Custom software vs SaaS for Italian SMEs: 7 criteria to decide

The right answer to “custom software or SaaS” is SaaS in 99% of cases. We say it as a company that builds custom software, and you might find that biased, but it is the truth. That remaining 1%, though, is where the future of a company is decided. The real question is not “custom or SaaS”. It is “are we in the 1% or in the 99%”. In this article we try to give you a framework to answer honestly.

TL;DR

• Standard SaaS is the right choice for the vast majority of Italian SMEs: predictable costs, outsourced maintenance, fast time-to-market.
• Custom becomes the right choice when the software supports a process that really is your competitive advantage, when existing systems do not integrate with SaaS, when volumes exceed SaaS tiers, or when specific compliance requirements are not covered.
• Seven concrete criteria to decide: process differentiation, scale, integration, compliance, 5-year TCO, evolution speed, tolerable lock-in.
• In 50% of real cases the answer is “both”: custom on 1-2 core processes, SaaS on everything else. It is often the healthiest architecture.

Why 99% of SMEs should choose SaaS

Five substantial reasons, not trade-show slogans.

Predictable costs in the budget. A SaaS license is negotiated, signed, renewed every year. Custom has high upfront costs, variable maintenance costs, technical debt that grows over time. For an SME doing financial planning on a 12-24 month horizon, the predictability of SaaS is worth real money.

Free automatic updates. SaaS evolves. Every 3-6 months new features come out that you did not pay for explicitly. With custom, every new feature is a deliberate development (and budget) decision. SaaS gives you a constant free upgrade; custom requires a team to keep up.

Outsourced maintenance. Security, patching, scalability, disaster recovery: these are jobs the SaaS vendor does with scale and specialization that an Italian SME cannot afford in-house. You pay the subscription and stop thinking about it.

A user community that drives discovery. If you use a SaaS with thousands of users, features land in your lap that you would never have thought to ask for, because somebody else asked for them first. With custom, you have to think of every possible improvement yourself.

Scalable pricing. You start small (few licenses), you grow (more licenses), or you shrink. Custom is almost always an irreversible commitment: the money spent in the first phase is not recovered if you change your mind.

When an SME comes to us asking “we want a custom business management system” and after 30 minutes of conversation we realize they could use TeamSystem, Zucchetti, ESA, Odoo, Microsoft Dynamics or an equivalent SaaS, we say so. Our job is not to sell ourselves at all costs: it is to tell you what you actually need.

The 1% case: when custom is the right answer

Four real situations where SaaS is not enough.

When your processes are your competitive advantage. If what you do is objectively different from your industry, and that difference is the reason clients choose you, then supporting it with standard software means eroding it. A manufacturing company with a unique product configuration process, a professional services firm with a distinguishing methodology, a distributor with dynamic pricing rules that no standard replicates: these are 1% cases.

When integration with existing systems is impossible via SaaS. You have an AS/400 running production, a proprietary MES on the line, a vertical quality control system. Standard SaaS does not integrate: either it requires custom bridges so expensive you might as well build the software custom, or it accepts a nightly batch sync that does not solve real-time problems.

When specific compliance rules exclude standard SaaS. Healthcare with patient data on Italian territory, finance with data custody constraints, public administration with AgID requirements and national cloud. Compliant SaaS solutions exist, but they are few and expensive, and sometimes they simply do not cover your use case. Custom on-premise or on Italian sovereign cloud becomes the only option.

When volumes are out of scale for SaaS tiers. SaaS products are designed for the market curve: the middle of the distribution. If you are very large, SaaS prices explode (cost-per-user that becomes prohibitive at 500 users), or technical limits (API rate limits, transaction throughput, storage) cut you out. Custom becomes the cheaper option again.

If none of these four cases describes you, you are in the 99%. Go with SaaS.

The 7 concrete criteria to decide

Seven questions in descending order of importance. Each has an implicit weight: if the answer leans toward custom, it counts more. If it leans toward SaaS, the same.

1. Process differentiation

Are your processes identical to those of your industry, or are they your competitive advantage? Concrete test: try describing to an external consultant from your industry how you handle orders, invoicing, warehouse. If they nod and say “like everyone else”, you are probably in SaaS-land. If they stop you every two minutes to ask “but why do you do it that way” and you answer “because it works”, you probably have a specificity worth preserving.

