Most comparisons of custom software and SaaS stop at “monthly subscription versus initial development cost.” This simplification leads thousands of companies to make decisions worth hundreds of thousands of euros using an incomplete spreadsheet. According to the 2025 Zylo SaaS Management Index, organizations waste an average of $21 million a year on unused SaaS licenses—52.7% of all licenses purchased. On the other side, a poorly scoped custom development project can overrun both its budget and schedule.
The truth lies in a calculation few decision-makers follow through to completion: five-year total cost of ownership (TCO), including visible costs, hidden costs, opportunity costs, and strategic risks.
This article provides a complete method for making that comparison honestly, with numerical data and decision frameworks you can apply to your circumstances.
TL;DR — SaaS costs less upfront, but its TCO often exceeds custom software after three years, especially when price increases of 10–20% a year according to Gartner, vendor lock-in, and unused licenses are included. A rigorous five-year TCO calculation covering seven cost categories frequently reverses the initial conclusion.
Why the Usual Calculation Is Flawed from the Start
The Upfront-Cost Bias
When a CIO or CFO compares a €500 monthly SaaS subscription with a €40,000 custom development quote, the conclusion seems obvious. SaaS appears six to seven times cheaper in the first year. This reasoning suffers from a documented cognitive bias: aversion to a high upfront cost.
The problem is that the calculation compares a cash flow—the subscription—with an investment—development. They follow different accounting rules and economic dynamics. The subscription is OPEX that creates no balance-sheet asset. Development is amortizable CAPEX that builds a software asset base.
The Seven Cost Categories Nobody Puts in the Spreadsheet
A complete TCO must include seven cost families that quick comparisons systematically omit:
- Acquisition costs: purchase price, initial subscription, setup fees
- Integration costs: connectors, APIs, middleware, gateway development
- Customization costs: adapting to business processes, advanced configuration
- Training costs: initial onboarding, ongoing training, internal documentation
- Maintenance costs: updates, fixes, functional enhancements
- Poor-fit costs: compensating manual processes, workarounds, duplicate entry
- Exit costs: data migration, team retraining, contractual penalties
Ignoring just one category distorts the result by 15–30%. Ignoring all of them, as most comparisons do, can create a gap of up to 200% between perceived and actual cost.
The Real TCO of SaaS: Beyond the Monthly Subscription
Built-In Price Escalation
SaaS prices are not fixed. According to Gartner, several major SaaS vendors raised prices by 10–20% in 2024, well above average IT budget growth of 2.8%. This trend is structural rather than temporary.
Vendors justify increases by pointing to AI features, restructured offerings, and regional adjustments. For a company with 50 users paying €50 per user per month, a 12% annual increase turns a €30,000 first-year bill into €52,940 in year five, without adding a single user.
Over five years, the company will spend €190,170 on subscriptions alone, compared with €150,000 if pricing stayed stable. The €40,170 difference is the cost of believing the price is fixed.
The Sinkhole of Unused Licenses
Zylo's 2025 report finds that 52.7% of purchased SaaS licenses are never used. For large companies, that represents an average annual waste of $21 million. Even at a 100-person SME, a 30% waste rate on a €200,000 annual SaaS portfolio means €60,000 thrown away each year.
Several factors drive this: employees leave without their licenses being canceled, tools adopted for a project are later abandoned, and departments take out duplicate subscriptions. The average company uses 419 SaaS applications, according to Zylo. Tool proliferation makes tracking virtually impossible without a dedicated management solution.
The Invisible Cost of a Poor Fit
A general-purpose SaaS tool covers an average of 70–80% of a company's business needs. The remaining 20–30% generates costs rarely counted: manual entry, exports and imports between tools, workarounds, and parallel Excel spreadsheets.
Consider a manufacturing SME using a standard SaaS CRM for technical quotations. The CRM handles neither product bills of materials nor automatic margin calculations by component. Two salespeople consequently spend 45 minutes each day reentering data in a spreadsheet. Over a year, that represents 375 hours of skilled work, or approximately €18,750 in fully loaded labor costs, invisible in the IT budget.
