You hit Export, expect a clean PDF of the conversation or a zip of your full data, and instead get a 0-byte file, a request that never finishes, or a PDF with only the system header and no messages. This happens often enough that most heavy Claude users have run into it at least once. The root cause is usually one of three: the conversation is too long for the renderer, it contains artifacts that fail to flatten into print, or the data-export pipeline is throttled or routed to a region your account cannot read from. Each has a different fix.
Common causes
Ordered by how often each one shows up.
1. Conversation is too long for the PDF renderer
Claude renders the entire transcript to HTML, then prints to PDF. Past roughly 200 messages or 500 KB of HTML, the renderer can silently abort and ship an empty file.
How to judge: If the conversation is under 50 turns, this is probably not it. Over 100 turns with code blocks and tables, it almost certainly is.
2. Artifacts in the thread fail to flatten
Artifacts (React previews, SVG diagrams, HTML pages) need to be rasterized into the PDF. If any artifact throws during render, the whole export can fall over.
How to judge: Open the offending conversation, scroll for artifacts. Try exporting a different conversation with no artifacts; if that one works, this is your cause.
3. Data export job is queued but never delivered
The full data export (Settings, Privacy, Export data) runs as a background job. The email link can take hours or fail silently if the job hit a backend error.
How to judge: Check Settings, Privacy, Export data history. If the latest entry shows “processing” for over 24 hours, the job is stuck.
4. Email link expired or went to spam
Data exports come as a one-time link valid for 24 hours. If it lands in spam or you miss the window, the file is gone and you must request again.
How to judge: Search your mailbox for “Anthropic” or “Claude export.” Check spam, promotions, and any aliases.
5. Browser blocked the download
Print-to-PDF triggers a download dialog. Aggressive popup blockers, restrictive Safari settings, or corporate browser policies can suppress it without an error toast.
How to judge: Try the export in a clean Chrome profile with no extensions. If it works there, your main browser is blocking it.
6. Region-restricted account on a tightened export endpoint
Some EU and APAC accounts have data export routed through a separate endpoint that occasionally lags or errors. The button works but no file ever arrives.
How to judge: Check status.anthropic.com for regional incidents on the day of the failure.
Before you start
- Decide whether you need a single-conversation PDF or a full account data export; the fix paths diverge.
- Make sure you are signed in to the same workspace where the conversation lives.
- Have a clean browser (Chrome or Edge, no extensions) ready as a fallback.
Information to collect
- Conversation length (rough message count) and whether it contains artifacts.
- Browser, version, and any active extensions (especially ad-blockers, privacy tools).
- Screenshot of the export dialog and any error toast.
- Timestamp of the export attempt for log correlation.
- Whether other conversations export cleanly from the same account.
- Account plan (Free, Pro, Team, Enterprise) — export limits differ.
Step-by-step fix
Step 1: Retry the export in a fresh tab
Close the conversation tab, reopen it from the sidebar, and trigger Export, Conversation as PDF again. About a third of failures clear on this alone because the prior render state was stale.
Step 2: Print to PDF via the browser instead
Open the conversation, press Cmd+P (Mac) or Ctrl+P (Windows). Choose “Save as PDF” as the destination. This bypasses Claude’s renderer and uses the browser’s, which handles long HTML better.
Step 3: Split the conversation for export
If the thread is huge, scroll to a midpoint, branch off a new conversation from that turn (Edit, then Send), and export the two halves separately. Slower but reliable.
Step 4: Request a fresh data export
Settings, Privacy, Export my data. Click Request export. Wait up to 24 hours. Watch the email tied to your account, including spam and category tabs.
Step 5: Disable extensions and try again
Open the conversation in an Incognito or Guest window with no extensions loaded. Trigger the export. If it works here, an extension in your normal profile is the culprit.
Step 6: Contact support with timestamps
If three attempts in 48 hours all fail, file a ticket at support.anthropic.com with the conversation ID (visible in the URL), timestamps, and a screenshot. Pro and Team plans get faster turnaround.
Step 7: Use the API as a last resort
For mission-critical archival, hit the messages list endpoint with your API key, dump JSON, and render to PDF locally with pandoc or wkhtmltopdf. This always works because it bypasses the UI entirely.
Verify
- The exported PDF opens cleanly and shows every message, not just the first turn.
- Code blocks render with monospace font and preserve indentation.
- Artifacts appear as static snapshots or clearly labeled placeholders.
- Data export zip extracts without errors and contains the expected conversations.json file.
Long-term prevention
- For research-grade threads, branch a new conversation every 50-100 turns to keep exports under the renderer limit.
- Run a monthly data export on Pro and Team accounts so you always have a recent archive.
- Whitelist anthropic.com in your popup blocker and ad-blocker.
- Keep one browser profile clean of extensions specifically for exports and downloads.
- For teams, consider an automated nightly API dump to S3 or Google Drive.
Common pitfalls
- Assuming an empty PDF means the conversation was lost — the source is still on Claude’s servers, only the render failed.
- Requesting the data export 5 times in a row. Each request queues separately and slows the pipeline further.
- Trusting that the email arrived “right after” — check the timestamp; old links expire.
- Exporting on mobile Safari, which often suppresses the download silently.
- Forgetting that artifacts in Pro mode may not survive PDF rasterization at all.
FAQ
- Why is my exported PDF blank? The HTML renderer aborted, usually on a long thread or a failing artifact. Try browser Print to PDF instead.
- How long should the data export take? Usually under 6 hours. Past 24 hours, the job is stuck and you should request again or contact support.
- Can I export an entire workspace at once? Not from the UI today. Use the Admin API on Team or Enterprise, or request a bulk export from support.
- Does export include attached files? Conversation PDF does not include file binaries. Data export zip does include attachments under a separate folder.
- Is there a file size limit? Soft limit around 50 MB per conversation export and roughly 2 GB per data export zip. Beyond that, split.
- Do artifacts survive export? They render as static snapshots in the PDF, not interactive. Source code blocks survive intact.
Related
- Claude artifact download issue
- Claude artifact renders blank
- Claude file upload stuck on processing
- Claude mobile app sync broken
- Claude PDF table extraction wrong
Tags: #Claude #Troubleshooting #artifact #PDF