What leaves with me? The question nobody asks a platform before they sign

Every SaaS evaluation I have watched runs the same way. Someone opens a spreadsheet, lists the features, weights them, argues about price, and picks. Two years later the same person is stuck on the tool they picked, and not one cell in that spreadsheet is why.

They are stuck because leaving turned out to be expensive, and nobody priced it on the way in.

I build a creator platform, so I see this from the wrong side of the table. Creators arrive at us carrying data from somewhere else, and the shape of what they carry tells you exactly how much their old platform respected them. Some arrive with a clean file. Some arrive with a PDF invoice list and no email addresses. One arrived with a screenshot.

This post is about what a real exit looks like, why it is usually missing, and what I think you should ask before you commit anything you care about to somebody else's database.

Export is not one feature, it is four

"Can I export my data?" is close to a useless question. Almost every platform says yes, and almost every yes means something different. Break it into four.

Identity. Do you get working contact details for your users, or do you get internal IDs? A platform that hands you user_84719 and a display name has technically exported your users and practically given you nothing. If you cannot contact a person after you leave, that person was never your relationship.

Money. Do you get the history? Who paid, how much, in what currency, on what date, on what plan, and where a subscription is in its cycle. Without it you cannot reconcile, cannot answer a tax question about last year, and cannot migrate an active subscription without asking every single customer to re-enter a card. That last one is where most migrations quietly die. Asking 400 people to re-subscribe is how you find out that only 250 of them were paying attention.

Content. Do you get the source, or the rendering? An export of your posts as HTML with absolute URLs pointing back at the platform's CDN is not an export, it is a hostage note with nice formatting. Check whether images come with it, and whether the paths still resolve.

Addresses. What happens to URLs people already have. Bookmarks, links in old newsletters, search results that took two years to earn. If those go to a 404 or, worse, to a page advertising the platform, you did not move, you restarted.

A platform can do the first one well and the fourth one terribly. Most do.

Why this gets built badly

It is tempting to read a bad export as malice. Sometimes it is. More often it is three ordinary engineering facts stacked on top of each other.

The first is that export is nobody's feature. It has no acquisition story, no retention story, and no dashboard. When it competes for a sprint against something a prospect asked for on a call, it loses, and it will keep losing every quarter forever unless someone protects it.

The second is that export is where your data model's sins become visible. If subscription state lives partly in your database and partly in Stripe, and the reconciliation only happens inside a webhook handler that has been patched nine times, then a truthful export requires you to define what a subscription actually is. Teams put that off. I have put that off.

The third is genuinely hard. Some of what a user thinks of as "their data" is not solely theirs. A comment thread involves other people. An email address you hold under one privacy policy does not automatically transfer under another. A payment method belongs to the processor, not to you or to them. There are real questions here, and they make a convenient hiding place for the two problems above.

What I would actually ask

If you are evaluating anything you plan to depend on, these five questions take ten minutes and tell you more than a feature matrix.

1. Show me an export file. Not the docs page. An actual sample, with the columns visible. If support has to ask engineering, you have your answer.
2. Is it self-serve? An export that requires a support ticket is a chokepoint, and chokepoints get slower exactly when relationships get worse.
3. What is the API story on the way out? Read access to your own historical data, without a plan upgrade, is the difference between a migration and an archaeology project.
4. What happens to my URLs? Ask specifically whether you can keep a custom domain and point it elsewhere, or whether your addresses are subdomains you are renting.
5. Has anyone actually left? Ask them to describe a migration off their platform. A vendor who can walk you through it calmly has built for it. A vendor who has never thought about it will improvise, and you will hear that.

Notice that none of these are about features. They are about whether the company has decided, structurally, that you are allowed to go.

Being honest about my own side

It would be easy to end here with a pitch. I would rather be useful.

DukieX is the thing I build. It puts a creator's posts, memberships, community, products, courses and livestreams in one Space, so that a person who buys a course is already a member rather than a row in a second system, and so the operator writes to their members from inside the same place the memberships live. On the paid plans the Space runs on the creator's own domain, which is the part that matters most for question four above, because a domain you control is the only address that survives you changing your mind.

Here is where we are not the answer. We have no discovery network. Nobody browses DukieX and finds you, which is a genuine reason to stay on Substack or Skool if that is where your growth comes from. Our courses are not a full LMS: no quizzes, no certificates, no drip scheduling, no progress tracking, so if you are teaching a cohort that needs assessment, a dedicated course platform beats us and I would rather say so than have you find out in month two. We publish the rest of that list, including the places competitors beat us on price, at https://dukiex.com/vs

The reason I put those in writing is the same reason I wrote this post. A platform that will not tell you where it is weak is unlikely to tell you how to leave.

The version of this you can act on today

You do not need to be evaluating anything to get value out of this. Go and export from whatever currently holds your users, right now, and open the file.

If you get working email addresses, payment history and content you could actually republish, you are in a good position and you should stop worrying about it.

If you get less than that, you have not discovered a crisis. You have discovered a number: how much it would cost you to change your mind. Knowing that number is worth more than any feature comparison, and it takes about two minutes to find out.