248 unsubscribe events / 123,750 delivered.
All tools
DENOMINATORS / UNCERTAINTY / SEGMENTS / CHURNMeasure the exits.
Measure the exits.
Find the pressure.
Calculate the delivered-rate signal, compare it with a real baseline, and locate where opt-outs concentrate before projecting list loss.
- Wilson interval
- Baseline significance
- Segment concentration
DELIVERED RATE0.200%95% interval 0.177%–0.227%
REVIEW THE CHANGE
The opt-out signal deserves a segmented review.
Unsubscribe rate describes recorded events per delivered message. It does not reveal recipient intent by itself and should be read with complaints, bounces, cadence, source, and expectation.
Human-scale frequency1 per 499Delivered messages per recorded unsubscribe.
A different denominator; keep the definition attached.
Complaints can be a higher-cost exit signal than unsubscribes.
Unsubscribe plus complaint events; overlap is assumed absent.
Comparable-period change
Two-proportion approximationCurrent0.200%248 / 123,750
Baseline0.140%154 / 110,000
Absolute change+0.060%Percentage points
Relative change+43.1%Rate ratio 1.43×
The observed rate increased beyond the model's sampling threshold.
p ≈ <0.001This comparison assumes independent delivered-message observations and comparable traffic. It does not establish campaign causality.
Segment concentration
100.0% of current delivered volume representedSegmentDeliveredUnsubscribesRateShare of exitsvs overall
Gmail recipients75,0001250.167%50.4%0.83×
Microsoft consumer26,000690.265%27.8%1.32×
Yahoo / AOL14,000280.200%11.3%1.00×
Other domains8,750260.297%10.5%1.48×
Other domains has the highest valid segment rate at 0.297% and contributes 10.5% of represented unsubscribe events.
Repeated-rate churn projection
Scenario, not forecastEstimated events / month9524 sends × modeled delivered audience × current rate
After 3 months277,1532,847 modeled cumulative subscriber losses
After 6 months274,3345,666 modeled cumulative subscriber losses
After 12 months268,78411,216 modeled cumulative subscriber losses
The model assumes the current event rate repeats, delivery performance stays constant, the declared audience share is representative, no acquisition occurs, and an unsubscribed person leaves future eligible sends. It is not a unique-person forecast when event feeds overlap or campaigns reach different people.
Interpretation and data checks
5 findingsCurrent denominator is usable. The primary result uses 123,750 delivered messages, while the sent-denominator rate remains separate.
The comparable-period rate increased. The observed change is +0.060% with an approximate two-sided p-value of <0.001. Investigate mix and measurement changes before assigning cause.
The rate is within the declared house ceiling. This comparison is an internal decision rule, not an industry certification.
Segment rows reconcile to the current totals. Weighted totals and contribution shares use the same event ledger.
Complaints add a second exit signal. Combined unsubscribe and complaint events equal 0.225% when overlap is assumed absent. Verify deduplication before calling this unique people.
KEEP THE DENOMINATOR ATTACHED
Export the rate, uncertainty, segments, and assumptions together.
A bare percentage loses its population, time window, comparison scope, and event definition. The handoff keeps those decisions visible.
Read the full agent guideTHE MEASUREMENT CONTRACT
The numerator is easy. The denominator changes the story.
Unsubscribe rate is commonly calculated from delivered messages because only delivered recipients had the opportunity to use the email's unsubscribe path. A sent denominator can still be useful for operational reconciliation, but it answers a different question.
Keep event uniqueness, campaign scope, delivery definition, complaint overlap, and comparison eligibility attached to every rate.
- 01Define
Document the event, denominator, time basis, campaign scope, and deduplication rule.
- 02Calculate
Use delivered messages for the primary rate and retain sent volume for reconciliation.
- 03Compare
Use a truly comparable baseline and inspect the uncertainty around both proportions.
- 04Segment
Find receiver, source, frequency, content, or cohort concentration before changing everything.
- 05Act
Review expectation, cadence, relevance, permission, and opt-out usability with complaints and bounces.
RATE QUESTIONS
What makes a comparison trustworthy?
Comparable traffic, stable event definitions, adequate volume, and the same delivery denominator matter more than a generic benchmark.
Should unsubscribe rate use sent or delivered messages?+
Delivered is the primary denominator in this tool because those messages created an opportunity to unsubscribe. The sent-denominator rate is shown separately for operational reconciliation.
What is a good unsubscribe rate?+
There is no universal healthy number. Compare against your documented house rule and comparable history, then segment by source, frequency, expectation, receiver, and campaign.
Why show a Wilson interval?+
The observed percentage is an estimate from finite data. The interval shows plausible sampling variation without relying on the weakest normal approximation for a single proportion.
Does statistical significance prove the campaign caused the change?+
No. The two-proportion approximation only evaluates whether the observed rate difference is difficult to explain by sampling variation under its assumptions. Mix shifts, tracking changes, seasonality, and frequency can still drive the result.
Can I add complaints and unsubscribes together?+
Only as a labeled event total when the feeds are mutually exclusive or deduplicated. The combined rate here explicitly assumes no overlap and should not be described as unique people without evidence.
Is the churn projection a forecast?+
No. It is a repeated-rate scenario with no acquisition and stable delivery, cadence, audience share, and event behavior. Use it to understand sensitivity, not to promise future list size.