New conversations in the last 24 hours, flagged when they run more than two standard deviations above your 30-day baseline — your early warning that something just broke or went viral.
At a glance
Volume Spike watches the rate at which new conversations land in your Intercom inbox and raises a flag when today’s inbound runs far above what’s normal for your store. It is an anomaly detector, not a counter: the headline is the count of new conversations in the last 24 hours, but its job is to tell you when that count is statistically unusual. For Blitz, a sudden surge almost always has a single root cause — a broken checkout, a delivery-carrier outage, a stock-out on a hyped drop, or a marketing email that landed wrong. Catching the spike early lets you fix the cause before the inbox becomes unmanageable.
| What it counts | New Intercom conversations created in the trailing 24 hours (conversations filtered by created_at), compared against the mean and standard deviation of daily new-conversation volume over the preceding 30 days. The card reports the 24-hour count and flags when it exceeds the baseline by more than 2σ. |
| Sample type | API-derived: a rolling 24-hour count benchmarked against a computed 30-day daily-volume distribution. |
| Why it matters | A volume spike is the leading indicator of an operational problem upstream of support — a payment failure wave, a shipping delay, a site bug, or a viral moment. It tells you to look outward (what changed?) before you drown in inbound. |
| Reading the value | A flagged spike means today’s inbound is abnormally high for your store, not just “busy”. Treat the flag, not the raw number, as the signal — what’s a spike for a quiet B2B desk is a normal Tuesday for a high-volume retailer, which is why the baseline is per-store. |
| Currency | number |
| Time window | 24H |
| Alert trigger | >2σ vs 30D baseline |
| Sentiment key | — |
| Roles | owner, operations |
Calculation
Vortex IQ counts conversations created in the trailing 24 hours via theconversations search endpoint filtered on created_at. In parallel it computes the mean (μ) and standard deviation (σ) of daily new-conversation counts over the preceding 30 days. The card flags an anomaly when the 24-hour count exceeds μ + 2σ — roughly the top ~2.5% of expected daily volumes, so genuinely unusual rather than merely high. The baseline self-adjusts as your normal volume shifts, so a growing store doesn’t get false alarms and a quiet desk still catches its smaller spikes.
Worked example
A representative reading of Volume Spike for Blitz. Blitz’s 30-day baseline is a mean of 90 new conversations a day with a standard deviation of 18, so the spike line sits at 90 + (2 × 18) = 126. On most days the inbox takes 70–110 conversations and the card stays quiet. On Thursday it reports184 and flags red — well past 126. The support lead opens the underlying tags and sees a cluster of “payment declined at checkout” conversations created since 14:00. That points straight at the payments stack; she cross-references the Support Spike on Failed Payments card, confirms a card-acquirer issue, loops in the founder, and posts a banner on the site to deflect duplicate contacts. To understand why the spike concentrated in one hour, she uses Vortex Mind; to ask “what tags are driving today’s volume?” she uses Ask Viq.
Sibling cards merchants should reference together
| Card | Why merchants reach for it |
|---|---|
ic_conversation_volume_trend | The full trend line behind the spike — see whether this is a one-day blip or the start of a sustained climb. |
ic_new_conversations_today | The raw today-count this spike detector is built on. |
ic_top_topics_tags | The fastest way to find what a spike is about — which tags surged. |
ic_xc_support_spike_failed_payments | The most common revenue-at-risk cause of a spike; cross-references payment failures. |
ic_alert_sla_breach | A spike usually causes breaches within hours — watch both to manage the downstream impact. |
Reconciling against the vendor’s own dashboard
Where to look in Intercom’s own dashboard: In Intercom, go to Reports → Conversations (or the Conversation volume report) and set the range to the last 24 hours, then compare against a 30-day daily view. Intercom shows you new-conversation counts over time but does not compute a statistical spike flag — so you’ll be eyeballing the bar height against recent days. To find what’s driving it, open Reports → Topics or filter the conversation list by tag andcreated_at = today.
Why the Vortex IQ value may legitimately differ:
| Reason | Direction | What to do |
|---|---|---|
| Spike is statistical, not raw. Vortex IQ flags relative to your 30-day baseline; Intercom shows only the raw count. A high day may not be flagged if your variance is naturally wide. | Conceptual | Read the flag, not just the number — the baseline is the point. |
| 24-hour rolling vs calendar day. Vortex IQ uses a trailing 24-hour window; Intercom reports often use calendar days in the workspace time zone. | Variable | Align the window when comparing exact counts. |
| Conversation vs message counting. This card counts new conversations; some Intercom views count inbound messages, which inflates the number for multi-message threads. | Vortex IQ may read lower | Compare against Intercom’s “new conversations” metric specifically. |
| Baseline warm-up. A connector live for under 30 days has a shorter baseline and may flag more readily until enough history accumulates. | Vortex IQ may over-flag early | Allow the baseline to mature for accurate sensitivity. |
Known limitations / merchant FAQs
Q: How often does Volume Spike update? The 24-hour count refreshes on the standard data refresh (typically every 30–60 minutes); the 30-day baseline recomputes daily. During an incident, force a manual refresh to see the latest count. Q: Why does my Intercom report not show a “spike”? Intercom reports raw volume; it does not compute a per-store statistical baseline or a 2σ flag. The spike detection is the value Vortex IQ adds — the same raw count can be a non-event for a high-variance store and an alert for a steady one. Q: I had a spike but the inbox felt fine — was it a false alarm? Not necessarily. A spike that gets absorbed by good staffing is still worth knowing about because the cause (a payment wave, a shipping delay) may still be hurting revenue even if support coped. Use the tags and cross-channel siblings to find the root cause. Q: Can I change the sensitivity? Yes. The>2σ threshold is configurable per profile in the Sensitivity tab — tighten to 1.5σ to catch smaller surges or loosen to 2.5σ for a noisy, seasonal inbox.