Weekly support quality review board for Gorgias tickets

Build me a support quality review board that my CX lead and reviewers open every week. It runs on Gorgias for the ticket data and a Google Sheets QA log that acts as the app's memory. There are two views: a review board where a reviewer reads closed conversations and scores them against a short rubric, and a scorecard that reads the log back to show how each agent is doing.

The review board is the main view. Load a weekly sample using the Gorgias List Tickets operation filtered to closed tickets and ordered newest first, and cap the sample so the board stays workable (default 25 tickets, with the cap and the lookback window configurable in the app). Only include tickets that actually received an agent reply, so auto closed and spam tickets never reach a reviewer. Show one row per ticket with the assigned agent, the channel, the ticket tags, the CSAT rating if one came back, the close date, and whether the ticket has already been reviewed.

Agent attribution comes off the ticket payload itself, from the assignee or user fields, since Gorgias has no documented operation for listing users. Take ticket tags from the ticket payload when they are present and fall back to List Ticket Tags per ticket. Pull CSAT with List Satisfaction Surveys, which returns surveys newest first, and match them to the sampled tickets by ticket id in a single pass rather than calling once per ticket. Gorgias rate limits API key integrations to roughly 40 requests per 20 second window, so fetch the list data once per load, cache it, and hydrate per ticket detail only when a row is actually opened.

Clicking a row opens the ticket in place. Render the full conversation from List Messages in reading order, clearly separating customer messages from agent replies and existing internal notes, so the reviewer never has to open Gorgias in another tab. Beside the conversation, show the scoring panel: a four part rubric covering tone, accuracy, completeness and policy, each scored 1 to 5, plus a free text coaching comment. Keep the rubric to these four dimensions by default, since scorecards with more categories wreck scoring consistency between reviewers, and make the dimension labels and their descriptions editable in one place.

Saving a review does three things. It appends a row to the Google Sheets QA log with Append Values. It posts the coaching comment back on the ticket with Create Message as an internal note, never as a customer facing message, and that is a hard rule everywhere the app writes to Gorgias. And it adds the qa-reviewed tag with Add Ticket Tags so the same ticket does not get scored twice.

The Sheets log is the only persistence the app needs, so give it a stable column layout: review id, reviewed at, ISO week, ticket id, subject, channel, agent name, agent email, closed at, CSAT rating, tone score, accuracy score, completeness score, policy score, total score, coaching comment, supporting quotes, source (human or agent drafted), status (draft or confirmed), and reviewer name.

Add a Score this batch button that kicks off a background agent over the current sample. For each sampled ticket the agent reads the conversation with List Messages, grades it against the same four part rubric, and must attach a short supporting quote from the conversation behind every single score, so a reviewer can check the reasoning instead of trusting a bare number. The agent writes its results into the same Sheets log with Append Values, with source set to agent drafted and status set to draft. Those rows appear in the board as pre filled drafts clearly marked as agent drafted, and a human confirms or overrides each one; confirming or editing a draft updates that row in place with Update Values rather than appending a duplicate. Agent drafted rows never count toward the scorecard until a human has confirmed them, and the board should show how many are still pending confirmation.

The second view is the scorecard, and it is purely a read of the log with Get Values. Show average score per agent per week, the trend across recent weeks, and each agent's weakest rubric dimension so coaching has an obvious starting point. Include a per agent page listing that person's reviewed tickets with their scores, the coaching comments, and a way back into the conversation view.

Apply per user visibility throughout. An agent signed into the app sees only their own scores and their own agent page, while CX leads see everyone and can compare across the team. Match the signed in user to log rows on their email against the agent email column, and keep the list of lead accounts configurable in the app.