The Real TCO of Custom Software: Beyond the Initial Quote
The Initial Investment: A Real but Amortizable Cost
Custom development requires a significant upfront investment. French market data puts typical SME business applications between €15,000 and €150,000, while complex enterprise projects can exceed €500,000.
That amount can be daunting. Yet in accounting terms, software developed internally or by a provider is an asset that can be capitalized and amortized over three to five years. A €60,000 investment amortized over five years creates an annual expense of €12,000, to compare with a much higher cumulative SaaS subscription over the same period.
Maintenance Costs: An Often-Underestimated Item
Custom software maintenance averages 6–10% of the initial cost annually. For a €60,000 project, that means €3,600–€6,000 a year. It covers bug fixes, security updates, and compatibility with new operating system and browser versions.
Decision-makers discovering this after delivery often consider it high. But compared with an equivalent SaaS subscription exceeding €30,000 annually, the ratio still favors custom software, provided maintenance was budgeted from the outset.
The Risk of Provider Dependence
Custom software has its own dependency risk: reliance on the provider that developed it. If that provider disappears or becomes unavailable, maintenance can become problematic.
Three measures manage this risk: rigorous technical documentation, clean source code following industry standards, and a clear contractual intellectual property clause. Well-built custom software using open-source technologies—React, Node.js, Python, PostgreSQL—can be taken over by any competent developer. Proprietary SaaS cannot be taken over by anyone.
TCO Methodology: The Five-Year Calculation Framework
A Seven-Dimension Comparison Framework
To compare SaaS and custom software TCO honestly, apply this framework over five years. Put a number against every row, even an approximate one.
| Cost category | SaaS (to calculate) | Custom software (to calculate) |
|---|---|---|
| Acquisition | Subscription × 60 months, with 10–15% annual increases | Initial development cost |
| Integration | Connectors, APIs, middleware, often charged extra | Integrated into development |
| Customization | Limited configuration plus possible specific developments | Included by definition |
| Training | Initial training plus adaptation to vendor updates | Initial training only; the tool does not change without an internal decision |
| Maintenance | Included in the subscription, but imposed rather than chosen | 6–10% of initial cost per year |
| Poor fit | Hours lost to workarounds × fully loaded hourly cost | Close to zero with a well-written requirements specification |
| Exit | Data migration plus contractual penalties and retraining | Zero cost: you own the code and data |
Worked Example: A Business Application for an SME with 40 Users
Consider an SME with 40 employees needing a field service management tool for scheduling, tracking, and customer reporting.
| Cost item | Specialized SaaS (5 years) | Custom software (5 years) |
|---|---|---|
| Acquisition / subscription | €156,000 (€65/user/month, 12% annual increase) | €55,000 (initial development) |
| ERP + CRM integration | €15,000 | €8,000 (integrated into development) |
| Customization | €12,000 (add-on modules) | €0 (included) |
| Training | €8,000 (initial plus vendor updates) | €4,000 (initial only) |
| Maintenance | €0 (included in subscription) | €22,000 (€4,400/year) |
| Poor fit | €25,000 (1 hour/day × 2 people × 250 days × 5 years × €10) | €3,000 (minor adjustments) |
| Estimated exit cost | €20,000 (migration plus penalties) | €0 |
| TOTAL FIVE-YEAR TCO | €236,000 | €92,000 |
In this example, SaaS TCO is 2.6 times higher than custom software over five years. The crossover point—when cumulative SaaS costs exceed the custom investment—generally falls between months 18 and 30, depending on user count and per-license pricing.
Vendor Lock-In: The Strategic Cost the Spreadsheet Misses
A Quantifiable Risk That Is Rarely Quantified
According to an NE Digital study, 74% of organizations undergoing digital transformation experienced vendor lock-in difficulties, particularly in ERP and CRM environments. The figure reflects a concrete reality: changing SaaS platforms is expensive, takes time, and introduces operational risks.
