How do I export my call transcripts and metadata?

TL;DR

Go to Data exports → New export, pick a date range, and start it. Discera fetches every call's transcript and HubSpot metadata for that window and packages them into a downloadable ZIP. Exports are a backup of your own data — they don't use your analysis call quota.

What an export is (and isn't)

An export is a raw data backup, not an analysis. There's no prompt, no model, and no executive summary — just your calls, captured exactly as Discera reads them. Use it to:

• ·Keep your own copy of transcripts and deal context outside Discera.
• ·Re-analyze later without re-fetching — a saved export can be fed straight back into an analysis, skipping the slow Gong/HubSpot fetch.

If you instead want findings, themes, and an executive summary, run an analysis (Your first analysis) — that produces a report, not a transcript archive.

Steps

1. 01Open Data exports in the sidebar, then click New export (or go to /exports/new).
2. 02Choose a date range — use a quick preset (Yesterday, Last week, Last month, Last quarter) or set custom From/To dates.
3. 03Leave Include transcripts on to bundle full transcripts, or turn it off for a lighter metadata-only archive.
4. 04Set Max calls if you want to cap the export (up to 10,000).
5. 05Click Start export. You'll land on a live progress page while Discera fetches and packages your calls.
6. 06When it finishes, click Download archive — or grab it anytime later from the Data exports list.

What's in the archive

The download is a single ZIP:

discera-export-<id>.zip
├── manifest.json        # date range, filters, counts, schema version
├── calls.jsonl          # one full record per call (the re-import format)
├── metadata.csv         # flat per-call summary for spreadsheets
└── transcripts/         # one plain-text file per call (if enabled)
    └── <call_id>.txt

• ·calls.jsonl is the full-fidelity record — participants, HubSpot account/contact/deal context, and the transcript — one JSON object per line. This is the format Discera can read back in to analyze the export later.
• ·metadata.csv is the human-friendly view: call title, date, duration, account, deal name/stage/amount, outcome, and whether a transcript was present. Open it in any spreadsheet.
• ·transcripts/ holds greppable plain-text transcripts, included only when Include transcripts is on.

Filters

An export honors the same HubSpot filters as an analysis (deal stage, company size, owner, amount range, and any configured property filters). Set them in the builder to export a focused slice instead of the whole window. Filters are optional — leave them empty to export everything in the date range. See How do analysis filters work?.

Analyze a saved export later

Because an export captures the exact records an analysis needs, you can re-run analysis against it without touching Gong or HubSpot again:

1. 01On the Data exports page, click Analyze on any export that has transcripts.
2. 02You're taken to the analysis builder with a snapshot banner — the date range is ignored (the export defines the call set) and the run reads from the archive.
3. 03Write your prompt and start the run as usual.

This is faster than a live run (no fetch step) and is the recommended way to iterate on prompts over the same set of calls. Two things to keep in mind: it does consume call quota (it runs the LLM), and it analyzes a point-in-time snapshot — HubSpot values reflect the moment the export was captured, not the present.

Quota, limits, and retention

• ·No quota cost to export. Exports never deduct from your monthly quota or add-on balance.
• ·One export at a time per running job slot — exports share your plan's concurrent-job limit with analyses.
• ·Up to 10,000 calls per export.
• ·Retention matches reports — older exports are hidden once they fall outside your plan's window and reappear on upgrade. The data isn't deleted. See Plans and quotas.