Insights · 5 min read
Most companies do not need a full ERP from scratch. They need a few modules that fit their workflow. Here is the build, buy, and extend math, and where ERP projects go wrong.
By GGP Editorial
Most companies do not need a custom ERP built from scratch. Most need a few modules that fit their actual workflow, connected to the tools they already use. The gap is real: off-the-shelf ERPs assume a standard process, and no growing business runs a standard process for long.
I want to be straight about the trade-offs here, because ERP projects go wrong more often than they should, and most failures come from scope, not from code.
An ERP touches the parts of a business that keep the lights on: orders, inventory, purchasing, accounting, HR, and reporting. Building one from zero is a large project, so most of what we ship is closer to a core system with integrations than a monolithic suite.
The typical build includes a few of these:
You rarely need all of it. The useful skill is knowing which modules matter and which ones you would pay for but never open.
Integration matters more than the modules themselves. A custom ERP that cannot talk to your payment processor, your bank feeds, or the CRM your sales team already likes becomes an island, and islands get abandoned. Plan the connections first and the screens second.
The decision usually comes down to three paths.
Buying suits companies whose process matches the software. If your workflow is close to standard, paying for a mature ERP and adapting your process is cheaper and faster than writing one.
Extending suits companies that already run an ERP but have one or two painful gaps. We have added custom modules, an extra reporting layer, and mobile apps that field staff actually use on top of systems clients already owned.
Building suits companies with a workflow the market does not serve: unusual pricing rules, specific compliance, or multi-currency settlement across regions. That is where a custom system earns its cost back.
| Path | Best when | Rough cost |
|---|---|---|
| Buy | Standard workflow, need it now | License plus setup |
| Extend | Existing ERP, a few gaps | $15,000 to $50,000 |
| Build | Unique workflow or compliance | $50,000 to $300,000 and up |
The build range is wide because a five-module system for one country is a different job from a multi-entity, multi-currency system.
The two variables that move the price most are the number of entities and the number of currencies. A single-entity, single-currency system is a contained job. Add subsidiaries in two countries with different tax rules and the integration and testing work roughly doubles.
Scope is the usual culprit. A team lists every feature it has ever wanted, and the project turns into a two-year science experiment. We push clients to cut to the modules that change daily operations and leave the rest for later phases.
Data migration is second. Moving years of messy spreadsheets into clean tables takes longer than anyone expects, and it is the line item nobody budgets for.
Adoption is third. A system the staff refuses to use is a very expensive filing cabinet. The build has to include training and a feedback loop, not just a go-live date.
We have built financial management systems, property rental systems, and a CRM layer that several clients run daily, and the same pattern repeats: start narrow, ship something real, then grow it. One client's financial management system replaced a set of linked spreadsheets that three people maintained. The first phase was only invoicing and expense tracking, live in six weeks. Later phases added payroll and reporting. The stack is mostly Java, Spring Boot, MySQL, and React, with Go and Python where the job calls for it.
We do not limit ourselves to one industry. If it is a legal business, we can build for it. A Brazilian billboard platform, a South African lifestyle app, and a multi-market brokerage are all running on code we wrote, and they have little in common beyond needing software that does not get in the way.
For clients outside China we run overlapping hours and a dedicated chat group, in English or Portuguese, so the daily decisions around an ERP rollout do not stall on a timezone. A deployment that needs sign-off at 9 am in Sao Paulo gets an answer the same morning.
Start with the two or three modules that would change your month, and resist the urge to boil the ocean.
Tell us what you are building and where you are today. We typically reply within 24 hours.