Should you replace SaaS with open source? A CFO's full-cost test

The question is usually asked at renewal time: the subscription has become expensive, so should we replace it with open source?

It is the wrong question if it begins and ends with license price.

Open-source software can remove a per-seat meter, give a company more control over its data, and fit a stable workflow very well. It can also create a new operating responsibility: hosting, backups, access control, upgrades, integrations, monitoring, support and a plan for when the person who understands it is unavailable.

For a CFO or COO, the useful comparison is not subscription versus free software. It is the full cost of doing the same work, at an acceptable level of reliability, over a defined period.

Start with a decision, not a product list

Before comparing tools, define the workflow in plain language.

For example: “The sales team needs to record client conversations, route qualified opportunities, report pipeline by owner, and retain the account history after a team member leaves.”

That is the job. A product name is only one way of doing it.

An apparent replacement often conceals a scope change. A low-cost alternative may cover the core workflow but not the reporting, permissions, mobile access, integrations, audit trail or support model that the team actually relies on. Conversely, a subscription may be paying for features nobody uses simply because the plan was chosen when headcount or needs were different.

The first decision is therefore one of 5:

• Keep the current tool because its demonstrated value and operating reliability justify the cost.
• Reduce the number of paid seats or move to a lower tier.
• Renegotiate the contract, terms or commitment period.
• Consolidate overlapping tools before replacing either one.
• Replace only when the full operating case is stronger than the current arrangement.

Open source belongs in the last option, not automatically at the beginning of the discussion.

The 5 costs that belong in the comparison

1. Current recurring cost

Use the actual renewal quote, not the list price. Include licenses, paid add-ons, storage, premium support, integration fees and the spend on adjacent tools bought to fill capability gaps.

Then separate active users from paid seats. A tool with 70 licenses and 42 regular users may have a seat-management problem before it has a replacement problem.

2. Replacement operating cost

Open-source software is not free to operate. A realistic plan includes hosting, backups, security updates, monitoring, domain and email services where relevant, support, and the accountable person or provider who runs it.

If a managed provider will operate the service, use its written proposal. If the internal team will run it, price the time honestly, including on-call responsibility and leave cover. A zero line item here is usually a missing cost, not a saving.

3. Migration and change cost

Count more than data export and import. The migration may require data cleanup, field mapping, role and permission design, template recreation, testing, parallel running, user training and revised operating procedures.

The financial question is not whether these activities are unpleasant. It is whether their one-time cost is recovered inside the decision horizon.

4. Reliability, security and compliance risk

The replacement must meet the requirements of the work it supports. That may include uptime expectations, restore testing, access reviews, encryption, retention, audit history, vendor or customer security requirements, and integration failure handling.

Self-hosting does not automatically make a company compliant with any framework. Compliance comes from the controls, evidence and operating practices around the system. If the team cannot own those controls, retaining a well-run SaaS product can be the lower-risk financial decision.

5. Exit and option value

Ask what happens if the workflow changes in 18 months. Can data be exported? Can the company change provider? Is the implementation documented? Could another engineer or managed operator take over without rebuilding it?

This is not abstract insurance. Undocumented customization and one-person dependency are costs that appear later, usually during a renewal, acquisition, outage or staff departure.

Use a 3-year comparison, not a monthly-price comparison

A simple full-cost model is:

• 3-year replacement cost = one-time migration cost + (3 × annual operating cost) + expected risk and support allowance.
• 3-year current cost = (3 × annual current recurring cost) + known renewal or add-on increases.

The allowance for risk should not be invented to force an answer. Use known support commitments, reasonable contingency for work that has not been scoped, and a clear record of what remains uncertain.

An illustrative example

Consider a 30-person business with a stable internal workflow. Its current subscription and necessary add-ons cost $12,000 a year. A managed open-source alternative is estimated at $3,600 a year, with a one-time migration and training cost of $4,000.

Before any contingency for unresolved risk, the 3-year current cost is $36,000 (3 × $12,000). The replacement cost is $14,800 ($4,000 + 3 × $3,600).

The apparent 3-year difference is $21,200. That is a reason to investigate, not a reason to sign off. The decision still depends on whether the replacement meets the actual workflow, reliability and security requirements, and whether the migration estimate survives a technical review.

If the alternative needs $12,000 of custom development, a dedicated internal owner, or a second product to restore a capability the team uses every day, the result changes. That is why the worksheet matters more than the headline.

When keeping SaaS is the financially correct answer

Keeping the current product is often rational when:

• The workflow changes frequently and the SaaS vendor absorbs the product-development burden.
• The product provides genuinely used specialist capability that would be costly or risky to reproduce.
• Contractual, security or reliability requirements cannot be met confidently by the proposed operating model.
• The team has no durable owner for the alternative.
• The migration would interrupt a revenue-critical process during a period when the business cannot absorb that risk.
• The real issue is unused seats, an overpowered plan or an unexamined renewal, not the product itself.

“Keep” should be recorded as a decision with evidence, not treated as inertia. The same is true of “replace.”

A 30-day way to test the case

Do not begin with a migration project. Begin with evidence.

1. Export the last 12 months of invoices, seat counts and renewal dates.
2. Ask the system owner for active-user and feature-usage data.
3. Write the 5 to 10 functions the workflow must perform, including the exceptions that cause operational trouble.
4. Obtain a scoped operating proposal for any alternative, including support and recovery responsibilities.
5. Price migration as a bounded project with named assumptions, rather than as a guess.
6. Put both options in the same 3-year worksheet and review the remaining risks with the accountable business owner.

At the end of that exercise, a company may keep the SaaS tool, reduce its spend, or make a well-scoped change. All 3 are useful outcomes. The bad outcome is treating a subscription invoice, or an open-source download, as the entire cost of the work.

FoundMargin’s 3-week audit makes these decisions from invoices, administrator usage data and the way work actually moves between systems. It identifies evidence-backed savings opportunities and distinguishes them from ideas that need more validation.