Turn last week's support chats into product backlog items

Every Monday at 8am, read last week's Freshchat support chats, cluster them into recurring themes, turn the themes worth acting on into Linear issues, and post a digest to Slack.

Start in Freshchat. Use List Users to gather the contacts who have been active recently. For each of those users, run List User Conversations, and for each conversation run List Conversation Messages to pull the actual transcript. Freshchat has no account-wide conversation feed, so this fan-out from users to conversations to messages is the only path to the transcripts. Use Get Conversation if you need conversation-level detail such as status or channel. Keep only messages from the past seven days.

Two Freshchat details to handle while gathering. Message text lives in a message_parts array rather than a single body field, so reconstruct each message from its parts before reading it. List endpoints paginate with page and items_per_page, and message lists default to 20 with a maximum of 50, so page through until the full seven-day window is covered instead of stopping at the first page.

Read across all of the transcripts together rather than one at a time, and cluster what customers said into recurring themes such as bugs, confusing or unclear flows, and feature requests. For each theme, track how many distinct customers raised it, and capture two or three short verbatim quotes in the customer's own words. Deliberately skip one-off questions, routine password reset and login help requests, and anything else that is a normal support interaction rather than a signal about the product, so only patterns worth acting on reach the backlog.

For every theme raised by at least three separate customers, check Linear before writing anything. Call Search Issues first, using the theme wording plus a few distinctive keywords from it, to see whether an issue already covers this ground. If a matching issue exists, use Update Issue to append the refreshed customer count and any new quotes to its description rather than filing a duplicate. If nothing matches, use Create Issue on the product team with the theme as the title and the quotes plus the affected customer count in the description. Treat this duplicate check as the most important step: the backlog should accumulate evidence on existing issues week after week, not fill up with near-identical tickets.

Finish with Slack Send a Message, posting a digest to the support channel. List each theme with its customer count, whether it was newly filed or updated this week, and a link to the Linear issue. It also helps to note how many one-off questions and password resets were skipped, so the team can see the filtering is working. If no theme cleared the three-customer threshold, post a short note saying so rather than staying silent.