How to migrate off Otter.ai without losing your meeting history
The reason people lose data when they leave a transcription tool is almost never a technical failure. It is an ordering mistake: they cancel first, then try to retrieve what they assumed they could still reach.
Export and verify while the subscription is still live. Switch second. Cancel third. Never any other order. Everything below is just detail on those three steps.
Step 1 — Inventory what you actually have
Before you plan anything, find out what is genuinely still reachable. Open Otter's conversation list and scroll to the bottom, letting it load as you go.
Write down two numbers:
- How many conversations are listed. This is your ceiling. No exporter can reach beyond it, because they all read the same session you do.
- The date of the oldest one. If that is much later than when you started using Otter, your older material is no longer visible to you.
If the second number surprised you, that is the whole reason to do this now rather than later. Lower-tier accounts don't surface your full history. The list shrinks from the back, quietly, without a warning email.
Step 2 — Export everything, in a format that outlives the tool
Do this before you change plans, not after. Your conversation list is at its longest right now.
The order of preference for output format, and why:
| Format | Survives 10 years? | Reasoning |
|---|---|---|
| Markdown | Yes | Plain text with light conventions. Any editor opens it. Reads fine even if every tool you use today disappears. |
| JSON | Yes | Keeps speaker turns, timestamps and structure intact for future processing. Not pleasant to read, essential to have. |
| CSV | Yes | Good for one row per utterance. Great for search and analysis, awkward for reading. |
| PDF / DOCX | Partly | Fine as a human-readable copy. Structure is cosmetic, not machine-readable, so it's a poor archive. |
Plain .txt |
Partly | Loses speaker labels and timestamps. You will regret this and it is expensive to reconstruct. |
If you only do one thing from this article, do this: take both Markdown and JSON. They cost nothing to produce together and the pair covers both “a human reads it in 2036” and “a script processes it tomorrow”.
Step 3 — Verify, properly
This step is where migrations are actually won or lost. Three checks, all of them cheap:
-
Count the files
It should match the number from Step 1 — exactly, not approximately. A shortfall almost always means pagination stopped early.
-
Check both ends of the range
Open your oldest file and your newest. If the oldest is much later than your Step 1 number, files are missing from the middle or the back.
-
Read the middle of three random files
Not the first line — the middle. Truncated transcripts fail in the middle, where a chunked fetch gave up silently and the last paragraph looks like the meeting just ended there.
You can do this while you still have access to re-export. That is the entire point of doing it now: a correction at this stage takes five minutes; the same correction after you cancel takes nothing, because there is nothing to correct from.
Get everything out in one pass, in both formats
Otter Exporter writes a single ZIP: one Markdown file per conversation, plus
index.csv for counting and conversations.json for
processing. It runs entirely in your browser and only ever reads your own
conversations. Free tier covers 10 conversations a month; Pro is $29 once.
Step 4 — Decide where the archive actually lives
An export sitting in your Downloads folder is one disk failure away from not existing. Pick a real home before you cancel anything:
- A notes app (Obsidian, Logseq, Bear, Notion). Best for daily reading and backlinking. Markdown drops straight in. Keep the folder in cloud sync so it is actually backed up.
- A git repository. Best for versioning and diffing. Text-only, no surprises, and the history is itself a backup if you push it somewhere.
- Plain cloud storage with a sane folder layout. Simplest, lowest effort. Make sure it is somewhere that will still exist when the current vendor pivot happens — i.e. not inside another startup's proprietary format.
Whichever you pick, open three files in the new home and confirm they read correctly before you touch your Otter plan. “I copied the folder” and “the files are readable in their new home” are different claims.
Step 5 — Now switch, then cancel
Only at this point is it safe to change anything about Otter itself. Order matters here too:
- Move your workflow to the new tool and use it for a week or two.
- Confirm the new tool actually covers the meetings you care about — especially any recurring call it might handle worse.
- Then downgrade or cancel Otter.
The reason to delay cancellation is that it is your undo button. Keeping one cheap month of access costs far less than discovering in three weeks that a recurring meeting type doesn't work in the new tool and you no longer have the originals to fall back on.
What you won't get back
Honest list, so you can decide whether you care before you leave:
- Live collaboration inside the transcript. If colleagues comment in-line on Otter transcripts, that shared annotation layer does not travel. Comments and highlights export inconsistently from Otter itself, and less reliably from anywhere else.
- The exact speaker diarisation. Speaker labels will be re-derived if the new tool re-transcribes. Names you assigned manually may or may not survive; plan to spot-check a few meeting types.
- Search across the old archive in the new tool. Most tools index only what they have ingested. If you keep the archive as flat files, you get filesystem search — which is fine, but it isn't the same as full-text search over a live corpus.
FAQ
Should I cancel Otter before or after exporting?
After. Always after, and only once you've verified the export by counting files and reading a few mid-file sections. Cancelling first is the single most common way people lose meeting history.
Will I get my old conversations back if I upgrade?
It depends on Otter's current retention terms and how your account is configured, and those terms have changed before. Don't plan a migration on the assumption that a future upgrade restores access you don't have today. Export what you can see, now.
How long does a migration take?
The export itself is usually minutes. The part that takes time is verification and settling into a new home for the archive. Budget an afternoon for the whole thing rather than assuming the export finishing means the migration is done.
Can I keep using Otter for new meetings and still archive locally?
Yes, and it's a reasonable setup: record in Otter, then run a periodic export to fold new conversations into your local archive. Just be aware the visibility limit means “periodic” needs to be frequent enough that your older conversations haven't aged out of view.