Pillar Guide

Email Engagement Metrics That Actually Drive Inbox Placement

Understand how opens, clicks, replies, and complaints shape your deliverability. Learn engagement-based filtering, segmentation strategies, and re-engagement campaigns.

Boxset TeamFeb 20, 202621 min read
engagementopen ratesclick ratesspam complaintsinbox placementdeliverability

Engagement Is the New Reputation

83%

of inbox placement decisions at Gmail are now influenced by recipient engagement signals

For years, email deliverability was an infrastructure problem. Get your SPF and DKIM records right, warm up your IP, keep your bounce rate low, and your emails would reach the inbox. That era is over. Between 2022 and 2025, every major mailbox provider completed a fundamental shift toward engagement-based filtering, and it has changed everything about how deliverability works.

Gmail led this transition. Starting with incremental algorithm updates in 2019 and accelerating through the bulk sender requirements enforced in February 2024, Gmail moved from a model that primarily evaluated sender reputation at the domain and IP level to one that evaluates recipient-level engagement patterns as the dominant filtering signal. In plain terms: Gmail does not just ask "Is this sender trustworthy?" It asks "Does this specific person want this specific email?"

The implications are profound. Two subscribers on the same list, receiving the same email, from the same domain and IP, can experience completely different placement outcomes. Subscriber A, who opens and clicks regularly, sees the email in their Primary tab. Subscriber B, who has ignored the last 15 emails, sees it routed to spam — or never delivered at all. The email is identical. The engagement history is not.

Outlook followed a parallel path. Microsoft's SmartScreen filter has long incorporated individual user behavior, but the weighting increased substantially in 2024-2025. Outlook now evaluates whether a specific user has previously marked emails from your domain as "not junk," moved them out of the Junk folder, or added you to their contacts. These per-user signals can override domain-level reputation in either direction — a strong domain reputation will not save you from spam placement if a specific user consistently ignores your messages.

Yahoo, after co-announcing the 2024 bulk sender requirements with Google, similarly increased the role of engagement in its filtering algorithm. Yahoo's system clusters recipients into engagement cohorts and adjusts filtering thresholds based on how each cohort has historically interacted with a given sender.

This shift from infrastructure-centric to engagement-centric filtering means that list quality and sending strategy now matter more than technical configuration for most senders who already have authentication in place. You can have perfect SPF, DKIM, and DMARC, a warmed dedicated IP, and pristine DNS records — and still land in spam if you consistently send to people who do not engage with your emails. The technical foundation is necessary but no longer sufficient.

Understanding this reality is the starting point for every strategy discussed in this guide: what engagement signals exist, how providers weight them, and what you can do to systematically improve them.

The Engagement Metrics That Mailbox Providers Track

Mailbox providers track a broader set of engagement signals than most senders realize. Your ESP reports opens and clicks. Mailbox providers see everything else — how long someone read your email, whether they deleted it immediately, whether they moved it to a folder, and whether they scrolled to the bottom. Each of these actions feeds into the filtering algorithm.

Here is a comprehensive breakdown of the engagement metrics that influence inbox placement, how heavily each is weighted, and which providers are known to track them:

Several things stand out from this breakdown. First, replies are the most powerful positive signal available. When a recipient replies to your email, it tells the mailbox provider unequivocally that this is wanted communication. One genuine reply carries more filtering weight than dozens of opens. This is why transactional emails and conversational-style emails consistently outperform broadcast-style marketing in inbox placement — they naturally generate replies.

Second, opens are a weaker signal than most senders assume, particularly since Apple's Mail Privacy Protection (MPP) launched in September 2021. MPP pre-fetches email content (including tracking pixels) for Apple Mail users, generating "opens" that did not actually happen. As of 2026, Apple Mail accounts for roughly 55-60% of email client market share. This means that for many senders, a significant portion of reported opens are phantom opens from Apple's proxy servers. Gmail, Outlook, and Yahoo are all aware of this and have adjusted how they weight open signals — relying more heavily on downstream actions like clicks, replies, and read time.

Third, the absence of engagement is itself a signal. When a recipient receives your email and takes no action — does not open it, does not delete it, does not do anything — mailbox providers interpret this as mild disinterest. When this pattern repeats over multiple emails, the cumulative signal of indifference builds, and the provider begins routing your messages to less prominent locations or filtering them entirely. This is why continuing to mail unengaged subscribers is not neutral — it is actively harmful.

Positive vs Negative Engagement Signals

Understanding the asymmetry between positive and negative engagement signals is critical for developing an effective email strategy. They are not mirror images of each other. Negative signals are weighted more heavily and take effect more quickly than positive signals. A single spam complaint does more damage than ten opens can repair.

