Engagement-Based
Email Segmentation

How Gmail and Yahoo Use Engagement to Score Domains

Gmail's spam filtering algorithm is not a simple content scan. It is a per-sender reputation system that evaluates, among many signals, how your specific subscribers respond to your specific emails. The primary question it asks: do people who receive email from this domain engage with it, or do they ignore it, delete it, or report it as spam?

The engagement scoring model ISPs use is per-sending-domain, not per-ESP. Every email you send from your domain contributes to that domain's reputation history at Gmail and Yahoo. Your domain builds its own score entirely independently of the thousands of other senders on your ESP's infrastructure. This is both empowering and sobering: your reputation is yours to build or damage.

Positive engagement signals include recipients opening emails, clicking links, forwarding messages, replying to campaigns, and moving emails from the spam folder into the inbox. Negative signals include recipients deleting without opening, marking as spam, and — most damaging — actively ignoring a sender's emails after previously engaging with them. Gmail interprets declining engagement as a signal that the subscriber no longer wants the email, and adjusts filtering accordingly.

Yahoo's complaint system is slightly more direct: it publishes Feedback Loop (FBL) complaint events that let senders see exactly which messages generated spam reports. Both ISPs converge on the same conclusion: a sender whose recipients engage positively deserves inbox placement; a sender whose recipients ignore or report their email does not.

What Counts as Real Engagement (and What Doesn't)

Before you can build an engagement-based segmentation model, you need an accurate definition of engagement. The problem is that two of the most widely tracked email metrics — open rates and click rates — have become significantly less reliable as signals of human attention since 2021.

Opens: Unreliable Since Apple MPP

Apple's Mail Privacy Protection (MPP), launched with iOS 15 in September 2021, pre-fetches email content including tracking pixels for all Apple Mail users — registering an "open" event even if the recipient never actually reads the email. Depending on your audience, Apple Mail users can represent 40-60% of your list. If half your list uses Apple Mail, half your "open" data does not represent human attention.

Security scanning tools compound the problem. Enterprise email security systems from Microsoft (SafeLinks), Barracuda, Cisco IronPort, and Proofpoint load email content to check for malware, often registering open events as a side effect. These automated opens provide zero positive reputation signal at Gmail — they are not a human choosing to engage with your email.

Clicks: Partially Reliable, with Caveats

Click data is more reliable than open data, but it is not immune to automation. The same security scanning tools that pre-fetch images also click links to check for malware. These bot clicks have specific fingerprints: they occur within seconds of delivery, they come from data center IP addresses, and they click every link in the email systematically. Human clicks happen later, from residential IPs, and usually focus on one or two links of interest.

Reliable Engagement Signals

• Human clicks — clicks that occurred more than 60 seconds after delivery, from non-datacenter IPs, not on every link simultaneously
• Replies — a subscriber replying to your email is one of the strongest positive reputation signals possible
• Purchases triggered by email — conversion data that confirms the email drove a real human action
• Forwards — a subscriber forwarding your email to someone else is a very strong positive signal
• Moving from spam to inbox — when a subscriber rescues your email from spam, Gmail treats this as a strong positive signal for that sender-recipient pair

Building a 4-Tier Engagement Segmentation Model

A four-tier engagement model divides your list based on recency of real engagement activity. The tiers determine which campaigns each subscriber receives and at what frequency. Applied correctly, this model ensures that your broadcast campaigns reach primarily your highest-engagement subscribers — the ones who give Gmail and Yahoo positive reputation signals when your email arrives.

Tier 1: Active

Definition: Clicked a link or completed a purchase triggered by email within the past 30 days.

This is your healthiest segment. These subscribers have demonstrated recent, real engagement. They are the strongest positive reputation signal you have. Send all campaigns, all automations, with no frequency limits beyond business sense. Monitor their complaint rate — it should be below 0.02%. If it rises above 0.05%, investigate your content or offer relevance rather than your frequency.

Tier 2: Engaged

Definition: Clicked a link or completed a purchase within the past 90 days, but not within the past 30 days.

These subscribers are engaged but their engagement frequency is lower. They are suitable for all standard broadcast campaigns, but apply normal frequency limits (no more than 4 campaigns per week for most senders). Their complaint rate should still be well below 0.10%. If it climbs, reduce frequency or improve relevance before expanding the volume going to this tier.

Tier 3: At-Risk

Definition: Opened (with credible signals — not MPP pre-fetches) within the past 180 days, but has not clicked in 90+ days.

These subscribers show weak positive signals at best. Their engagement is declining, and their complaint rate is likely higher than Tiers 1 and 2. Apply a 50% frequency reduction — if you typically send 2 campaigns per week to Tiers 1 and 2, send 1 per week to Tier 3. Add them to a re-engagement flow that runs parallel to your regular campaigns. If they re-engage within the re-engagement window, move them to Tier 2. If they don't engage with the re-engagement flow, move them to Tier 4.

Tier 4: Inactive

Definition: No credible engagement signal in 180+ days.

Exclude Tier 4 from all regular broadcast campaigns. Send a win-back sequence of 2-3 emails with strong subject lines ("We haven't heard from you in a while," "Last chance before we unsubscribe you"). If they don't engage with the win-back sequence, suppress them from future sends. Do not delete them from your system — keep them in a suppression list so they cannot be accidentally re-imported or re-subscribed without confirmed opt-in.

