INEYA Leaving SaaS
The method
Leaving SaaS without breaking what works.
Leaving a software stack is not decided by conviction, it is decided by inventory. Here is the method we apply, including the part where the conclusion is that you should stay.
Inventory first, conviction never
Most exits fail in the same place: they start from a principle instead of a list. Someone decides that everything must come back in house, starts with the most visible tool, and six weeks later there is a half-migrated mailbox and a team that has lost confidence.
The right first day is boring: twelve months of bank statements, every recurring debit tied to a tool, every tool tied to an internal owner and a renewal date. In a fifty-person company that list is almost always longer than management expected, and part of it serves nobody any more.
The reversibility test, tool by tool
Four questions. A tool that fails all four is a tool you will not leave painlessly, and it is better to know that now than on the way out.
What exactly do you get back?
Not the list of exports promised in the documentation: ask for a real export, today, and open the file. The difference between the two is the measure of your dependency.
How long does your business last without it?
An hour, a day, a week? The answer sets the switching order far better than the size of the invoice.
Who, internally, knows how it is configured?
If the answer is one person, or worse nobody, the tool already holds you, whoever publishes it.
What happens to your customers if you change it?
A tool carrying a public address, a form, or a link sent to thousands of people does not swap out like an internal one.
The three-column triage
The whole difficulty is holding the third column. An exit plan made only of replaceable tools is not a plan, it is a brochure.
| Category | Replaces well | What actually blocks |
|---|---|---|
| Sales and CRM | Yes, and it is the best first project: the gain is immediate and the scope is clear. | The automations and sequences, which do not export and are rebuilt by hand. |
| Project and task management | Yes. The data is simple and habits transfer quickly. | Integrations with the rest of the stack, usually more numerous than anyone thinks. |
| Time tracking and internal invoicing | Yes, and the gain often beats expectations, because these tools are billed to the entire headcount. | History, which you need to compare one year against the last. |
| Customer support and knowledge base | Partly. The tool replaces, the indexed public addresses far less easily. | Links already sent to customers, and the search ranking of public articles. |
| Office suite and file storage | With difficulty, and rarely first. The disruption usually outweighs the gain. | Files shared with outsiders, and real-time co-editing, which remains a genuine advantage. |
| Technically yes; humanly, the most sensitive project of all. | Sending reputation, which rebuilds slowly and breaks in a day. | |
| Video conferencing | Yes for internal use. No for external, as long as your customers use something else. | The link a customer can open without installing anything. |
| Accounting and payroll | No, and that is not a failure: it is a regulated trade that moves with the law. | Legal liability and the permanent updating of rates and thresholds. |
| Banking, payments and legal invoicing | No. These are infrastructure, not subscriptions you swap. | Licensing, banking interoperability and invoicing compliance. |
| Electronic signature | Partly, depending on the level of proof your contracts require. | Evidential value, which depends on the certification provider and not on the software. |
The order in which to switch
Order matters more than pace. One successful project funds two more; one failed project cancels them all.
Start with what costs a lot and never touches the customer
The CRM and internal delivery tools meet both criteria. You demonstrate the gain before asking for any trust.
Continue with what is billed to the whole headcount
Those lines are the most sensitive to growth: they are what explains the gap between the constant-headcount projection and the plus 30 % one.
Keep what the customer sees for last
Email, video, public addresses. You touch those once the rest runs and the team is confident, never before.
Leave regulated tools where they are
Accounting, payroll, evidential signature, payment collection. Bringing them in house saves almost nothing and exposes a great deal.
The mistakes that sink an exit
Switching with no overlap
A few weeks of running both costs a fraction of what a rushed rollback costs. The overlap is not waste, it is the insurance.
Confusing export with reversibility
A clean export guarantees nothing about automations, templates and conversation history. Those are what hurt, not the records.
Leaving three weeks before the deadline
An exit is prepared six months before renewal. Otherwise you pay for another year, and you pay for it having already started to leave.
Bringing software home with no operating plan
Software running at your place needs backups whose restore has been tested, monitoring that alerts, and automatic restart. Without that you have not gained sovereignty: you have changed risk.
Forgetting the person who knows
Every tool has a power user who configured it. If they are not in the loop from day one, they become the most effective opponent the project has.
Frequently asked questions
How do you leave a SaaS product without interrupting the business?
Start with a full inventory of recurring debits, put every tool through a reversibility test, then switch in this order: first what costs a lot and never touches the customer, then what is billed to the whole headcount, last what the customer sees. Every switch runs both systems in parallel for a few weeks.
Which SaaS tools should not be brought in house?
Accounting, payroll, payment collection and evidential signature. These are regulated trades that move with the law: bringing them home saves little and exposes a lot.
How much can you save by leaving SaaS?
It depends on the number of billed seats and on how fast you hire, never on a general percentage. The gap is computed line by line, and it widens with every hire since the price follows headcount.
Which tool should you start with?
The one that costs the most per seat and that the end customer never sees. In a professional services firm that is almost always the CRM.
The next step
The method is public. The work itself takes five days.
We apply this method to your stack: full inventory, three-year projection in both scenarios, data mapping, switching plan and its cost. €1,500 excl. VAT, deducted from the first deployment.
Read next