The positive signals that matter most:

Replies are the gold standard. When someone replies to your email, it creates a bi-directional communication signal that almost guarantees future inbox placement for that recipient. Gmail, in particular, treats replied-to senders as trusted contacts, even without the recipient explicitly adding you to their address book. Marketers who find ways to encourage replies — asking questions, soliciting feedback, using conversational CTAs — see measurable inbox placement improvements. Even automated reply mechanisms (like "Reply YES to confirm") generate this signal.

Moving an email from spam to inbox is the second most powerful positive signal. When a recipient actively rescues your email from the spam folder, it overrides the provider's filtering decision and signals that the algorithm made a mistake. Gmail treats this as a strong correction signal that influences future placement for that specific sender-recipient pair and, in aggregate, can improve placement across your broader subscriber base.

Adding sender to contacts and moving to the Primary tab (in Gmail) are explicit trust signals. They tell the provider that this recipient has consciously decided they want this sender's emails in their most prominent inbox location. These signals persist across sessions and create durable positive filtering effects.

Clicks are a reliable positive signal because, unlike opens, they cannot be easily spoofed by privacy proxies. A click requires deliberate human action and indicates genuine content engagement. Clicks on multiple links within the same email carry additional weight, as they suggest deep engagement rather than an accidental tap.

Read time — the duration a recipient spends viewing your email — is a subtler signal that Gmail and Outlook have increasingly incorporated into their models. An email that is opened and read for 15+ seconds generates a stronger positive signal than one that is opened and closed in under 2 seconds. This metric rewards content quality and relevance.

The negative signals that cause the most damage:

Spam complaints (marking as spam / reporting junk) are the most destructive engagement signal in the email ecosystem. A single complaint is weighted orders of magnitude more heavily than a single open. We cover this in depth in the next section, but the key point here is that complaints do not just affect placement for the complaining recipient — they contribute to your domain-level reputation and affect placement for all recipients.

Delete-without-reading is a signal that providers can detect when a recipient deletes your email without opening it (or opens it for under 1-2 seconds and immediately deletes). This pattern, when repeated across multiple emails, builds a strong negative engagement profile. Gmail is particularly sensitive to this signal and uses it to gradually shift a sender's emails toward spam for that recipient.

Sustained ignoring — receiving multiple emails from a sender and never interacting with any of them — creates cumulative negative pressure. Providers interpret this as the recipient having lost interest (or never having had interest) in the sender's content. After enough ignored emails (the threshold varies by provider but is typically 5-15 consecutive sends), the provider begins downgrading placement.

Gmail organizes engagement data into what deliverability experts call "engagement clusters." Rather than evaluating each recipient in isolation, Gmail groups recipients with similar engagement patterns and applies cluster-level filtering adjustments. If a large cluster of your recipients consistently ignores your emails, the negative signal from that cluster can spill over and affect inbox placement for recipients in adjacent clusters — even those who have historically engaged. This is the mechanism by which sending to unengaged subscribers damages deliverability for your engaged subscribers. It is also why engagement-based segmentation is not optional; it is a structural requirement for maintaining inbox placement at Gmail.

Spam Complaint Rate: The Most Dangerous Metric

Of all the engagement metrics mailbox providers track, spam complaint rate occupies a unique position: it is the only metric with a publicly stated, hard enforcement threshold that triggers automatic filtering. It is also the metric most likely to cause sudden, catastrophic deliverability collapse.

Google's guidelines are explicit and unambiguous. Senders must maintain a spam complaint rate below 0.10% (1 complaint per 1,000 emails delivered to the inbox). Exceeding 0.10% generates a warning-level response — your domain reputation in Google Postmaster Tools may downgrade, and you will experience intermittent spam filtering. Exceeding 0.30% (3 complaints per 1,000) triggers severe filtering: the majority of your Gmail-bound email will be routed to spam, and recovery can take 30-90 days of corrected sending behavior.

Exceeding Google's 0.30% spam complaint rate threshold can result in bulk filtering of all your email to spam across Gmail's entire user base. This is not a gradual degradation — it can happen within 24-48 hours of crossing the threshold. Once triggered, recovery requires reducing volume to engaged-only segments, maintaining complaint rates below 0.05% for weeks, and waiting for Google's reputation models to recalculate. There are no expedited recovery paths, no support tickets to file, and no way to buy your way back. Prevention is the only viable strategy.

Yahoo enforces a similar threshold, though their published guidance places the ceiling closer to 0.30%. Microsoft does not publish a specific number but their Junk Mail Reporting Program (JMRP) data and delisting requirements suggest a practical threshold in the 0.20-0.40% range. Regardless of provider-specific thresholds, best practice is to target a complaint rate below 0.05% across all providers.

