Your 7-day customer-satisfaction score, flagged when it falls more than 5 points against the prior period — the early signal that something in your support experience has slipped.
At a glance
CSAT Drop Alert watches your customer-satisfaction score for a sudden decline. CSAT is the share of rated conversations that customers marked positive, and a falling score is the clearest leading indicator that your support experience is deteriorating — before it shows up in churn or reviews. This card compares the last 7 days against the prior 7 and raises a flag when satisfaction drops by more than 5 percentage points. For Blitz, a CSAT dip during a busy sales week usually means agents are rushing, a product issue is generating frustration, or wait times have crept up. Catching it within days, not at the end of the month, is what makes it actionable.
| What it counts | The percentage of positively rated conversations over the trailing 7 days, drawn from Intercom’s conversation_rating (a positive rating is the top of the scale), compared against the same metric for the prior 7 days. The card surfaces the current CSAT and the point change. |
| Sample type | API-derived from rated conversations on the conversations endpoint; period-over-period comparison of two 7-day windows. |
| Why it matters | CSAT is your customers grading your support in their own words. A sustained drop predicts refund requests, negative reviews, and churn — and it usually has a fixable cause (slower replies, an unhappy product cohort, a single under-performing agent). The alert turns a slow-moving lagging metric into something you can act on this week. |
| Reading the value | Read the direction and size of the change, not just the absolute score. A drop greater than 5 points trips the alert. A small, single-survey wobble on low response volume is noise; a multi-point fall on healthy volume is real. |
| Currency | percent |
| Time window | 7D vsP |
| Alert trigger | drop >5pts |
| Sentiment key | csat |
| Roles | owner, operations |
Calculation
Vortex IQ pulls conversations rated in the trailing 7 days from the Intercomconversations endpoint, reading each conversation_rating value. CSAT is computed as positive ratings ÷ total ratings × 100 (Intercom’s rating scale tops out at the happiest face; “positive” is typically the top one or two points, per your profile setting). The same calculation runs over the prior 7-day window, and the card reports the current CSAT plus the percentage-point delta. A delta below −5 points trips the alert. Because the metric is sensitive to response volume, the card also carries the rating count so you can judge whether a move is statistically meaningful.
Worked example
A representative reading of CSAT Drop Alert for Blitz. In the prior 7 days Blitz scored92% CSAT across 140 ratings. In the latest 7 days it scores 84% across 155 ratings — an 8-point drop on healthy volume, so the alert fires. The support lead segments the negative ratings and finds they cluster around one product line (a running shoe with a sizing-guide error) and around conversations handled late in the evening. She fixes the sizing guide, adds a canned macro that sets expectations on fit, and adjusts evening cover. The next 7-day window recovers to 90%. To trace which conversations and agents drove the drop, she opens Vortex Mind; to ask “show me every negative-rated conversation tagged sizing this week” she uses Ask Viq.
Sibling cards merchants should reference together
| Card | Why merchants reach for it |
|---|---|
ic_csat | The headline CSAT figure this alert monitors — the current standing without the period comparison. |
ic_csat_trend | The trend line behind the alert — confirm whether a drop is a blip or a sustained slide. |
ic_negative_rated_conversations | The actual conversations dragging the score down — your action list when the alert fires. |
ic_csat_by_agent | Pinpoints whether a drop is concentrated on one agent or spread across the team. |
ic_rating_response_rate | Tells you if the score moved on real volume or a thin sample — essential context before reacting. |
Reconciling against the vendor’s own dashboard
Where to look in Intercom’s own dashboard: In Intercom, open Reports → Customer satisfaction (the CSAT report). It shows your satisfaction score over a chosen range, the rating distribution, and breakdowns by agent and team. Set the range to the last 7 days and compare with the prior 7 to reproduce the period-over-period change this card flags. The negative-rating breakdown there is the same population the Negative-Rated Conversations sibling lists. Why the Vortex IQ value may legitimately differ:| Reason | Direction | What to do |
|---|---|---|
| Positive-rating definition. Vortex IQ counts the top of the rating scale as positive; if your profile counts the top two faces (or vice versa) differently from Intercom’s report setting, the percentages diverge. | Variable | Confirm which rating values count as positive in both systems. |
| Rating-date basis. Vortex IQ buckets by when the rating was given; Intercom can bucket by conversation creation or close date. | Variable | Match the date basis (rating date vs conversation date). |
| Low response volume. With few ratings, one survey swings the percentage hard; the two systems may sample slightly different windows. | Magnified at low volume | Check the rating count before treating a divergence as real. |
| Time zone / window edges. A 7-day window starting at a different hour captures different ratings at the boundary. | Marginal | Align the reporting time zone and window start. |
Known limitations / merchant FAQs
Q: How often does CSAT Drop Alert update? The 7-day windows refresh on the standard data refresh (typically every 30–60 minutes). New ratings flow in continuously, so the score moves as customers respond to survey prompts. Q: Why does my Intercom CSAT report show a different score? The usual reasons are a different definition of “positive” rating, a different date basis (rating date vs conversation close date), and low response volume making both samples swing. Set Intercom’s CSAT report to the same 7-day windows and confirm the positive-rating definition matches. Q: The alert fired but only a handful of people rated — is it real? Be cautious. On low volume, a single negative rating can move the score several points. Check the rating count (and the Rating Response Rate sibling) before acting; a drop on a thin sample warrants watching, not a fire drill. Q: Can I change the 5-point drop threshold? Yes. Thedrop >5pts trigger is configurable per profile in the Sensitivity tab — tighten it for a high-volume desk where small moves are meaningful, or loosen it for a low-volume queue where the score is naturally jumpy.