The Scenario
You manage QA for a support team. Every month you score call interactions — listening to recordings, reviewing transcripts, rating the conversation against a rubric. This month there are 30 conversations from last week that your team flagged for review. The conversation IDs are in a Google Sheet. The transcripts are in Intercom.
Getting the transcripts out is normally a manual task: open each conversation, scroll to the call recording section, copy the transcript, paste it into the sheet. You did this last month. It took most of a morning and you still ended up with formatting that made the text hard to read.
The bad version:
- Open Intercom conversation #1, find the call section, expand the transcript, select all, copy.
- Paste into Google Sheets. Watch it collapse into one cell because it contains line breaks.
- Spend time unwrapping the text, formatting it as readable prose in column B.
- Do this 29 more times. Eat lunch at your desk.
QA scoring was supposed to start this afternoon.
The Easy Way: One Prompt in SheetXAI
SheetXAI is an AI agent that lives inside your Google Sheet. It reads each conversation ID and pulls the transcript text directly into the adjacent column.
For each conversation ID in column A of this sheet, retrieve the call transcript from Intercom and write the transcript text and call date into columns B and C — leave B blank if no transcript is available.
What You Get
- Transcript text written to column B for each conversation that has a call recording.
- Call date written to column C.
- Column B left blank (not an error) for conversations where no transcript is available — so you know immediately which ones to handle differently.
- Text formatted as plain readable prose, not a JSON blob or a single collapsed cell.
What If the Data Is Not Quite Ready
Some conversation IDs in the sheet are wrong or have already been deleted
A few IDs were entered manually and may be stale.
For each conversation ID in column A, attempt to retrieve the Intercom call transcript — if the conversation does not exist or has no call, write "not_found" or "no_transcript" to column B rather than leaving it blank.
You want the transcript broken into speaker turns, not as one block
The QA rubric scores by turn — you need agent turns and customer turns separated.
For each conversation in column A, retrieve the call transcript and write speaker-labeled turns to column B — format each turn as "AGENT: [text]" or "CUSTOMER: [text]" on separate lines.
You need to compute word count per transcript for scoring time estimates
The rubric assigns scoring effort based on conversation length.
Retrieve call transcripts for each conversation ID in column A — write transcript text to column B and call date to column C — then compute the word count of column B and write it to column D.
Pull transcripts, label speaker turns, compute word counts, and flag short calls
Calls under 200 words are too short to score against the full rubric and get a different process.
For each conversation in column A, retrieve the Intercom call transcript, format it as labeled speaker turns in column B, write the call date to column C, compute word count to column D, and write "short_call" to column E for any transcript under 200 words — otherwise leave column E blank.
Try It
Get the 7-day free trial of SheetXAI and open any Google Sheet with a list of Intercom conversation IDs — ask it to pull the transcripts directly so your QA team can start scoring without the manual copy-paste step. For managing the conversations themselves, see the bulk-assign-conversations spoke and the Intercom hub overview.