Buy when a SaaS product fits at least 80 percent of the process and the process is not what makes you different. Customize the ERP only at configuration level, when the data must live there. Build custom when the process is the business, the integrations are the hard part, or licensing at your seat count exceeds ownership over five years. A core platform costs $200,000 to $600,000 (€175,000 to €520,000, £150,000 to £445,000) to build plus 15 to 25 percent per year to run; SaaS at 300 seats often passes that by year four.
Prices are in US dollars. Euro and pound figures in parentheses are approximate, rounded conversions. Quotes are issued in the currency of your contract.
Build versus buy is the question every enterprise software project starts with, and most organizations answer it with a year-one comparison that makes buying look cheap and building look reckless. The honest comparison runs five years and includes the costs nobody puts on the slide: the workarounds, the per-seat growth, the integration layer, the customizations that break on every vendor upgrade. This post gives you that comparison for the three real options, custom software, a SaaS product, and customization of the ERP you already have, and ends with a rule for deciding. It is one chapter of our buyer's guide to enterprise software development.
What are the three options, honestly described?
- Buy SaaS. Subscribe to a product that does most of the job, configure it, and integrate it with your other systems. Fast to start, predictable in year one, and you are renting a process the vendor designed for their median customer.
- Customize the ERP. Extend the system of record you already run, through its configuration layer, its scripting environment or its partner ecosystem. Keeps data in one place, keeps your ERP partner employed, and every customization is a liability at the next major upgrade.
- Build custom. Design and build a system around your process, integrated with the ERP and the rest. The highest year-one cost, the lowest per-user cost at scale, and you own the roadmap, the data model and the code. Who owns the code explains what ownership has to mean in the contract.
What does each cost over five years?
The table models a core operational system for 300 users, with four integrations and moderate compliance needs. Your numbers will differ, but the shape of the comparison holds.
| Cost line | SaaS | ERP customization | Custom build |
|---|---|---|---|
| Year-one setup | $30,000 to $80,000 (€26,000 to €70,000, £22,000 to £60,000) implementation and configuration | $100,000 to $300,000 (€85,000 to €260,000, £75,000 to £220,000) partner customization | $200,000 to $600,000 (€175,000 to €520,000, £150,000 to £445,000) build including discovery and migration |
| Licenses, five years | $300,000 to $900,000 (€260,000 to €785,000, £220,000 to £665,000) at $200 to $600 per year (€175 to €520 per year, £150 to £445 per year) per seat, rising with headcount and tier | $50,000 to $150,000 (€44,000 to €130,000, £37,000 to £110,000) in added ERP modules or user licenses | $0 (€0, £0) for the platform; $10,000 to $50,000 per year (€8,500 to €44,000 per year, £7,500 to £37,000 per year) for third-party components |
| Integration | $60,000 to $200,000 (€50,000 to €175,000, £44,000 to £150,000) for a middleware layer, plus rework when the vendor changes their API | Low initially; rises as the ERP becomes the hub for everything | Included in the build; $8,000 to $25,000 (€7,000 to €22,000, £6,000 to £18,000) per additional integration after |
| Workarounds and gaps | The 20 percent that does not fit: spreadsheets, manual steps, a second tool, lost time | Processes bent to fit the ERP's model; slow screens, staff workarounds | None by design; the gaps are features in the backlog |
| Upgrades and maintenance | Included, but each vendor release can change your configuration | Each major ERP upgrade retests every customization: $30,000 to $100,000 (€26,000 to €85,000, £22,000 to £75,000) per upgrade cycle | 15 to 25 percent of build per year for support and maintenance |
| Five-year total, indicative | $450,000 to $1,300,000 (€390,000 to €1,130,000, £335,000 to £960,000) | $350,000 to $900,000 (€305,000 to €785,000, £260,000 to £665,000) | $400,000 to $1,200,000 (€350,000 to €1,045,000, £295,000 to £890,000) |
Read the totals as overlapping ranges, because they are. The point is that SaaS is not cheap at 300 seats over five years, ERP customization has a cost that arrives at every upgrade, and custom is front-loaded but flat. The year-one comparison hides all three facts. How much enterprise software development costs gives the build figures in detail.
What are the hidden costs of each option?
SaaS
- Seat growth. Per-user pricing means the system gets more expensive as the company succeeds. Tier jumps, where a needed feature sits in the enterprise tier, can double the bill.
- The 20 percent. Every process has a part the product does not handle. That part lives in email, spreadsheets and people's memory, and it is where errors happen.
- Data access. Your data is in their database. Reporting, migration and exit all depend on their export tools and their API limits.
- Roadmap risk. The vendor is acquired, the product is sunset, pricing changes. You are a line in their revenue, not a stakeholder in their plan.
ERP customization
- Upgrade drag. Each customization must be retested and often rewritten at a major upgrade. Organizations end up years behind on their ERP version because the customizations make upgrading too expensive.
- Partner lock-in. The customizations are written in the ERP's own tools by a partner who knows them. Changing partner means rediscovering everything.