The 90-Day Engagement Window Explained

The 90-day window is not arbitrary — it reflects how ISPs weight recency in engagement scoring. Gmail's reputation system gives significantly more weight to recent engagement signals than older ones. A click from yesterday matters more than a click from six months ago. The 90-day mark is roughly where the decay in engagement signal value becomes steep enough to affect your deliverability decisions.

For your segmentation model, the practical implication is that subscribers who haven't engaged in 90+ days are beginning to be a reputational liability when mailed at full volume. Their potential positive reputation contribution is low, their potential negative contribution (via complaint or deletion without reading) is growing. The 90-day split between Tier 2 and Tier 3 is where you start applying frequency restrictions.

The 30-day window for Tier 1 reflects a different principle: these are your most recently active subscribers, and they are most likely to generate positive signals on your next send. Identifying them separately allows you to prioritize them for time-sensitive campaigns, VIP offers, or higher-frequency sends where you need maximum reputation insurance.

The 180-day boundary for Tier 4 reflects the point at which continued mailing is likely net-negative for reputation. Research from multiple deliverability consultants consistently shows that subscribers who haven't engaged in six months have complaint rates 5-10 times higher than actively engaged subscribers. Suppressing them from campaigns is almost always the right deliverability decision.

The Bot Open Problem: Why Your Engagement Data May Be Wrong

Here is the uncomfortable reality: if you built your Tier 2 and Tier 3 segments using open data, you have almost certainly over-populated those segments with subscribers who have never actually read your email. Apple MPP inflates opens for all Apple Mail users. Security scanners inflate opens for enterprise users. The result is that your "engaged" segments contain a significant proportion of contacts who are functionally unengaged — they just have automated systems that happen to load your emails.

When you mail those inflated segments as if they were real Tier 2 contacts, you get worse deliverability outcomes than expected. Your complaint rate is higher than your open rate would suggest. Your inbox placement is lower than your engagement data would predict. This is the bot open problem: garbage data produces garbage segments, which produce poor deliverability outcomes.

How to Identify the Inflation

Compare your open rate to your click-to-open rate (CTOR). If your open rate is 45% but your CTOR is 3%, that is a sign that many of your "opens" are not leading to human attention. A healthy CTOR for engaged commercial email is typically 10-25%. CTOR below 5% suggests significant automated open inflation.

Look at your email client breakdown in your ESP analytics. If Apple Mail and webmail clients represent 70%+ of your opens, and you know a large portion of your list is on mobile Apple Mail, MPP is almost certainly inflating your open data substantially.

Switch your Tier 2 and Tier 3 definitions from open-based to click-based, even temporarily, and observe what happens to your segment sizes. If your "engaged in 90 days" segment shrinks by 40% when you switch from opens to clicks, that is approximately how much your segment was inflated by automated opens.

Sending Strategy Per Engagement Tier

The four-tier model produces its greatest deliverability benefit when applied consistently across every broadcast campaign — not just occasionally. Here is how to implement the sending strategy for each tier.

Tier 1 Strategy: Full Campaigns, Priority Placement

Send every broadcast campaign and every automation to Tier 1. When you have a new campaign to deploy, send to Tier 1 first and let it run for 4-6 hours. Check Gmail Postmaster Tools for any sudden reputation changes. If Postmaster shows stable or improving reputation, proceed to send to Tier 2. This "warming" approach protects you if a campaign has a content problem that generates unexpected complaints.

Tier 2 Strategy: Standard Campaigns, Frequency Cap

Send standard broadcast campaigns to Tier 2 alongside Tier 1, subject to a frequency cap. If your usual cadence is 3 campaigns per week, consider 2 per week for Tier 2. Reserve the third weekly send for Tier 1 only. Monitor Tier 2's complaint rate separately if your ESP allows segment-level analytics. If it climbs above 0.05%, reduce frequency to Tier 2 first before making other changes.

Tier 3 Strategy: Reduced Frequency + Re-engagement

Exclude Tier 3 from all high-frequency campaigns. If you mail Tiers 1 and 2 twice a week, send Tier 3 once per week — and only your highest-value, most relevant campaigns. Simultaneously, run a re-engagement automation for everyone in Tier 3. The re-engagement sequence should be 3-4 emails over 2-3 weeks with compelling subject lines, a clear reason to click, and a preference center option for people who want to reduce frequency rather than unsubscribe entirely.

Tier 4 Strategy: Suppression + Win-Back Only

Zero broadcast campaign sends to Tier 4. Run one win-back sequence, 2-3 emails maximum, with approximately a week between each email. The first email should be honest and direct: "We haven't heard from you in a while — are you still interested?" Make the call to action simple (a single click to confirm they still want your emails). If a Tier 4 subscriber clicks, move them immediately to Tier 2 and start rebuilding engagement. If they don't engage with the win-back sequence after the final email, add them to your suppression list.

The Deliverability Impact

The most impactful deliverability move most senders can make is simply not sending to Tier 3 and Tier 4 on every broadcast campaign. Analysis consistently shows that segmenting broadcast campaigns to Tier 1 and Tier 2 can improve inbox placement rate by 15-40% — not through any technical change, just through better segment targeting. Gmail rewards senders whose recipients engage. The best way to ensure your recipients engage is to only send to the ones who do.