CSAT, Intercom

Percentage of positive Intercom conversation ratings (4-5 stars) over all rated conversations in the last 30 days. How to read it, why it matters, and how to act on it.

Metrics type: Key Metrics • Category: Customer support

% positive conversation_rating (4-5★) over rated conversations in the window.

At a glance

CSAT is the customer satisfaction score from Intercom’s post-conversation rating. When a conversation closes, the customer can rate it 1-5 stars; this card reports the share that rated 4 or 5 — happy — over every conversation that was rated in the last 30 days. It is the clearest read on whether Blitz’s support is actually leaving customers satisfied, not just fast. The founder treats it as the quality counterweight to the speed and backlog cards: a desk can be quick and busy and still be making customers unhappy, and CSAT is where that shows up.

What it counts	The percentage of conversations rated 4 or 5 stars (conversation_rating.rating in ) out of all conversations that received any rating in the window. Unrated conversations are excluded from the denominator.
Sample type	API-derived. Computed from the conversation_rating object Intercom attaches to rated conversations on the Conversations API.
Why it matters	It is the headline quality signal. Speed and volume tell you whether the desk is keeping up; CSAT tells you whether customers are happy with the help they got. A falling CSAT is an early warning of churn and bad word-of-mouth that no operational metric catches.
Reading the value	Higher is better, shown as a percentage on a gauge. Read it against your own baseline; most healthy support desks sit comfortably above 85%. The alert fires below 70%.
Currency	percent
Time window	30D
Alert trigger	<70%
Sentiment key	csat
Roles	owner, operations

Calculation

Vortex IQ counts conversations closed and rated in the last 30 days, splitting them by conversation_rating.rating. Ratings of 4 and 5 are treated as positive; the score is positive ratings divided by all rated conversations, expressed as a percentage. Conversations that were never rated do not count either way — they affect Rating Response Rate, not CSAT. Because the score only reflects customers who chose to rate, a low response rate makes the figure noisier; read the two cards together.

Worked example

A representative reading of CSAT for Blitz over the last 30 days. The gauge shows 78% — above the 70% alert line but down from a steady 90% the month before. The support lead opens CSAT by Agent and Negative-Rated Conversations. Two things stand out: a new seasonal hire is sitting at 61% CSAT, and a cluster of 1-2 star ratings all carry the “delivery” tag. The story is twofold — the new agent needs coaching on tone, and a courier delay is generating frustration support cannot fix but is absorbing the blame for. The lead books a coaching session and adds a proactive shipping-delay macro so customers hear about delays before they have to ask. CSAT recovers to 88% the following month. To find the exact comments behind the dip, open Negative-Rated Conversations; to ask directly, try Ask Viq “what’s dragging our CSAT below 80% this month?”

Sibling cards merchants should reference together

Card	Why merchants reach for it
ic_csat_trend	Plots CSAT over 90 days so you can tell a one-off dip from a slide.
ic_csat_by_agent	Splits the score by agent to find who needs coaching or recognition.
ic_negative_ratings	Lists the low-rated conversations with their comments and tags — the why behind the number.
ic_rating_response_rate	Tells you how many customers actually rated, so you know how much to trust the CSAT figure.
ic_health_score	The composite this satisfaction score feeds as its CSAT input.

Reconciling against the vendor’s own dashboard

Where to look in Intercom’s own dashboard:

Open Reports → Customer satisfaction. Intercom shows the CSAT score, rating distribution and response rate there. Match the date range and confirm how Intercom is bucketing ratings (some configurations treat 3 stars as neutral rather than negative) to compare against this card.

Why the Vortex IQ value may legitimately differ:

Reason	Direction	What to do
Positive bucket definition. This card treats 4-5 as positive; if Intercom’s report counts 3 as positive or neutral, the percentages differ.	Variable	Confirm the rating buckets match.
Window. The card uses a rolling 30 days; an Intercom report on a calendar month covers different conversations.	Variable	Align the date range.
Rated-only denominator. Both exclude unrated conversations, but if an Intercom view includes them, the denominator differs.	Variable	Check the report counts rated only.
Time zone. Intercom reports in the workspace time zone; Vortex IQ aligns to the merchant reporting time zone.	Marginal	Confirm time zone match.

Cross-connector reconciliation: when a CSAT dip clusters on a delivery or stock tag, cross-check Complaints on Out-of-Stock SKUs to see whether the cause is operational rather than service quality. For divergence investigations, use Vortex Mind.

Known limitations / merchant FAQs

Q: How often does CSAT update? It refreshes on the standard data refresh (typically every 30-60 minutes) as newly closed conversations are rated.

Q: Why does my Intercom report show a different CSAT? The most common reason is how 3-star ratings are bucketed (positive, neutral or negative) and the date range. This card treats 4-5 as positive over a rolling 30 days.

Q: Why does CSAT look volatile some weeks? Because it only reflects customers who chose to rate. When the Rating Response Rate is low, a handful of ratings can swing the figure — read the two cards together.

Q: Can I customise the alert threshold? Yes. The default fires below 70%, but set it per profile in the Alert Rules tab to match the satisfaction bar your brand holds itself to.

---

Tracked live in Vortex IQ Nerve Centre

CSAT is one of hundreds of KPI pulses Vortex IQ tracks across Intercom and 70+ other ecommerce connectors. Nerve Centre runs the detection layer; Vortex Mind investigates the cause when something moves; Ask Viq lets you interrogate any number in plain English.

Start for free or book a demo to see this metric running on your own data.