TextTrace
English
TürkçeEnglishTürkçeDeutschFrançaisEspañolPortuguês (BR)Italiano

Journalism and research · Interview archive

Keep every quoted chat passage connected to the source you reviewed.

Preserve the original export, build a searchable working record, log consent and scope separately, and check every quote against the source. Use CSV for analysis and PDF for human reading, but de-identify only on a working copy.

Updated 2026-09-078 minTextTrace

Separate four records

SourceThe unchanged TXT, ZIP, screenshots, recording, or supported backup reference.
Working recordThe searchable transcript used for coding, review, and navigation.
Consent and scope logWho agreed to what use, which dates and participants are included, and any restrictions.
Publication datasetThe minimum excerpts or rows required for the story or study, de-identified where appropriate.

Quote verification loop

  1. Search a distinctive phrase in TextTrace.
  2. Open the result with preceding and following messages.
  3. Compare spelling, punctuation, sender label, date, and attachment context with the source.
  4. Record the source reference in your notes.
  5. Have a second reviewer check high-risk quotations when policy requires it.

Research limits and privacy

A row-per-message CSV is useful for coding, but it can expose people who did not consent to participate and can flatten replies, reactions, or attachments. Plan exclusion and anonymization before analysis. Keep identifiers out of filenames, restrict access, and delete working outputs according to your newsroom, institution, ethics-board, or source agreement—not according to a generic retention promise.

Questions, answered precisely

Frequently asked questions

Can TextTrace anonymize a research dataset automatically?

Do not assume full de-identification. Review names, numbers, message text, attachments, and indirect identifiers in the exported working copy.

Is CSV always better for qualitative analysis?

CSV supports rows and filters; PDF preserves a more readable chronology. Many projects keep both, with the unchanged source stored separately.

Articles

Continue with a practical guide

Put each message into a row you can filter.
Conversation → CSV

Put each message into a row you can filter.

Import the conversation, review the reconstructed record, then export CSV for sorting by people, visible dates, text, and other available fields.

4 min · Updated 2026-09-05
Do not read a 50,000-message chat from the beginning.
Long conversations · Search workflow

Do not read a 50,000-message chat from the beginning.

Define the question, then narrow the record by person, visible date, phrase, attachment type, source, direction, or available transcript. Open each result in surrounding context before deciding what belongs in the handoff.

7 min · Updated 2026-09-07
“Private” should describe a data path, not a slogan.
Privacy · Processing boundaries

“Private” should describe a data path, not a slogan.

TextTrace uses different processing routes for different sources. Supported transcript files can be parsed directly, while visual extraction uses Google Cloud Vision through the TextTrace service. Review the in-app disclosure and current privacy policy before importing sensitive material.

8 min · Updated 2026-09-07