Vendor lock-in operates at three levels:
Data level: Your data is stored in a proprietary format. Exporting it completely, including its history and relationships, is often technically complex and sometimes contractually restricted.
Process level: Your teams have adapted working processes to the tool. A switch requires reorganization extending well beyond technical migration.
Financial level: Multiyear contracts, automatic renewal clauses, and early termination penalties create deliberately designed exit barriers.
The Cost of Vendor Dependence
A SaaS vendor can unilaterally change its interface, remove a feature, raise prices, or alter its terms of use. In 2024, several private-equity-owned vendors imposed price increases of up to 900% on certain products. Customers had no choice but to pay or migrate, both expensive options.
With custom software, you own the source code, database, and infrastructure. No third party can change the rules. This technological sovereignty has real financial value, even if it appears in no accounting line.
When SaaS Remains the Better Choice—and We Should Acknowledge It
Use Cases Where SaaS Clearly Wins
An honest comparison must recognize situations where SaaS is objectively better in both TCO and suitability:
Standardized, non-differentiating functions: accounting, email, videoconferencing, file storage. These functions are the same across businesses. Developing a custom accounting tool would be absurd. Solutions such as Pennylane, Google Workspace, or Microsoft 365 offer an unbeatable functionality-to-price ratio.
Very small teams with fewer than ten users: with few users, SaaS subscription costs stay moderate and the TCO crossover moves beyond five to seven years. Custom software becomes hard to justify economically.
An immediate start is needed: if time to market is critical and business processes are still stabilizing, SaaS lets you start within days. Custom software requires prior scoping.
Areas with frequent regulatory change: payroll, electronic invoicing, tax compliance. Specialized SaaS vendors absorb regulatory complexity and constant updates.
The Decision Matrix
| Criterion | SaaS advantage | Custom software advantage |
|---|---|---|
| Standard process | Yes | — |
| Specific business process | — | Yes |
| Fewer than 10 users | Yes | — |
| More than 30 users | — | Yes |
| Immediate need (< 1 month) | Yes | — |
| Expected use > 3 years | — | Yes |
| Complex IT integration | — | Yes |
| Limited upfront budget | Yes | — |
| Data sovereignty | — | Yes |
| Competitive advantage linked to the tool | — | Yes |
The Most Common Reasoning Errors
Error 1: Comparing a Monthly Subscription with a Total Quote
This is the most common trap. A €2,000 monthly subscription looks trivial beside a €60,000 quote. But €2,000 × 60 months = €120,000, before price increases. Always convert both options into cumulative costs over the same period.
Error 2: Forgetting Additional Users
SaaS charges per user. Every hire increases the bill. Custom software costs no more with 40 or 80 users, aside from server infrastructure whose additional cost is marginal. For a growing business, this difference is strategic.
Error 3: Overestimating Custom Software Maintenance
Annual maintenance of 6–10% of initial cost is a market average. A well-architected application using mature frameworks and stable technologies can fall to 3–5%. Conversely, a poorly designed application can reach 15–20%. Initial development quality determines long-term maintenance costs.
Error 4: Ignoring Opportunity Cost
A general-purpose SaaS covering 80% of requirements forces teams to compensate for the missing 20%. Time spent on workarounds is time unavailable for value-adding activities. Over five years, a five-person team losing 30 minutes daily incurs an opportunity cost of €156,250, valuing an hour at €50.
Error 5: Not Discounting Cash Flows
A euro today is not worth a euro in five years. Custom development is paid immediately or over a few months, while SaaS subscriptions are spread out. Applying a 5% discount rate reduces the present-value cost of SaaS by approximately 8% compared with the raw subscription total. This favors SaaS but generally does not reverse the conclusion when the TCO gap is large.
Conduct Your Own TCO Analysis in Five Steps
Step 1: Define the Exact Functional Scope
List every required feature in three categories: essential, important, and desirable. For each, assess coverage by the SaaS under consideration. Overall coverage below 75% is a strong warning in favor of custom development.
