Short answer: the erp vs custom software decision is not about flexibility or scalability. It is about two numbers — how many people need a licence, and what fraction of your processes are actually different from everyone else’s. Below is the arithmetic, using published prices, so you can work out your own answer instead of trusting a pros-and-cons table.
The question almost everyone asks is the wrong one
Search this comparison and you get a dozen agency pages with the same table: ERP is faster to deploy but rigid, custom is flexible but expensive. All true, all useless. It tells you nothing about which one your business should buy, because it never touches the two variables that decide it.
Here is a better framing. Split your processes into two buckets:
- Standard processes. General ledger, accounts payable, payroll, GST filing, statutory reporting. Your version is not different from anyone else’s, and being different would be a liability, not an advantage. Never build these. Someone has already solved them, kept them compliant through a decade of regulation changes, and will keep doing so.
- Differentiated processes. The thing your customers actually pay you for. Your production scheduling logic, your pricing model, your quality workflow, the reason clients pick you over the firm down the road. Never buy these. Configuring a generic system to imitate your advantage is how you spend two years arriving at what your competitors already have.
Almost every company has both. So the real question is not which one you pick. It is where the seam between them goes, and how much each side costs.
What each actually costs
Published prices, for a 25-user company, over five years. Every figure is either sourced or a stated assumption.
| Option | Licence, 5 years | Implementation | 5-year total |
|---|---|---|---|
| Dynamics 365 Business Central Essentials | ₹99.8 lakh ($105,736) | ₹29.9 lakh ($31,721) | ~$137,500 |
| Odoo Standard | $37,350 | ~$20,000 | ~$57,400 |
| Custom build | — | $36,000 build | ~$79,500 |
Licence prices come from the vendors: Microsoft lists Business Central Essentials at ₹6,655 per user per month for India on annual billing, GST extra, and Odoo lists Standard at $24.90 per user per month on yearly billing.
The rest are assumptions, stated so you can change them. ERP implementation is taken at 1.5x the first-year licence, which is a common mid-range for a straightforward deployment and can run far higher with heavy customisation. Odoo implementation is a flat $20,000 estimate. The custom figure uses 2,000 engineering hours at a blended $18 per hour, plus 17.5 percent of build cost per year for maintenance and $200 a month for hosting — the same model derived from published salary data in our piece on custom software development cost in India.
The number that decides it is your headcount
Look at the shape of those costs rather than the totals, because that is where the actual decision rule lives.
ERP cost scales with people. Custom cost scales with scope. Add ten employees to an ERP and you add ten licences, every month, forever. Add ten employees to a custom system and you add nothing — the build is already paid for.
That gives you a break-even you can compute in one line. Over five years, a single Business Central Essentials licence costs about $4,229. A single Odoo Standard licence costs about $1,494. Against a custom system with a five-year total of roughly $79,500:
- Versus Business Central, custom breaks even at about 19 users.
- Versus Odoo, custom breaks even at about 53 users.
Below those thresholds the ERP is cheaper. Above them, the licences alone exceed what building the thing would have cost. This is why the honest answer to erp vs custom software changes with company size and nothing else changes as sharply — and why advice from a 10-person consultancy and a 500-person manufacturer will contradict each other while both being correct.
What that arithmetic does not capture
Cost is the easy half, and treating the break-even as the whole answer would be a mistake. The ERP is not just a licence. It is a general ledger that already balances, a tax engine that already files, an audit trail auditors already recognise, a permissions model, a reporting layer, and roughly two decades of edge cases someone else already hit.
Building all of that yourself is not a $36,000 project. It is a decade-long product company, and it is not the business you are in.
So the break-even number is not telling you to build an ERP. It is telling you that once you pass roughly 20 to 50 users, paying per seat for the parts of the system that are genuinely yours starts to look expensive — which points at the seam rather than at a winner.
Where the seam goes
The pattern that works for most mid-size firms:
- ERP as the system of record. Finance, procurement, inventory, HR, compliance. Buy it, configure it lightly, and resist the urge to customise. Every customisation is a bill you pay again at every upgrade.
- Custom as the system of differentiation. The 10 to 20 percent of your operation that is genuinely yours. Build it as a separate application that reads from and writes to the ERP through its API.
The discipline that makes this work is refusing to customise the ERP. Teams get into trouble by bending the standard system to fit a differentiated process, which produces the worst of both: ERP licence costs, custom development costs, and an upgrade path that breaks every year. Keep differentiated logic outside the ERP, and the ERP stays cheap to own.
A scored decision matrix for erp vs custom software
Score each row 1 to 5, where 1 means strongly ERP and 5 means strongly custom. Total it.
| Question | 1 (ERP) | 5 (Custom) |
|---|---|---|
| How many users need access? | Under 20 | Over 60 |
| How different is your core process from your industry’s norm? | Not at all | It is the reason clients choose you |
| How stable are your requirements? | Settled for years | Changing every quarter |
| Do you have in-house technical ownership? | None | A team who will own it |
| How urgent is go-live? | Weeks | Six months is fine |
| What is your compliance burden? | Heavy and standard | Light or unusual |
| Do you need to integrate with systems you do not control? | Many, all standard | Few, or all yours |
Under 17: buy the ERP and do not customise it. Over 28: build, and buy accounting off the shelf. Between 17 and 28 — which is where most companies land — you are a hybrid, and the work is deciding exactly where the seam sits, not which side wins.
Frequently asked questions
Can we start with an ERP and build custom later?
Yes, and it is usually the right sequence. Buy the system of record first, run it for two quarters, and let the friction tell you which processes genuinely need custom work. Building first, on assumptions, is how you get an expensive system nobody uses.
What about customising the ERP instead of building separately?
Light configuration is fine. Deep customisation is the trap: you pay ERP licences and development costs together, and you pay the development cost again at every major upgrade. If a process needs real logic of its own, build it outside the ERP and integrate through the API.
Is a custom system harder to hire for?
Usually the opposite. Hiring a general web developer is easier and cheaper than hiring a certified consultant for a specific ERP. The real risk is not hiring, it is documentation — an undocumented custom system is genuinely hard to inherit.
How long does each take to go live?
A lightly configured ERP is typically weeks to a few months. A custom system of comparable scope is typically four to nine months. If you need something running this quarter, that alone may settle it.
Working out your own number
Take the break-even model above, substitute your real headcount and the licence price you have actually been quoted, and you will know within an hour whether you are on the buy side or the build side — before any vendor tells you. For the related decision one step earlier in the process, see our comparison of custom software versus SaaS.
Ethersofts builds custom software and AI systems from Mohali, Punjab, and integrates them with the ERPs our clients already run. If you want this arithmetic run against your actual headcount and process map, see how we work and send us the brief — we will tell you honestly if the answer is to buy.