Resend to Non-Openers Under a Frequency Cap
A home-goods retailer is about to send a 48-hour flash-sale email to 1.2 million opted-in contacts. The campaign is scheduled for Tuesday 10:00 a.m. in each contact's local time zone. The account has a global frequency cap of 4 marketing messages per contact per 7 days, and 18% of the target audience has already received 4 messages in the last 6 days. The previous flash sale had a 21% open rate and a 0.08% complaint rate; the retailer wants to resend to non-openers 24 hours later with a different subject line. The email team proposes a straight resend to all non-openers, but the compliance team flags that the last campaign to non-openers produced a 0.19% complaint rate. The retailer also has a 30-day suppression list for recent complainers and a requirement that all promotional sends respect quiet hours from 9 p.m. to 8 a.m.
1. Should the retailer send the 24-hour non-opener resend to every non-opener? Explain your decision and the controls you would apply before send.
Read the worked answer
No. A non-opener is not a clean audience. Some non-openers are already at the 4-message weekly cap, some are on the 30-day complainer suppression list, and some will fall into quiet hours depending on their time zone. A blanket resend would push capped contacts over the limit and increase complaint risk beyond the 0.19% already observed.
Before send, I would apply these controls:
- Frequency cap: exclude anyone who has received 4 marketing messages in the last 7 days. The 18% at cap cannot legally or responsibly receive another promotional send.
- Suppression: exclude the 30-day complainer list, global unsubscribe, hard bounces, and any account-level suppression.
- Quiet hours: schedule the resend at 10:00 a.m. local time, but if the journey or campaign cannot enforce local quiet hours, hold contacts whose send time falls outside 8 a.m.–9 p.m. until the next eligible window.
- Holdout: keep the existing 5% holdout out of the resend so lift can be measured.
- Volume guardrail: send in batches and monitor complaint and unsubscribe rates after the first 10–15% of volume. If complaint rate exceeds 0.12%, pause.
The decision rule is to resend only to eligible non-openers who remain under the cap and outside suppression, not to the full non-opener pool.
2. Design the A/B test for the resend subject line and define the decision rule. Include the primary metric, guardrails, and what you would avoid.
Read the worked answer
I would randomly split eligible non-openers 50/50 between subject line A (control, same as original) and subject line B (new angle). The test unit is the contact, and the split must happen after all eligibility filters so both arms have the same frequency-cap and suppression profile.
Primary metric: click-through rate to the sale landing page, not open rate. Opens are increasingly unreliable because of mailbox-provider privacy features and prefetching, and the business outcome is a click or purchase.
Guardrails: complaint rate, unsubscribe rate, and hard-bounce rate. A subject line that lifts clicks but doubles complaints is not a win.
Decision rule: run until the test has enough volume to detect a 10% relative lift in click rate at 95% confidence; do not peek and stop early on a favorable day. If B shows a statistically significant click lift with no guardrail breach, use B for the remainder or next send. If B is flat or worse, keep A and document the result.
Avoid: testing on open rate alone, changing the sender name and subject line at the same time, or sending both subject lines to the same contact. Also avoid declaring a winner from the first few hours of data.
3. After the resend, what would you check in Insights & Analytics to confirm the resend did not harm list health?
Read the worked answer
I would compare the resend cohort with the holdout and with the original send on four dimensions:
- Complaint rate and unsubscribe rate: these are the earliest warning signs. A resend that lifts revenue but raises complaints above the original campaign's 0.08% baseline needs review.
- Suppression-list growth: count new complainers and unsubscribes attributable to the resend, not just the original send. If the resend adds a disproportionate share of new suppressions, the non-opener audience is too risky for a second touch.
- Revenue and conversion per recipient: compare resend recipients with the holdout to confirm incremental lift, not just attributed revenue. The resend should not cannibalize purchases that would have happened anyway.
- Deliverability signals: bounce rate, spam-trap hits if available, and domain-level reputation trends. A complaint spike can affect future sends beyond this campaign.
I would review these at 24 hours for early warning and at 7 and 30 days for list-health impact. If complaint rate exceeds the guardrail, I would suppress the remaining non-openers and revise the resend policy.