Step 2: Project SaaS Costs over Five Years, Including Increases
Never treat today's price as a fixed baseline. Apply annual increases of 10–15%, the range Gartner observed in 2024. Include expected workforce growth. If you plan to grow from 40 to 60 employees over five years, calculate costs using 60 licenses rather than 40.
Step 3: Quantify Both Options' Hidden Costs
For SaaS: integration, customization, update training, poor fit through lost hours, and a possible exit. For custom software: annual maintenance, foreseeable functional enhancements, and hosting.
Step 4: Assess Strategic Risks
Assign monetary values to risks: vendor lock-in, using the cost of migration if the vendor closes or changes its terms; provider dependence, using the cost for another developer to take over the code; and technological obsolescence, using the cost of rebuilding over a seven-to-ten-year horizon.
Step 5: Calculate the Crossover Point
Identify the exact month when cumulative SaaS subscriptions exceed the custom investment—development plus cumulative maintenance. If that point falls before month 36, custom software is almost always the better financial choice. Between months 36 and 60, strategic criteria determine the decision. Beyond month 60, SaaS can remain suitable for standard needs.
The Market Confirms the Trend: The Macroeconomic Figures
The global custom software development market was worth $43.16 billion in 2024 and is expected to reach $146.18 billion by 2030, an annual growth rate of 22.6% (Grand View Research). This acceleration reflects a broader shift: companies with five to ten years of accumulated SaaS subscriptions are starting to calculate the real cost.
In France, SaaS accounts for more than 56% of software revenue, with a market value of $8 billion in 2024. But signs of fatigue are visible: large French companies spend an average of €4.1 million annually on SaaS to manage 419 tools. Rationalization is underway.
AI accelerates this trend in two ways. SaaS vendors raise prices to finance AI investment. Meanwhile, AI substantially reduces custom development costs and timelines, moving the TCO crossover further in custom software's favor.
FAQ
What Is the Average Cost of Custom Software for an SME? In France, a custom SME business application costs €15,000–€80,000 depending on complexity. Annual maintenance represents 6–10% of that amount. Over five years, total TCO—development plus maintenance—generally remains below cumulative equivalent SaaS subscriptions for more than 20 users.
How Many Users Make Custom Software Cost-Effective? The crossover generally occurs between 15 and 30 users, depending on the compared SaaS unit price. Beyond 30 users, with an expected usage period exceeding three years, custom software is almost always more economical in TCO terms.
How Long Does Custom Software Take to Develop? A functional MVP can be delivered in two to six weeks by an experienced team using current AI tools. A complete application requires two to four months. These timelines have fallen substantially thanks to generative AI, which accelerates development phases by 40–70%.
Is SaaS Always Less Risky than Custom Software? No. SaaS carries specific risks: unilateral price increases, feature removal, vendor closure, and vendor lock-in. Custom software gives you control of the code, data, and product evolution. Its development-quality risk is managed through provider selection and technical documentation.
How Do You Calculate Custom Software ROI? Use the formula: ROI = (Benefits generated − Total cost over the period) ÷ Total cost × 100. Benefits include hours saved, fewer errors, faster processes, and additional revenue. According to Gartner, companies implementing custom solutions report an average five-year ROI of 55%, compared with 42% for SaaS.
Can You Gradually Migrate from SaaS to Custom Software? Yes, and this is often the most prudent strategy. Start with modules where SaaS shows its limits through poor fit or excessive costs. Retain SaaS for standard functions such as email and accounting. This hybrid approach makes the transition safer while keeping the budget under control.
AI Coder Squad: Custom Software TCO Without Unpleasant Surprises
Comparing SaaS and custom software requires rigorous analysis of actual costs over time. When the result favors custom development, provider quality directly determines long-term maintenance costs—and therefore the validity of the entire TCO calculation.
AI Coder Squad designs custom applications and AI agents for companies that want to move quickly without sacrificing quality—with senior developers and an AI-powered approach.
→ Start your project and discover how AI Coder Squad can accelerate your next delivery.