cost.
• Annual maintenance as a percentage of the original build cost. Budget 18-25% and you will rarely be surprised.
• Cost of switching if the choice turns out wrong in twelve months — data export, retraining, re-integration work.

If you cannot put a number on three of these, you are not ready to decide. You are ready to measure for a month.

What does a build really cost?

A build costs roughly two to three times the first invoice you approve. The quote usually covers development; it rarely covers requirements you will rewrite twice, QA, and integrations with Tally, your payment gateway or the GST portal — each of which has its own quirks. Add your own team's time: a 20-person distributor typically spends 250-350 internal hours across specification, testing and training on a mid-sized internal tool, spread over eight to twelve weeks. Budget the maintenance line separately, because a tool nobody maintains becomes a liability within a year. If you want a sense of what scoped builds actually run to in India, our pricing page lists typical ranges rather than a single headline number.

What does buying really cost?

Buying costs the subscription plus an admin tax that never appears in the proposal. In practice that means per-seat pricing that stings once you cross 30 users, 5-8% annual escalation written into most Indian SaaS contracts, 18% GST you can usually claim as input credit if you are registered, and setup fees for anything touching your ERP. The bigger cost is fit: bought tools force your process into their data model, and the workaround you invent to cope becomes the new hidden expense. A ₹900-per-seat tool at 30 seats is ₹27,000 a month before GST — real money, but predictable money, which matters when cash flow is tight. Ask for the exit cost too: what does exporting your data look like on the day you leave?

When does building actually pay back?

Building pays back when the software encodes something you cannot buy — your pricing logic, dispatch rules, or a customer workflow specific to your industry. Do the arithmetic before the emotion. A 35-person company paying ₹900 per seat for 30 seats spends about ₹3.24 lakh a year; with 7% annual escalation that is roughly ₹10.4 lakh over three years. A custom build at ₹9 lakh plus 18-25% annual maintenance comes to about ₹14-15 lakh over the same period. Buy wins on pure cost. Build only wins if seat count will grow past 60, the vendor raises prices sharply, or the tool is genuinely core to how you compete — and if it is core, the honest question is why you are renting it at all.

Which numbers tell you to stop building?

Three signals say stop: the scope has grown more than half beyond version one, only two people can safely change the code, and nobody has requested a feature in six weeks. The first means you are building a product, not a tool. The second is key-person risk — the week one of them resigns, you own a black box. The third usually means the problem was smaller than it felt. There is also a compliance layer now: anything holding customer data falls under the DPDP Act, so you need access controls, a retention rule and an audit trail whether you built it or bought it. We have written up a few of these builds in our case studies, including the ones that should have been a subscription.

Where to start

Pick one painful workflow, measure it for four weeks, and write down the five numbers above before you talk to any vendor. If the payback is beyond 30 months, buy. If the workflow is genuinely yours and the maths supports it, build small, ship in six weeks, and hold back the second phase until someone asks for it. Maysan Labs does this kind of scoping and build work for Indian SMBs, and we will tell you when the answer is a subscription.