Email Engagement Metrics: What ISPs Actually Track

Engagement is the dominant 2026 deliverability signal. ISPs read recipient behavior — opens, clicks, replies, deletions, spam reports — to decide where future mail lands. Senders see some of these signals; ISPs see all of them. Understanding what's measured and how to influence the aggregate is the difference between healthy inbox placement and a slow slide to spam.

This guide covers the engagement metrics senders can see, what ISPs use internally, and how to improve the aggregate.

What senders can see

Metric	Source	Reliability
Unique opens	ESP open tracking pixel	Medium (Apple MPP contamination)
Total opens	ESP	Medium
Unique clicks	ESP click tracking	High
Total clicks	ESP	High
Replies	Some ESP integrations	Low-medium (often manual)
Forwards	Limited ESP tracking	Low
Unsubscribes	ESP	High
Spam complaints	Feedback loops (FBL, JMRP)	High (for participating ISPs)
Time on email	Limited ESPs	Low

What ISPs see (that you don't)

Signal	Used for filtering	Visible to sender
Open (real)	Yes	Partially (via tracking pixel, contaminated by MPP)
Click	Yes	Yes
Reply	Yes (strong positive)	Only for replies to owned domains
Marking as spam	Yes (very negative)	Via FBL/JMRP, lagging
Marking as not-spam	Yes (positive)	Generally no
Move from spam to inbox	Yes (positive)	No
Add to contacts	Yes (positive)	No
Delete without opening	Yes (negative)	No
No interaction over time	Yes (negative)	Inferable from absence
Read time	Some clients	No
Forward to another mailbox	Some clients	No

The asymmetry matters: ISPs have richer data than senders. Aggregate engagement reputation is built from signals you cannot directly observe.

The Apple MPP problem

Apple Mail Privacy Protection (since 2021) pre-loads images on behalf of Apple Mail users from Apple's proxy infrastructure. This means:

• Open tracking pixels load automatically when mail arrives at the Apple Mail user
• Apparent open rate inflates by 30-50% for Apple-heavy audiences
• "Opens" don't represent actual reads

Mitigations:

• Treat opens as a directional signal, not absolute
• Compare opens across cohorts with similar Apple/Android/web distribution
• Weight clicks more heavily than opens
• Use reply rate as the most reliable engagement signal (for outreach)

See Apple Mail Privacy Protection.

Practitioner note: Apple MPP broke open-rate-based filtering inside ESPs. Sunset rules based on "no opens in 90 days" now drop subscribers who actually do open in Apple Mail. The fix: combine opens with clicks for active determination. A subscriber who hasn't opened OR clicked in 90 days is genuinely inactive. Opens alone overcounts MPP-inflated "engagement."

The signals that matter most

In rough order of weight in ISP filtering decisions:

1. Per-recipient interaction history — Has this recipient engaged with this sender before? Very high weight.
2. Aggregate engagement ratio — Domain-wide engagement signals over time. Very high.
3. Recent send engagement — Last 7-30 days of engagement. High.
4. Complaint rate — Very high. Hard threshold at 0.3% (Gmail/Yahoo).
5. Click rate — Real engagement, not contaminated by MPP.
6. Reply rate — Strong positive signal, especially for B2B.
7. Open rate — Contaminated by MPP but still tracked.
8. Bounce rate — Negative signal, threshold-based.
9. Spam trap hits — Negative, can trigger blocklist.

How to improve aggregate engagement

The most impactful interventions:

Sunset inactive subscribers

Removes dead weight from denominator. Engagement rate jumps immediately. ISPs see improved aggregate even though absolute engaged count doesn't change.

Recommended threshold: drop subscribers who have not opened OR clicked in 12 months, after one re-engagement attempt. See sunset policies guide.

Reduce frequency for low-engagement segments

Subscribers who haven't engaged recently are unlikely to engage with the next send and likely to ignore or complain. Reduce frequency by 50-75% for low-engagement segments.

Segment by recency

Send different content treatments to recently engaged vs lapsing subscribers. Highly engaged tolerate more frequency; lapsing need re-engagement style.

Improve content-subject alignment

If subject line promises one thing and content delivers another, subscribers stop opening. Audit subject lines for honesty.

Implement preference centers

Let users choose frequency and topics instead of unsubscribing. Per-topic granular subscriptions produce higher engagement per send.

Run pre-sunset re-engagement

One-time "we miss you" sequence to subscribers approaching sunset. Recovers 5-15%; drops the rest. See list cleaning guide.

A simple engagement segmentation

For most programs:

Segment	Definition	Treatment
Highly engaged	Opened OR clicked in last 30 days	Full frequency, all content
Engaged	Opened OR clicked 30-90 days ago	Standard frequency
Lapsing	Opened OR clicked 90-180 days ago	Reduced frequency, relevant content
Inactive	No engagement 180+ days	Re-engagement attempt, then sunset

Apply consistently. Document the policy.

What to report on

Track engagement at three levels:

1. Per-send: Open, click, unsubscribe rate for the campaign
2. Per-cohort: Engagement by segment, source, signup date
3. Aggregate: Domain-wide trend over 90-day windows

The aggregate trend predicts deliverability. Per-send numbers are noisy.

Practitioner note: I work with clients whose per-send open rates look fine (25%+) but whose aggregate engaged-subscriber percentage has dropped from 60% to 35% over a year. The next deliverability crisis is baked in even though the per-send dashboard looks healthy. Track aggregate active percentage as a leading indicator.

Engagement vs deliverability

The chain is direct:

Better engagement → better ISP reputation → more inbox placement →
More opens and clicks → better engagement metrics → reinforcing loop

The reverse loop is equally direct and equally hard to break:

Worse engagement → worse reputation → more spam placement →
Even fewer opens → worse engagement → reinforcing loop

This is why early intervention matters. Once the negative loop starts, recovery requires deliberate disengaged-subscriber removal — the opposite of what most senders' instincts (and stakeholders' KPIs) push for.

For broader context see email engagement signals, email marketing metrics guide, and Google Postmaster Tools guide.

If you need help designing an engagement-based segmentation, sunset policy, or recovery from declining engagement, book a consultation. I work with senders on engagement repair weekly.

Sources

• Google Postmaster Tools Help
• Apple Mail Privacy Protection Documentation
• Google Email Sender Guidelines
• M3AAWG Sender Best Common Practices
• Litmus: State of Email Engagement Report
• RFC 8058 — One-Click Unsubscribe

---

v1.0 · May 2026