Under Files to download in the extension, tick the formats you want: Spreadsheet (CSV), Readable transcripts (TXT) and Structured data (JSON). Every plan has all three, and the extension remembers your choice. Only the files you tick are built, so fewer files means a faster export.
- One file ticked: it downloads on its own, for example
inbox-export-yourstore-2026-10-30.csv. No unzipping. - Two or more: they come together in one zip, like
inbox-export-yourstore-2026-10-30.zip.
conversations.csv
Choose the layout under Spreadsheet rows in the extension. It remembers your choice.
- One row per conversation (the default): the whole chat sits in one cell, oldest message first, one line per message like
[2026-10-01 10:01] Customer 1: Hi…. Best for reading and counting conversations. - One row per message: a conversation with 8 messages takes 8 rows, and the conversation’s details repeat on each row. Best for filtering by who wrote what.
Columns in both layouts
| Column | What it contains |
|---|---|
conversation_id |
Shopify’s ID for the conversation |
conversation_status |
open or closed |
handled_by |
automated if Inbox’s automatic replies handled it, otherwise staff |
conversation_started |
When the chat began (UTC) |
topics |
Topic tags Shopify added, such as “Shipping” or “Returns / Delivered orders” |
customer… |
Customer columns, depending on your plan: see customer details by plan |
One row per conversation also has
| Column | What it contains |
|---|---|
last_message_at |
When the last message was sent (UTC) |
message_count |
Number of messages in the conversation |
conversation_text |
The whole conversation. A spreadsheet cell holds about 32,000 characters, so a very long chat continues in conversation_text_2, _3 instead of being cut off. |
One row per message also has
| Column | What it contains |
|---|---|
message_number |
1 for the first message in the conversation, 2 for the next… |
message_sent_at |
When the message was sent (UTC) |
sender_role |
Customer, Staff or Automated |
sender_name |
Staff member’s name, or the customer’s name or label |
message_text |
What was written. Attachments, products and discounts are described in brackets, like [Attachment: …]. |
Which messages are included
- Everything people wrote or sent: customer messages, staff replies and automated replies, including attachments, product links and discounts.
- No system events. Things Inbox records by itself, such as “conversation assigned” or “customer viewed a page”, are left out.
- Away messages only when they’re last. An away reply is kept, marked “(away message)”, only when it ends the conversation, because then it shows the customer was left waiting.
The transcript file
Every conversation written out in order, with a header line for its date, status and topics, then one line per message: Customer: …, Staff (Elena): …. It’s the easiest file to read, search with Ctrl+F, or share with someone who doesn’t want a spreadsheet.
conversations.json
The same data in a structured format for developers and other tools. It also records which plan and level of customer detail the export used.
Large exports come in parts
So your computer stays responsive, a big export is saved in parts of up to 5,000 conversations each: …-part-1.zip, …-part-2.zip and so on, or …-part-1.csv, …-part-2.csv if you ticked only the spreadsheet.
- Every file carries its part number, such as
conversations-part-2.csv, so a file you open or upload on its own still shows which part it is. - The transcript header and the JSON file also give the part number and the dates that part covers.
- “Customer 7” is the same person in every part of one export.
An export that fits in one part keeps the plain names, like conversations.csv.