How to track complaint rate accurately:

The most reliable source for Gmail complaint data is Google Postmaster Tools, which reports the spam rate directly — the percentage of your emails delivered to the inbox that recipients subsequently marked as spam. This is the metric Google uses internally for reputation calculations, so it is the authoritative source.

For Outlook, enroll in the Junk Mail Reporting Program (JMRP), which sends real-time notifications when an Outlook user marks your email as junk. Processing these notifications and calculating complaint rates against your send volume gives you an accurate picture of your Outlook complaint performance.

For Yahoo and other providers that support the Feedback Loop (FBL) standard (ARF format), register your sending domain or IP and process incoming FBL reports. Yahoo's CFL is one of the most reliable in the industry. Suppress every address that generates a complaint — immediately and across all sending platforms.

Common causes of elevated complaint rates:

The leading cause is not malicious behavior — it is sending to people who do not remember opting in or who have lost interest. A subscriber who signed up eight months ago, received no emails for six months, and then suddenly gets a promotional blast is highly likely to hit "spam" rather than search for an unsubscribe link. Other common causes include unclear or hidden unsubscribe mechanisms, content that does not match the subscriber's expectations (they signed up for product updates and received promotional offers), purchase or third-party lists (where opt-in was never granted), and sending frequency that exceeds what the subscriber anticipated.

The fix is structural: implement visible one-click unsubscribe headers (required by Gmail and Yahoo since 2024), honor unsubscribe requests within 48 hours, set clear content and frequency expectations at opt-in, and segment by engagement so that dormant subscribers are not subjected to the same cadence as active ones.

Segmentation Strategies for Better Engagement

Engagement-based segmentation is the single highest-leverage strategy available for improving inbox placement. It is conceptually simple — send more frequently to people who engage, less frequently to people who do not — but the implementation details determine whether it works or backfires.

The foundation of engagement segmentation is categorizing your list into tiers based on recency, frequency, and monetary value of interactions — a framework borrowed from direct marketing known as RFM segmentation. For email specifically, the relevant dimensions are recency of last engagement (open, click, or reply), frequency of engagement over a trailing period, and depth of engagement (click vs open vs reply).

The percentages above will vary by industry, sending frequency, and list age, but the pattern is consistent across virtually every email program: a significant minority of your list drives the vast majority of your engagement. The temptation is to keep mailing the full list because "they might open eventually." This thinking is precisely what causes deliverability problems. Every email sent to a Dead-tier subscriber generates a negative engagement signal (ignore or delete-without-reading) that degrades your domain reputation for everyone else.

Implementing tiered segmentation operationally:

Start by defining your engagement events and their recency thresholds. Opens are the weakest signal (and unreliable on Apple Mail), so use clicks as your primary engagement indicator when possible. If a subscriber has clicked a link in the last 30 days, they are Active. If they have opened but not clicked in the last 30 days, they are marginally Active. If their last click was 45 days ago but they opened 20 days ago, they are Lapsing.

Build these segments dynamically in your ESP or marketing automation platform. Most modern platforms (Klaviyo, HubSpot, Customer.io, Braze) support time-based engagement segments natively. Update segment membership daily or in real-time based on engagement events.

Adjust your sending calendar so that Active subscribers receive your full cadence, Lapsing subscribers receive your best-performing campaigns only (based on historical open and click rates), Dormant subscribers receive only a structured re-engagement sequence, and Dead subscribers are suppressed from all regular sends.

Frequency optimization within tiers:

Even within your Active tier, sending frequency requires calibration. The optimal frequency depends on your content type, industry, and subscriber expectations. An e-commerce brand with daily deals can sustain higher frequency than a B2B SaaS company sending thought leadership content. The universal principle is that frequency should match value: every email must deliver value to the recipient. The moment frequency exceeds the value your content provides, engagement drops and complaints rise.

Monitor your unsubscribe rate and complaint rate at each frequency level. If increasing from 3x/week to 5x/week causes your unsubscribe rate to jump from 0.1% to 0.4%, the additional two sends are destroying more value than they create. Find the frequency ceiling where incremental sends still produce positive engagement, and hold there.

Re-Engagement Campaigns: Win Back or Let Go

Re-engagement campaigns are your last opportunity to salvage relationships with subscribers who have stopped interacting with your emails. Done well, they recover 5-15% of lapsing and dormant subscribers. Done poorly — or not done at all — they leave dead weight on your list that drags down deliverability for everyone.

The fundamental question a re-engagement campaign answers is: "Does this person still want to hear from us?" If the answer is yes, you reactivate them into your engaged segments with renewed interest. If the answer is no — or if they do not answer at all — you remove them from your list. Both outcomes are positive. The only bad outcome is doing nothing and continuing to mail people who do not engage.