2. Current scale and 3-5 year projection

Current volumes and projected growth. Under 50 users SaaS is almost always cheaper in the long run. Between 50 and 500 users it depends on the vendor’s pricing model. Above 500 users custom often becomes cost-effective again. For transactions, above one million transactions a month you risk falling out of standard SaaS tiers.

3. Integration with existing systems

How many systems need to talk to the new software, and what level of maturity they have. Modern systems with REST APIs: SaaS integrates. Legacy systems without modern APIs (old AS/400, vertical Italian on-premise business management systems, custom ERPs from the 2000s): integration becomes the dominant constraint. Often custom is mandatory or, alternatively, modernization of legacy systems has to be tackled first.

4. Specific compliance

Baseline GDPR is covered by almost all serious SaaS products. Advanced compliance is not. NIS2 with sector-specific requirements, AgID, the specific Italian Privacy Code, healthcare (FSE, DPCM 178/2015), finance (Bank of Italy/Consob regulations), national cloud: each additional layer reduces the number of suitable SaaS products. If you are in a heavily regulated sector, start from the compliance analysis and see what remains.

5. 5-year TCO

Total cost of ownership over 5 years, not just licenses. Include: licenses, implementation, training, maintenance, customization, decommissioning and final migration. With custom you pay much more in the first 2 years (initial development) but the curve flattens. With SaaS the first 2 years cost less but TCO grows linearly. Between year 4 and year 5 the curves often cross for medium-to-large scenarios.

6. Required evolution speed

How fast do features need to change? SaaS evolves at the vendor’s pace: 2-4 releases a year, rarely tailored to your case. Custom evolves at the pace of your development team: weekly if needed. If your business changes often (e.g. a startup in market discovery mode, sectors with fast-changing regulation), custom gives flexibility that SaaS cannot replicate.

7. Tolerable lock-in

What happens if the SaaS vendor shuts down, gets acquired and changes pricing, or deprecates features you use intensively? The most well-known recent example is Heroku for its free tier services deprecated in 2022. For business-critical software, lock-in can be a strategic risk. Custom gives total ownership, but also total responsibility.

Direct comparison table

Dimension	Custom	SaaS
Upfront cost	High (50-500k+)	Low (10-300 euro/user/month)
5-year cost	Stable after the investment	Linearly growing
Time-to-market	6-18 months	1-4 weeks
Customization	Total	Limited to the vendor’s plug-ins
Maintenance	Yours	The vendor’s
Scalability	To be designed	Included in the service
Lock-in	Technological (chosen stack)	On the single vendor
Data ownership	Yours	Shared with vendor
Compliance	Tailored configuration	Limited to the vendor’s
Legacy integration	Always possible	Possible if connectors exist
Evolvability	Your team’s speed	The vendor’s speed
Exit cost	High if abandoned	Medium (data export)

An honest reading of the table: every row rewards a trade-off. Custom rewards ownership, customization, evolvability. SaaS rewards speed, upfront cost, outsourced maintenance. There is no winner. There is a choice that depends on what weighs more in your context.

The mistake every SME that chose custom (but should not have) makes

The pattern repeats itself. An SME with 80-150 employees decides “we need a business management system built specifically for us” because the SaaS they tried (TeamSystem, Zucchetti, or even just Odoo) “does not do exactly what we want”. They commission a custom build from a vendor who accepts enthusiastically. They spend 200-400 thousand euro over 18 months. They go live.

In the next 3 years this is what happens: the custom software needs maintenance, but the original vendor has lost key people or raised their rates. Every small change costs 5-15 thousand euro. The software ages in its dependencies (framework, libraries, runtime) and nobody wants to upgrade because “it works”. After 5 years the software is obsolete, expensive to maintain, and the company faces the choice of rebuilding it from scratch or going back to SaaS.

The honest diagnosis in retrospect: the rejected SaaS would have covered 85% of the needs. The company went custom for the 15% of customization it believed was non-negotiable. It was negotiable.

To avoid this: before committing to custom, run a rigorous exercise. List the 20 features you need most. For each one, check which SaaS products cover them natively. If 17 out of 20 are covered, custom is a mistake. The 3 missing ones can be handled with a mix of process (adaptation), light integrations, or accepting a small compromise.

The mistake every SME that chose SaaS (but should not have) makes

The symmetric mistake. An SME with unique processes gets convinced that “SaaS is configurable enough anyway”. They buy SAP Business One, Microsoft Dynamics, an enterprise edition of TeamSystem. They invest in implementation (50-150 thousand euro), train the staff, go live.

