Sudden bounce surge = list contamination from a bad import. Cap sends until investigated.
At a glance
A 24-hour anomaly detector for bounce-rate spikes. Fires when last-24h bounce rate exceeds the 30-day rolling baseline by more than 2 standard deviations. A sudden spike almost always means list contamination from a bad import, a re-engagement send to a dormant tier, or a deliverability event. Cap sends until investigated. Compares against Bounce Rate’s slower-moving 30D rate; this alert catches the spike before the 30D rolling number registers it.
| What it counts | Last-24h bounce rate computed as SUM(bounces in 24h) ÷ SUM(emails_sent in 24h) × 100, compared against the trailing 30-day mean and standard deviation. Alert fires when 24h rate > mean + 2σ. |
| API endpoint | Marketing API v3, GET /3.0/reports?since_send_time={24h ago} for the recent send window, plus a 30-day historical pull for baseline computation. |
| Audience-based scope | Aggregates across every audience. Per-audience spike detection not surfaced; account-level alert. A spike from a single audience’s send still fires this alert because the blended 24h rate breaches. |
| Channel scope | Email only. SMS bounces tracked separately. |
| Hard vs soft bounces | Combined. Mailchimp’s bounces.hard_bounces + bounces.soft_bounces summed; the spike rule treats both equally because either type indicates immediate deliverability risk. |
| Attribution model | Not applicable, deliverability metric. |
| MPP impact | None. Bounces happen at SMTP layer before MPP. |
| Statistical baseline | 30-day mean and standard deviation computed from daily bounce-rate readings. New accounts (less than 14 days of history) use a fixed 5% threshold instead of statistical baseline; the alert is suppressed entirely on accounts with <7 days of sends. |
| 2σ choice | 2σ corresponds to ~95% confidence interval. Roughly one false-positive per ~20 sends if the baseline is well-behaved, acceptable for a spike-alert. Lower thresholds (1.5σ) are too noisy; higher (3σ) miss meaningful spikes. |
| Time window | 24H (last 24 hours vs 30D baseline) |
| Alert trigger | >2σ vs 30D baseline (drives the spike-detection insight). |
| Roles | owner, marketing, engineering |
Calculation
Calculated automatically from your Mailchimp data. See the At a glance summary above for what the metric tracks and the worked example below for a typical reading.Worked example
A small DTC accessories brand on Shopify with Mailchimp Standard, single audience of 24,000. Steady-state metrics over the prior 30 days:| Metric | Value |
|---|---|
| 30D mean bounce rate | 1.4% |
| 30D standard deviation | 0.5 pp |
| 2σ alert threshold | 2.4% |
| Metric | Value |
|---|---|
| 24h bounce rate after send | 5.8% |
| Alert state | FIRED (5.8% > 2.4% threshold) |
| Hard bounces from import contacts | 220 (out of 3,500 imported) |
| Hard bounces from existing audience | 165 (out of 24,000 existing, 0.69%) |
- The alert fired exactly as intended. Without the alert, the merchant would have continued sending under the assumption everything was fine. The 30D rolling rate would only register the breach over the next 5-7 days, by which time Gmail and Yahoo would already be throttling the domain. The 24h spike alert exists to catch the bad import before reputation damage compounds.
- The bad import contributed 6.3% bounce rate on its own slice. 220 hard bounces out of 3,500 imported = 6.3%, well above the 5% Mailchimp Compliance ceiling. Trade-show CSVs almost always degrade faster than expected: emails written by hand on a clipboard contain typos, throwaway addresses given to avoid spam, role-based aliases that bounce on bulk sends.
- The existing audience’s bounce rate at 0.69% is healthy, the spike is import-only. This diagnostic separation is the value of the alert. If both slices showed elevated bounces, the cause would be deliverability (sender reputation crisis); seeing only the import slice spike means the cause is list contamination.
