Sentiment trend
Tracks whether customer sentiment is improving, stable, or declining across recent interactions
One of 51 free, open-source prompts from Sonora — copy-paste-ready AI prompts for customer success and account management, organized by customer lifecycle stage. Copy this into any AI assistant, or run it on autopilot inside Sonora.
Analyze the sentiment trend in recent communications from [customer name] to determine whether customer sentiment is improving, stable, or declining.
Step 1: Pull recent interactions
Gather the last 90 days of customer interactions across:
- Meeting transcripts and eCSAT scores
- Support tickets and cases
- Emails and other communications
Step 2: Split into two periods
Compare sentiment across:
- Recent: Last 45 days
- Prior: 45–90 days ago
Step 3: Assess sentiment in each period
Look for:
- Positive signals: Praise, satisfaction, proactive engagement, high eCSAT (≥ 4)
- Negative signals: Complaints, frustration, escalations, low eCSAT (< 3)
- Neutral: Routine operational communications with no strong tone
Step 4: Classify
- Improving: Recent period has notably more positive signals or fewer negative signals than prior period.
- Stable: No significant change in sentiment between periods.
- Declining: Recent period has notably more negative signals or fewer positive signals than prior period.
Return exactly one of: Improving, Stable, Declining
Include the key evidence (e.g., “Two escalations in the last 30 days after a clean prior period.”) in the reason field.
Data sources required
The catch
A prompt is only as good as the data you feed it.
Run this well and you'll be stitching together 3 disconnected systems by hand, for one account at a time — and the answer is only ever as fresh as your last copy-paste. Sonora already has the context, live:
- Meeting transcripts — connected and continuously synced
- Support tickets — connected and continuously synced
- Customer communications — connected and continuously synced
Inside Sonora this prompt runs grounded in your real customer data, on a schedule, across every account at once — and flags what needs you, before you have to ask.
More Monitor prompts
Bad call
Detects customers with poor eCSAT scores (less than 4) in the last 2 weeks
Build VoC report
Synthesizes themes from calls, tickets, and sentiment data into a Voice of Customer artifact for product teams
Call count (90d)
Tallies the total number of meetings held in the last 90 days
Competitor mentions
Identifies customers who have mentioned competitors in recent communications
Customer health brief
Synthesizes health signals, risks, and opportunities into a 1-page executive brief with three prioritized next steps
Diagnose account issues
Surfaces and prioritizes everything going wrong at an account with root cause analysis and a recovery plan