Nobody reads the offboarding documentation on the way in. You evaluate features, pricing, maybe security certifications — and the day you finally leave, years later, is the day you discover what “your data” meant in the contract all along.
None of what follows is an accusation. Major vendors publish real export tools and honour them. But the details decide whether leaving is an afternoon or a quarter, and the details are worth knowing while you’re calm.
Exports exist. Complete exports are rarer.
Google Takeout, Microsoft’s admin export tools, Dropbox’s folder archives — these are real and they work. The question is what arrives in the archive.
What travels well: raw files, mail messages (usually as mbox or PST), contacts and calendars in standard formats. If it has an open format, you’ll probably get it.
What usually doesn’t: the structure around the content. Sharing permissions, version histories, comment threads, form responses wired into sheets, app scripts, integrations. Documents in a vendor’s native format convert on the way out — usually well, not always perfectly, and at volume “not always” is a real number.
A useful mental model: exports give you your content; they rarely give you your system. The system — who can see what, what connects to what, what happened when — is what you rebuild on the other side.
Retention windows are shorter than memory
When a subscription ends, your data doesn’t sit waiting indefinitely. Vendors retain data from closed accounts for a limited window and then purge it — details vary by vendor, plan, and how the account ended, and policies change over time, which is exactly why the clause is worth reading for your stack rather than trusting a summary.
Two patterns worth knowing. First, the export has to happen before or shortly after cancellation — the window is measured in days or weeks, not quarters. Second, “we’ll just downgrade to the free tier” is not an exit plan: free tiers have storage caps and feature limits that can lock the account into a read-mostly state, with the data effectively stranded above the cap.
The migration is a project, not a download
Even with a clean export in hand, moving in somewhere new is real work.
Mail imports are the easy part; folder hierarchies, aliases, and shared mailboxes take planning. Files need their permission structures rebuilt around new accounts and groups. And every integration — every form, webhook, and automation that pointed at the old system — has to be found and re-pointed, including the ones nobody remembers building.
This isn’t an argument against leaving. It’s an argument for understanding the exit while it’s hypothetical, because the cost only grows — and a cost you’ve never measured quietly becomes a reason to stay.
Audit your own exposure: a checklist
An afternoon of work, worth doing even if you never leave anything:
- Inventory the tools. Every service holding company data, who administers it, and what it costs. The list is always longer than expected.
- Find each export. Locate the actual export function — not the marketing page about it. Note the formats it produces.
- Test one for real. Run a full export of one core tool. Open it somewhere else. Could you actually work from this? Put a quarterly reminder on it.
- Read the retention clause. For each critical tool: what happens to data after cancellation, and how long do you have?
- Map the integrations. For each tool, list what breaks the day it disappears. This is your real switching cost, and it’s invisible until written down.
- Name an owner. One person accountable for keeping these answers current. Unowned checklists decay.
File the results next to your insurance documents — same genre.
Whether you ever leave a vendor is a business decision, and staying is often the right one. But being able to leave is an engineering property, not a mood. You either have it or you don’t — and it’s decided long before the day you need it.