Six months later it emerges that 30% of the business processes do not fit in the SaaS. The workarounds begin: parallel Excel sheets, double data entry, manual processes. The staff works against the software instead of with it. The software has been “implemented” but is not really being used for what it was bought for. They are back to the fragmented data from before SaaS, only with an extra license cost on the balance sheet.

The honest diagnosis in retrospect: the differentiating processes, that 30% that does not fit in the SaaS, were in fact the company’s added value. They needed to be supported with dedicated software, not sacrificed to a generic product.

To avoid this: before committing to SaaS, run the inverse exercise. List the 5 most important processes for your competitive advantage. For each one, ask the vendor: how exactly is this done in your software? If the answer is “you do it with a workaround”, “it requires customization” or “you do not really do it”, be careful: you are signing a contract that will force the company to change its own processes to fit inside the tool.

Hybrid cases: custom + SaaS together

In most real situations, the real answer is not binary. It is a hybrid architecture.

The sensible split is almost always this one: SaaS for the commodity, custom for the differentiator. Invoicing, payroll, accounting, HR: SaaS, because they are processes standardized by law or universal convention. The software that supports your specific competitive process: custom, because there is no alternative.

In between sits integration, which is the real technical work. Having a SaaS for invoicing and a custom system for order management means making them talk: APIs, data sync, exception handling when one of the two is down. A well-built integration costs 10-30 thousand euro per serious handshake. It is worth it on 2-4 integration points. Above that, you enter a zone where maybe the hybrid is not the optimal choice.

The hybrid architecture has another advantage: risk is distributed. If the invoicing SaaS changes painfully, you replace it without touching the custom system. If the custom system becomes obsolete, you rebuild it without throwing away the SaaS. The blast radius of every change is contained.

FAQ

When is it right to choose custom software?

When the software supports a process that is your competitive advantage, when integrations with existing systems are critical and impossible with SaaS, when volumes are out of scale for standard SaaS, or when you have specific compliance requirements not covered by off-the-shelf products. If none of these conditions apply, choose SaaS. The healthy default is SaaS, custom is the motivated exception.

How much more does custom cost compared to SaaS over 5 years?

Typically 3-5 times more in the first 2 years, then the curves converge between year 4 and 5. For high-scale scenarios (over 500 users) custom becomes cheaper than SaaS after year 3. Under 50 users SaaS almost always stays cheaper in the long run. The TCO calculation has to be done over 5 years, not 1: annual comparisons are misleading.

Can I switch from SaaS to custom later?

Yes, but with significant migration cost: typically 50-300 thousand euro for medium-sized companies, of which 30-40% goes into data migration and the rest into building the new system. It is always preferable to make the right choice from the start. The reverse migration (from custom to SaaS) is even more expensive because critical custom features are often lost and need to be handled with workarounds in the SaaS.

How heavy is SaaS lock-in?

It depends on the vendor and the market. Vendors with direct competitors (e.g. CRM where Salesforce, HubSpot, Pipedrive, Zoho all exist) are lower risk: you can migrate to a competitor. Niche vendors without alternatives (e.g. vertical SaaS for specific regulated sectors) are high lock-in risk. Always check the ability to fully export your data (standard formats, documented APIs) before signing multi-year contracts.

Is it worth running a custom POC before committing?

Often yes. A custom POC on a narrow use case (3-6 weeks of work, 8-20 thousand euro) lets you validate two things: that the company’s vision is technically feasible, and that the chosen development team is the right one. Skipping the POC and going straight to the full project worth hundreds of thousands of euros is the classic recipe for failed custom projects.

Conclusion: there is no universal answer, there is your case

Custom software or SaaS is not an ideological choice. It is an analytical exercise on the seven criteria we have seen, applied to your specific reality. The companies that do it well fall into two groups: those that honestly conclude “we are in the 99%, let us go SaaS” (and they do well), and those that conclude “we really are in the 1%, we need custom” (and they do well too, because they entered consciously, not by inertia).

If you are evaluating custom or SaaS and want an honest assessment of your case, we can help you make one. Regardless of where the decision takes you: if the answer is SaaS, we tell you without a pitch. Let’s talk.

To go deeper on specific dimensions: the pillar page security-aware custom software for regulated cases, the page custom ERP replacement for those already in the middle of the problem, and the page Excel to custom software conversion for those stuck in the spreadsheet-as-de-facto-ERP limbo.