- User experience. ERP interfaces are built for finance and operations power users. Field staff, customers and occasional users struggle, and adoption suffers. User adoption for internal software is where this bill is paid.
Custom build
- Discipline. Custom software needs an owner, a backlog and a budget every year. Without them it becomes the legacy system of 2032. What is technical debt describes how that happens.
- Time to first value. Three to twelve months before users touch it, against weeks for SaaS. Phasing mitigates this but does not remove it.
- Vendor quality. The outcome depends on the team. The wrong team produces a system that costs as much as SaaS and works worse. How to choose an enterprise software development company is the guard against that.
When does each option win?
| Choose | When | Examples |
|---|---|---|
| SaaS | The process is standard across your industry, the product fits 80 percent or more, seat count is modest, and the process is not how you compete | HR, payroll, CRM for a conventional sales team, help desk, expense management, email and documents |
| ERP customization | The process is close to standard, the data must live in the ERP, the customization is small and configuration-level rather than code, and you have a long-term ERP partner | An approval workflow, an additional field set, a report pack, a country-specific tax rule |
| Custom build | The process is the business, the SaaS fit is under 70 percent, the integrations are the hard part, many users are external or occasional, or licensing at your seat count exceeds build cost by year three | Order and fulfilment logic specific to your products, customer and partner portals, scheduling and dispatch, pricing engines, anything the ERP shows as a workaround |
What about the hybrid?
Most mature organizations end up with all three, and the useful question is which option for which layer. Standard back-office processes run on SaaS and the ERP. The processes that differentiate the company run on custom systems that integrate with both. The integration layer, which is where the three meet, is designed once rather than grown. Enterprise application integration covers how to design it so the custom system does not become a second ERP.
What is the decision rule?
- Score the fit. List the process steps. Mark each one as handled, partially handled or not handled by the best SaaS candidate. Under 70 percent handled is a build signal.
- Price five years, not one. Licenses at projected headcount, integration, upgrade cycles, support, and a cost for the gaps. Use the table above as a template.
- Ask whether the process is strategic. If a competitor could buy the same SaaS and get the same process, it is not. If the process is why customers choose you, renting it from a vendor is a risk, not a saving.
- Check the exit. For SaaS, how do you get the data out and what does it cost. For ERP customization, what happens at the next major version. For custom, who owns the code and could another team maintain it.
- Phase whichever you choose. A custom build starts with one workflow. A SaaS rollout starts with one team. The ERP customization starts with configuration before code.
Rule of thumb
Buy the processes that every company in your industry runs the same way. Customize the ERP only at the configuration level. Build the processes that make customers choose you, and build them around the ERP rather than inside it.
How 7L approaches build versus buy
7L International is a custom software development and digital transformation firm with hubs in New York, London and Athens, and we have told clients to buy more often than you might expect from a firm that builds. Our discovery phase includes the fit scoring and the five-year model above, and ends in a written recommendation that is sometimes a SaaS product plus an integration layer rather than a build. When the answer is build, we deliver end-to-end, from design through launch and support, and we design the integration with the ERP first so the new system stays a system of engagement rather than a second system of record. If you are weighing the options for a specific process, describe it to us and we will give you a straight answer on which column it sits in.
Frequently asked questions
Is custom software always more expensive than SaaS?
In year one, almost always. Over five years, often not. A SaaS product at 300 seats costs $300,000 to $900,000 (€260,000 to €785,000, £220,000 to £665,000) in licenses alone over five years, before integration and the cost of the parts it does not handle. A custom core platform costs $200,000 to $600,000 (€175,000 to €520,000, £150,000 to £445,000) to build and 15 to 25 percent per year to run. The crossover for a mid-sized operational system is usually between year three and year five.
What fit percentage justifies building instead of buying?
If the best SaaS candidate handles under 70 percent of the process steps you actually run, build is usually the better five-year answer. Between 70 and 85 percent, buy and accept the gaps, or buy and build a small custom layer around it. Above 85 percent, buy and configure. Score the steps, not the feature list.
Should we customize our ERP instead of building a separate system?
Only at the configuration level: fields, workflows, reports and rules the ERP supports natively. Code-level customization inside the ERP is retested and often rewritten at every major upgrade, which is why so many organizations run versions years out of date. Processes that need code are better built as separate systems that integrate with the ERP through its API.
How do we avoid a custom system becoming legacy?
Give it an owner, a yearly budget of 15 to 25 percent of the build for maintenance, and a backlog that is reviewed quarterly. Keep the technology choices boring and documented, keep the code and the cloud accounts in your name, and make sure a second team could take it over. Systems become legacy when nobody is responsible for them, not because they are custom.
Can we start with SaaS and build later?
Yes, and it is often the right sequence. Run the process on SaaS for a year, learn what the gaps actually are, then build around or replace it with a clear scope. The risk is data: make sure the SaaS contract gives you a full export in a usable format, or the migration later will cost more than it should.