Missed London? Beyond the Map goes live online on Dec 8.

Pull in surveys and feedback from Amazon S3

This article is for teams who already use the Amazon S3 integration and want to bring qualitative data, such as survey responses, open-text feedback, customer interviews and support logs, into the Data Hub. Until now the S3 pipeline imported structured records: solutions, insights, opportunities, and metrics. It can now also import survey and feedback responses, interview transcripts, and support logs such as tickets, calls and support chats. All of them produce Quotes in the Data Hub, ready to explore with the TheyDo Agent and to turn into insights with AI mining. This article covers what each import adds, how to prepare your export, and how the data flows in.

Before you start

These imports run on top of your existing Amazon S3 integration, so you need that connection in place first. If you have not set up Amazon S3 yet, follow Amazon S3 integration, then come back here.

Like the rest of the S3 integration, this is set up together with the TheyDo team, and you will need your IT or data team to prepare the export. What is new is that, alongside any structured files you already send, you can now export qualitative data as JSON files that follow TheyDo's published schemas.

Choose the right format for your data

Each kind of qualitative data has its own format:

Surveys and feedback: many responses in one file, such as a survey export or a feedback form stream.
Interviews: one file per interview, such as a research interview or a recorded customer call.
Support logs: one file per ticket, call or support chat.

Interviews and support logs each come in two versions. Use the conversation version whenever your source system keeps track of who said what. This covers most chat and transcription tools. Each turn is exported as a speaker plus what they said, for example interviewer and interviewee, or customer and agent. Use the text version only when your transcript is one block of text with no separate speaker turns, such as a ticket body or a plain transcript file.

Tip: For support chats, choose the conversation version. With separate customer and agent turns, TheyDo leaves the agent's replies out when pulling Quotes, so your Quotes reflect what the customer said.

How survey and feedback data lands in TheyDo

When you import a survey or feedback file, each response becomes one or more Quotes in the Data Hub, rather than a structured record. You control how that happens:

Let TheyDo's AI extract one or more Quotes from a longer response, so a single detailed answer can yield several distinct Quotes.
Or import each row verbatim as a single Quote, when you want the response preserved exactly as written.

Field mapping is flexible. You can bring each column in as plain text, map it to a tag group, map it to a persona, or ignore columns you do not need. When you map a column to a tag group, TheyDo uses that column's values to tag the Quotes from each response. If the tag group does not exist in your workspace yet, it is created for you. Persona values are matched to personas that already exist in your workspace. New personas are never created.

How interviews and support logs land in TheyDo

Each interview or support log file becomes exactly one data source in the Data Hub, with the full transcript stored as readable text. Ten interviews give you ten data sources, each titled with the interview or ticket title you export (or its ID if there is no title) and grouped under the source system you name, for example "Gong - EU interviews" or "Zendesk - EU support".

As each transcript comes in, TheyDo's AI reads the conversation and pulls out Quotes automatically. One interview or support chat can produce several distinct Quotes, all linked back to the transcript they came from. You do not need to run anything extra: the Quotes appear in the Quotes Library alongside those from your surveys and feedback.

You can attach context to each transcript in the file:

Tags: add tags grouped by category, such as Region or Touchpoint. Tags apply to the data source. Any tag or tag group that does not exist in your workspace yet is created for you.
Personas: name the personas the transcript relates to. These are matched by name to the personas already in your workspace (upper or lower case does not matter). New personas are never created. A name that does not match an existing persona is skipped, so check that the names in your export match your workspace.

Quotes do not inherit the tags on their data source. TheyDo's AI tags each Quote based on what it says. So if you filter the Quotes Library by a tag, you will only see Quotes that carry that tag themselves, not every Quote from a source with that tag.

Note: Tags and personas are applied when a transcript is first imported. If you change them later in the Data Hub, your changes are kept.

How the import works

Set up the Amazon S3 integration with the TheyDo team if you have not already. Select Contact us on the Amazon S3 card in Settings > Integrations to start.
Choose the format for each kind of data you want to bring in (see Choose the right format for your data).
Prepare your export as JSON, following the matching TheyDo schema (see the next section).
Validate the export with the TheyDo team, so you can confirm the data, tags, and personas come through as expected before it goes live.
Upload your files to the S3 bucket at the frequency you choose. Most teams sync daily.
TheyDo validates each file against the schema and ingests it. Surveys, feedback, interviews and support logs all produce Quotes in the Quotes Library. Each interview and support log also appears as its own data source in the Data Hub.

Note: The Amazon S3 integration validates every file against the schema on upload. If a file does not match, the errors are reported so your team can fix and re-upload. Note: Give interviews and support logs their own S3 connection, one for each. If a connection carries both JSON and CSV files, the CSV files will not process correctly. A connection dedicated to one transcript type also gets a faster processing queue.

Prepare your export

Surveys and feedback

Your survey or feedback file follows a published JSON schema. At a high level it includes:

Source metadata: a display name and a stable source identifier for the survey or feedback stream, plus the field definitions.
Response data: for each response, a unique response identifier, a UTC submission timestamp, and the mapped field values.
Field mapping: for each field, whether it comes in as plain text, a tag group, or a persona, or is ignored.

To update an existing source, re-upload a file that uses the same survey or feedback identifier. TheyDo adds the file to the existing data source rather than creating a duplicate, and each upload appears as its own file entry in that source.

Interviews and support logs

Export one JSON file per interview, ticket, call or chat. Each file includes:

Source system: a stable ID and a display name for the tool the transcripts come from.
Transcript details: a unique transcript ID that stays the same across exports, an optional title, and the date and time the conversation happened, in UTC.
The transcript itself: either a list of conversation turns (each with a speaker and what they said), or a single block of text for the text version.
Tags and personas (optional): up to 50 tags and up to 20 personas.

Note: Each transcript is identified by its source system ID and transcript ID together. If you upload a transcript again with the same pair after it has been processed, the new file is ignored and any edits you have made in the Data Hub stay as they are. To send a corrected transcript, export it with a new transcript ID. This creates a new data source. The original is not removed automatically, so delete it from the Data Hub yourself.

Schemas and tools

You can find the schema definitions, example files, and a command-line tool for validation and upload in TheyDo's public repository at github.com/theydo/theydo-s3-integrations.

Tip: Have your data team run the export through the CLI tool before the first real upload. It validates the file against the schema and tests the connection, so mapping problems surface on their machine rather than after ingestion.

Work with the imported data

Once your data is in the Data Hub, its Quotes behave like any other Quotes there. You can:

Explore and analyze the Quotes with the TheyDo Agent.
Run AI mining to turn feedback, interviews and support logs into insights mapped onto journey steps.
Select the source when you create a new journey, or use it to enrich an existing one.

Because you can re-upload on a schedule, new responses, interviews and tickets keep flowing in, and AI mining can keep surfacing insights as fresh data arrives.

Tip: Amazon S3 already brings your structured data, such as metrics from a BI tool, into TheyDo. Adding surveys, interviews and support logs through the same pipeline means the numbers and the reasons behind them arrive side by side, ready to sit together on a journey.