How We Score Exit Reports
A score you cannot argue with is a score you should not trust. Here is the entire rubric, the grade bands, and the things we deliberately leave unscored.
On this page
A score with no visible method is an opinion wearing a number. So here is the whole thing, including the parts that are inconvenient.
The six criteria
Each is scored 0 to 5, for a maximum of 30. Every point has to trace back to something the vendor has published, which is quoted on the report and linked in its sources.
1. Self-service
Can you get your data out yourself, from inside the product, without asking anyone?
Five means you click a button. Lower scores reflect gates: needing to contact support, needing an administrator to act, or a role restriction that stops ordinary users exporting. Zero means there is no route out that does not involve a sales conversation.
This matters because the moment you most need an export is usually the moment your relationship with the vendor is worst.
2. Format openness
Is the export in a format another product can actually read?
CSV, JSON and Markdown score well because they are documented and universally supported. A proprietary archive that only the vendor's own software can open scores low even if it contains everything, because completeness you cannot read is not portability.
PDF sits awkwardly here. It is open, and it is also where structured data goes to die — an export that flattens a database into a document has technically exported it and practically lost it.
3. Completeness
Does the export carry the hard parts?
The easy parts always travel: names, titles, dates, the fields visible in a list view. The hard parts are the ones that carry the accumulated value — comments and discussion history, attachments and uploaded files, custom fields, and the record of what changed and when.
Five means those come out. Lower scores mark each omission. This is the criterion where products differ most, and the one buyers almost never check.
4. Not paywalled
Can you export fully on the plan you are already on?
Some products place export, or the useful export format, on a higher tier. That converts leaving into a purchase, which is a meaningful thing to know before you are in the position of making it.
Five means the full export is available at every tier including free. Lower scores reflect formats or scopes locked behind an upgrade.
5. Documentation clarity
Does the vendor plainly say what is and is not included?
This is the criterion most likely to surprise people, because it rewards honesty rather than capability. A product with a limited export that publishes its limits clearly scores better here than one with a good export nobody can describe.
The reasoning is practical: a documented limitation is one you can plan around. An undocumented one is discovered during the migration.
6. Caps and expiry
Are there hard limits that can bite you at the worst moment?
Record caps, action-history ceilings, download links that expire, limits on how many times an archive may be downloaded, processing times measured in days. These are rarely prominent and they are exactly the things that turn a planned migration into a failed one.
Five means no caps we could find. Lower scores name the specific limit.
Grade bands
| Score | Grade | What it means |
|---|---|---|
| 26–30 | A | You can leave with essentially everything |
| 21–25 | B | You leave with most of it; the losses are known and survivable |
| 16–20 | C | Real gaps. Plan around them before you commit data |
| 11–15 | D | Significant lock-in. Assume some things stay behind permanently |
| 0–10 | E | Leaving means losing a substantial part of your history |
What we deliberately do not score
This list matters as much as the rubric, because a score that quietly includes guesswork is worse than no score.
Anything we would have to test with an account we do not have. Export speed, whether a file actually opens cleanly, how a specific import behaves. We are reading documentation, not running migrations, and dressing one up as the other would be exactly the kind of unverifiable claim this site exists to avoid.
Things the vendor has not published. Silence is recorded as a gap in its own section on each report and has no effect on the number. Inferring a score from absence would make the total look better-evidenced than it is.
Support quality, refunds, or contract terms. Real considerations, not data portability, and lumping them in would make the score mean less.
Whether the product is any good. An exit report is not a review. Several products we recommend elsewhere on this site score poorly here, and that is not a contradiction — it is the point of measuring two different things separately.
The honest limitations
Three, stated plainly.
Documentation can be wrong or out of date. Vendors change export tooling without updating every help page. If what you see in the product differs from what we report, the product is right and we want to hear about it.
Documentation can be incomplete in ways we cannot detect. A help page that does not mention a limitation may mean there is no limitation, or may mean nobody wrote it down. We cannot always tell which, and where we suspect the latter we say so.
These facts expire. Which is why every report leads with the date its sources were checked rather than a vague freshness signal. A dated fact you can verify beats an undated one you cannot.
Disagreeing with a score
The reasoning sits next to every number precisely so you can. If you think a criterion is weighted wrongly, or that we have read a vendor's documentation too generously or too harshly, the argument is right there on the page to be had.
That is deliberate. A rubric you cannot argue with is one you should not trust — and the numbers here are meant to be a starting point for your own judgement, not a substitute for it.
Corrections go through the correction policy, and every report links to it. The reports themselves are indexed on the exit reports hub.
Questions people ask
How is the exit score calculated?
Six criteria, each scored zero to five, added together for a maximum of thirty. The criteria are self-service access, format openness, completeness, whether export is paywalled, documentation clarity, and caps or expiry. Each score appears next to the reasoning that produced it.
Do you actually test the exports?
No, and we would rather say so than imply otherwise. These reports are built by reading each vendor's published documentation and quoting it. That is a real limitation - documentation can be out of date or incomplete - but it has one advantage over a claimed hands-on test: you can check every source yourself in about a minute.
What happens when a vendor does not document something?
It goes in the "what we could not verify" section on the report and does not affect the score. Inferring a score from silence would make the number look more complete than the evidence behind it, which is the failure mode this whole method exists to avoid.
Does a low exit score mean I should avoid a product?
No. Plenty of excellent software has a poor exit, and a low score is a reason to plan differently rather than to walk away - keep the irreplaceable things somewhere portable, and run an export early while nothing depends on the answer.
How often are the scores rechecked?
Each report shows the date its documentation was last checked at the top of the page. Export tooling changes quietly and often, so we treat that date as the most important fact on the page rather than something to bury in a footer.
Spotted something out of date or incorrect? Tell us and we will fix it.