Every vendor we've ever signed with has an export button, and every vendor rep has used the same line when we asked about getting out later: "you own your data, you can export it anytime." It's true, in the narrowest possible sense. Press the button and a file lands in your downloads folder. The lie isn't in the button. It's in the word "useful" that we quietly attach to the word "export" without anyone ever promising it.
An export is a snapshot of one table. A business is not one table. It's a contact in the CRM connected to a proposal connected to a deposit connected to a job connected to an invoice connected to a renewal date six months out. The connections are the business. The rows without the connections are just a filing cabinet that's been knocked over.
The mechanism: reassembly cost
Here's the plain term worth saying out loud in your own Monday meeting: reassembly cost. It's the price of turning a pile of exports back into a coherent history, and it's almost always higher than anyone budgeted, because nobody budgets for it at all. We budget for migration, which sounds like plumbing. We don't budget for reassembly, which is closer to archaeology.
The CRM export gives you a contact ID that means nothing to any other system. The invoicing export gives you a customer name, spelled slightly differently because someone typed it fresh into a different form. The scheduling export gives you a date and a job site but no reference back to the deal that created it. None of these files share a key. A key is what lets you say "this row and that row are about the same customer, the same job, the same event." Without it, you don't have ten years of company history. You have ten years of unrelated facts that happen to mention the same people.
This is why "you own your data" is true and also beside the point. Ownership of disconnected fragments isn't ownership of a history. You can hold every fragment in your hand and still not be able to answer the question a new owner, a new hire, or a bank asks on day one: what actually happened with this customer over the last three years.
Where the standard advice runs out
The standard advice is "export your data regularly, from everything, just in case." It's not wrong, it's just incomplete, and treating it as sufficient is how people get burned. Regular exports protect you against a vendor disappearing. They don't protect you against the harder problem, which is that even with every export sitting on your drive, nobody can rebuild the joins after the fact. You'd need someone to sit down with the CRM export and the invoicing export and manually match three hundred customer records by hand, guessing at which "J. Martinez" is which, and which date next to which dollar figure corresponds to which job. That's not a backup. That's a research project, and research projects don't get done during a renewal push.
It's also worth saying where consolidation is the wrong call, because the instinct to over-correct is real. If you're a small shop running one well-chosen tool for scheduling and nothing else touches it, there's no reassembly problem to solve, because there's nothing to reassemble against. If your current stack is genuinely integrated, meaning the systems already share a customer key and sync in both directions, the fact that it's technically "a dozen vendors" matters less than whether the joins survive a switch. The number of logins isn't the risk. The number of disconnected keys is.
A composite case
Picture an operator who runs a multi-crew services business. The complaint, as she puts it in the owner's words: "We've got ten years of customer history, we can export everything, switching tools should be simple." On paper, true. Every system she uses has an export button and she's used all of them.
The real diagnosis, once someone actually tried to rebuild a single customer's timeline: the CRM listed the customer under a company name, the invoicing system listed the same customer under the owner's personal name, and the scheduling tool listed jobs by site address with no customer reference at all. Three systems, three keys, zero joins. The handoff between the person who closed the deal and the person who scheduled the job had always happened by memory and a sticky note, not by a shared record. Nobody had noticed because nobody had ever needed to look backward across all three at once. The renewal conversation only ever needed last quarter.
What changed wasn't a wholesale rip-and-replace overnight. It was a decision to stop adding new history to systems that couldn't be joined, and to route new customers, proposals, jobs and invoices through a shared customer key from that point forward, in a place where the key was structural rather than a convention someone had to remember to follow. The old history got reassembled once, by hand, as a one-time project, and then frozen as read-only reference rather than living data anyone had to keep reconciling.
What improved was not dramatic, it was procedural, which is the kind of improvement that actually holds. The Monday meeting stopped starting with "whose version of this customer are we looking at." The renewal conversation could pull one timeline instead of three screens. The scorecard review stopped needing a translator between what sales thought closed and what invoicing thought got paid. None of that shows up as a headline. It shows up as a meeting that ends on time.
What ownership actually has to mean
If "you own your data" is going to mean anything, it has to mean you can reconstruct your company's history without a research project. That's a much narrower claim than the marketing version, and it's the one worth testing before you sign anything new, not after you try to leave something old.
The test is simple enough to run this week: pick one real customer, one with a few years of history, and try to produce a single timeline of everything that happened with them using only exports from your current tools. Proposal, job, invoice, renewal, in order, in one place. If someone has to guess, match by hand, or ask around to fill the gaps, you don't have ten years of history. You have ten years of exports, and those are not the same thing. The export button was never the problem. The absence of a shared key was.