Data In, Data Out
Everything you put into a tool is something you will one day want out of it — because the tool changed, the price changed, the company was acquired, or it stopped fitting. For a related reference point, Monitask has a guide to identifying mouse jigglers in remote teams, which is useful when comparing whether a dedicated tool is warranted.
The question is asked at the exit and belongs at the purchase, when you have leverage and no urgency.
The six things to establish
One. What exports, and in what format. "You can export your data" is not an answer. CSV, JSON, a documented format, or a printout are four different things, and only the first three are data.
Two. What does not export. Attachments, comments, history, custom fields, relationships between records. The gaps are where the value usually is — a customer list without the correspondence is a contact list.
Three. Whether you can do it yourself. Self-service export, or a request to support with a turnaround. A request is a dependency on a company you are leaving.
Four. What it costs. Some tools charge for export, and some restrict it to higher tiers — which is a step you will hit at the worst moment.
Five. What happens after cancellation. A window during which you can still get in, or immediate lockout. Windows are commonly thirty to ninety days and are worth knowing before rather than after.
And six. Whether an import route exists anywhere else. An export nobody can read is an archive, not a migration.
Asking before buying
One email, before the trial ends.
"Before we commit — can you tell me what a full export contains and what it does not, whether we can run it ourselves, and how long we have access after cancelling?"
Three questions. A vendor with good answers replies specifically and quickly. A vendor with bad answers replies vaguely or redirects to a sales call, and that response is itself the answer.
Ask in writing and keep the reply. It belongs with the purchase record, and it is what you will read in three years.
What a bad answer looks like
"You can print reports."
"Export is available on request."
"That's not something customers usually need."
And silence, which is the most informative of the four.
None of these means the tool is bad. They mean leaving will be expensive, and that belongs in the decision alongside the price.
Data going in
The less discussed direction, and it decides whether you can start at all.
Can you import what you have? A tool requiring manual re-entry of two years of records is a tool with a hidden setup cost, and the cost is somebody's fortnight.
In what format? Usually CSV, and the shape it expects is usually specific.
And what will not come across — the same gaps as on the way out, discovered on the way in.
The thing that makes all of this cheap
Export once, early, while you are still happy.
A month after adopting a tool, run a full export and look at it. You learn what the file contains, whether it is usable, and whether the process works — at a moment when nothing is urgent and you are not annoyed.
Then repeat it annually. Fifteen minutes, and it converts the exit question from a risk into a known quantity.
Most people first attempt an export on the day they have decided to leave, which is the worst possible combination of time pressure and unfamiliarity.
Where it matters most
Records with a long life. Customer conversations, accounts, anything with a retention obligation.
Anything regulated, where you may need to produce records years later regardless of which tool you now use.
And anything that took a long time to accumulate. A year of task history is annoying to lose; a decade of customer correspondence is a different category of loss.
For genuinely disposable data — a scratch tool, a short project — none of this matters, and treating every purchase as though it did is its own waste of time.
Who else holds it
A separate question from export and frequently more consequential.
Where the data rests, which country and on whose infrastructure. For anything holding customer information this may be constrained by law rather than by preference.
Which subprocessors are involved — the analytics, the support desk, the mail sender — each of which is another party with access.
And what notice you get if that list changes. The right to be told before a new subprocessor is added is a term worth having and is standard in better contracts.
These are not paranoid questions for a small firm. They are the questions that determine whether you can honestly answer a customer who asks where their information is kept.
The tool that will not let you leave
Rare, and recognisable in advance.
No documented export. A proprietary format with no reader. Pricing that rises steeply once your data volume is large. And integrations that are one-way — data flows in from your other systems and nothing flows back.
Each of these individually is a design decision. Together they are a strategy, and the strategy is that leaving should be expensive.
Notice them at the trial, which is when they are visible and cost nothing to act on.
Where the export lands
Somewhere that is not the tool.
An export saved into the same account it came from protects against the tool changing and not against losing the account, which is the more common failure.
Download it, put it with your other records, and treat it as any other document you would want in three years. For broader context, see WIRED.
The short version
- The export question belongs at the purchase, when you have leverage and no urgency
- Establish what exports, what does not, whether you can do it yourself, what it costs, the post-cancellation window, and whether anything can import it
- The gaps are where the value is: a customer list without the correspondence is a contact list
- Ask three questions in writing before committing; a vague reply or a redirect to sales is itself the answer
- Run a full export a month after adopting, while nothing is urgent, and repeat it annually
- It matters most for long-lived, regulated or slowly accumulated records, and not at all for disposable ones