- The remediation sequence is well-defined. Step 1: pause sends to the imported contacts immediately. Step 2: suppress the hard-bouncing 6.3% of imports (220 contacts) explicitly. Step 3: send a confirmation email to the remaining 3,280 imports asking them to opt in (double-opt-in). Step 4: only the confirmed survivors return to the audience proper. Step 5: monitor the next 3 sends; if the 24h alert stays cleared, you’re recovered.
- Mailchimp’s automation depth is light enough that this manual remediation is required. Klaviyo has a built-in “list cleanse on import” flow that flags this pattern automatically. Mailchimp does not. Vortex IQ’s audit MC-DEL-001 (sustained bounce >5%) and this 24h spike alert together cover the gap, but the response is still merchant-driven.
Sibling cards merchants should reference together
| Card | Why pair it with Bounce Spike Alert |
|---|---|
| Mailchimp Bounce Rate | The slower-moving 30D rate. This alert catches spikes before that rate registers them. |
| Mailchimp Spam Spike Alert | The complaint-rate twin. List contamination often causes both spikes simultaneously. |
| Mailchimp Sender Reputation Alert | Composite alert; this 24h spike is one of its inputs. |
| Mailchimp Suppressed Members | The cumulative cost of bounces. Watch suppression spike alongside this alert firing. |
| Mailchimp Audience Size | A spike usually shrinks the audience as Mailchimp auto-cleans hard bouncers. |
| Mailchimp Audience Growth Rate | If growth came from a single import that’s now bouncing, growth and this alert correlate. |
| Mailchimp Email-Attributed Revenue | Revenue follows deliverability with 2-3 week lag. Today’s spike predicts later revenue dip. |
| Mailchimp Top Campaigns by Revenue | Identify which recent campaign triggered the spike. |
Reconciling against the vendor’s own dashboard
Where to look in Mailchimp’s own dashboard: Mailchimp does not surface a “spike alert” natively. The closest views are Mailchimp → Reports → individual recent campaign → Performance for the per-campaign bounce rate of the suspected trigger, and Mailchimp → Account → Sender Reputation for the blended reputation score that lags behind the spike. This alert is a Vortex IQ-derived statistical detection layered on top of Mailchimp’s raw bounce data. Why our alert state may legitimately differ from a hand-built calculation:| Reason | Direction of divergence |
|---|---|
| Time-zone. 24h windows in UTC; Mailchimp’s per-campaign view in account tz. The 24h boundary differs slightly. | <0.1 pp on the 24h rate |
| Baseline window. 30 days of rolling history. New accounts (<14d) use a fixed 5% threshold; the alert may be muted when statistical baseline is unreliable. | None when baseline is stable |
| Page caps. Engine pages campaigns 10 per call. >50 distinct sends in 24h may truncate; rare but possible. | Vortex IQ slightly under-counts on extreme volume |
| Customer Journey contributions. Engine includes both campaigns + Customer Journey email steps in the 24h window. Mailchimp’s UI per-campaign view doesn’t blend them. | None on the alert; affects how merchants verify |
| Suppression mid-window. Mailchimp auto-suppresses hard bouncers after first bounce. A bad import that bounced 1 hour ago has already had its hard bouncers suppressed; subsequent sends in the same 24h window won’t bounce again from those addresses. | The 24h rate falls naturally as suppression takes effect; alert may clear before merchant intervenes |
| Card | Expected relationship | What causes legitimate divergence |
|---|---|---|
| Gmail Postmaster Tools (no connector) | Gmail’s “domain reputation” should track downward in the days following a bounce spike. Postmaster lags by 24-48h. | Postmaster Gmail-only; Mailchimp all-ISP. |
klaviyo.klv_alert_bounce_spike | When both ESPs run, a list-contamination event affects both if both have the bad addresses. | Migration history matters. |
shopify.total_revenue | Revenue dips with 2-3 week lag after sustained spikes. No direct numerical relationship. | Other channels can compensate. |