Hey everyone, I've been using Fireflies.ai for a few weeks to transcribe my team meetings. I noticed the sentiment scores in the dashboard, but I wanted to track them over time alongside other project data.
So I built a small Python script that uses the Fireflies API to pull the meeting sentiment into a Google Sheet. It runs daily. Now I can easily see trends and correlate sentiment with specific discussion topics. Has anyone else tried something like this? I'm curious if there are better ways to structure the data or what other metrics might be useful to pull.
Interesting idea to track sentiment over time. Most teams stop at just reviewing the scores in the moment.
The biggest thing I'd watch out for is API call costs if your team volume grows. Fireflies' pricing tiers often have hard caps, and scripts like this can quietly blow through them. You might want to add a simple counter to your script to track usage against your plan.
Beyond sentiment, consider pulling data on action items or decisions flagged. That's where you'll see real ROI - correlating sentiment trends with actual project completion rates.
—hd
You're right about the API costs, it's easy to miss that when you're just automating something cool. I set up a quick alert in my spreadsheet that flags when the month's calls hit 80% of our limit.
>Beyond sentiment, consider pulling data on action items or decisions flagged.
That's the gold right there. I started adding a column for 'action item count' from the same API pull, and seeing a negative sentiment score paired with a spike in action items is a huge red flag. It usually means we've just dumped a bunch of stressful tasks on someone without a clear plan.
Have you found a reliable way to categorize the action items? I'm trying to bucket them by department or project type for better correlation.
Data > opinions
Glad you're moving beyond just glancing at the dashboard. A key metric you're missing is speaker distribution - the percentage of talking time per participant.
You'll find that a stable sentiment score is meaningless if one person is dominating 80% of the conversation. The trend you should watch is sentiment variance among multiple speakers, not just the meeting aggregate.
Consider pulling participant-level sentiment and talk time into separate columns. It's a more honest picture of team health.
Your cloud bill is 30% too high
Nice setup, pulling sentiment into Sheets is the first step I took too. The trends are way more useful than static scores.
Have you thought about mapping sentiment against specific agenda points in your meetings? I started tagging transcript segments in Fireflies before the API pull, then matching those tags to rows. It shows which topics tank morale instantly.
Might be overkill for your team size, but adding a simple 'topic sentiment' column helped us spot recurring pain points in project kickoffs.