A critical mistake many senders make is running re-engagement campaigns but never following through on the sunset. They send the three emails, see a 10% re-engagement rate, and then... keep mailing the other 90% anyway. This defeats the entire purpose. The value of a re-engagement campaign comes equally from recovering interested subscribers and from removing uninterested ones. If you are not prepared to suppress non-responders, do not run the campaign — you will generate complaints from dormant subscribers who are annoyed by the additional emails without gaining any of the list hygiene benefits.

Send Time and Frequency Optimization

When you send matters almost as much as what you send. A perfectly crafted email that arrives at 3:00 AM when the recipient is asleep competes with a full inbox of messages by the time they wake up. The same email delivered at 10:15 AM when they are checking email between meetings has a dramatically higher chance of being seen, opened, and engaged with.

Send Time Optimization (STO) algorithms analyze each subscriber's historical engagement patterns to predict the time of day and day of week when they are most likely to engage. Rather than sending a campaign to your entire list at 10:00 AM, STO distributes sends across a window — perhaps 6:00 AM to 2:00 PM — with each subscriber receiving the email at their predicted optimal time. Most enterprise ESPs (Braze, Salesforce Marketing Cloud, Iterable) and several mid-market platforms (Klaviyo, Customer.io) offer STO capabilities natively.

The engagement lift from STO is real but modest: most studies show a 5-15% improvement in open rates and a 3-10% improvement in click rates. The compounding effect on deliverability, however, is larger than the direct engagement lift suggests. Higher open and click rates generate stronger positive engagement signals, which improve inbox placement, which leads to higher visibility, which drives further engagement improvements. The virtuous cycle amplifies the initial STO gain over time.

Frequency fatigue is the inverse problem. Every additional email you send within a given timeframe has diminishing marginal engagement and increasing marginal complaint risk. The first email of the week might achieve a 25% open rate. The third email in the same week might achieve 12%. The fifth might achieve 6% with a complaint rate twice your baseline. The incremental revenue from emails four and five may not cover the deliverability damage they cause.

Pro Tip from Boxset Team

Implement frequency caps at the subscriber level, not the campaign level. Rather than asking "How many campaigns should we send this week?" ask "How many emails should any individual subscriber receive this week?" A frequency cap of 3-4 emails per week per subscriber (across all campaign types — promotional, editorial, triggered) prevents fatigue without requiring you to reduce your overall campaign calendar. Subscribers who hit the frequency cap simply skip the lowest-priority send, ensuring they receive your best content without being overwhelmed. Most marketing automation platforms support per-subscriber frequency capping as a built-in feature.

Preference centers offer the most sustainable solution to frequency optimization: let subscribers choose how often they want to hear from you. A well-designed preference center allows subscribers to select their preferred frequency (daily, weekly, monthly), the content types they want (product updates, promotions, educational content), and their preferred channel (email, SMS, push). Subscribers who control their own experience complain less, engage more, and remain on your list longer. Preference centers also reduce unsubscribes by offering a "receive less" option as an alternative to "receive nothing."

The operational key is to actually honor preference center selections in your sending logic. A surprising number of senders collect preferences but do not enforce them in their campaign workflows, sending daily emails to subscribers who selected weekly. This is worse than not having a preference center at all, because it violates an explicit promise and drives complaints.

How Boxset Tracks Engagement Across ESPs

Most businesses do not use a single email platform. A typical email operation might use SendGrid for transactional email, Klaviyo or HubSpot for marketing campaigns, Customer.io or Braze for product-triggered messages, and a separate tool for warm-up sends. Each of these platforms reports its own engagement metrics in its own dashboard, using its own definitions, its own time zones, and its own methodology for calculating rates.

The result is fragmentation. Your SendGrid dashboard shows a 22% open rate. Your Klaviyo dashboard shows a 28% open rate. Are these comparable? Are they measuring the same thing? Is one platform performing better, or are they sending to different segments with different engagement baselines? Without a unified view, these questions are unanswerable.

Boxset solves this by aggregating engagement data from all connected ESPs into a single, normalized analytics layer. When you connect your SendGrid, Mailgun, Amazon SES, Postmark, Klaviyo, HubSpot, or any supported platform, Boxset pulls event-level data — sends, deliveries, opens, clicks, bounces, complaints, unsubscribes — and normalizes it into a consistent schema. This means you can compare open rates across ESPs on an apples-to-apples basis, identify which platform delivers the best engagement for which audience segment, and spot anomalies that are invisible in siloed dashboards.

The Seltra Score — Boxset's proprietary engagement health metric — synthesizes engagement data across all your ESPs into a single, actionable number. It weights positive signals (clicks, replies) more heavily than opens, accounts for Apple MPP inflation, penalizes complaint rate trends, and factors in engagement velocity (is engagement improving or declining week over week?). A declining Seltra Score triggers proactive alerts before the downstream effects — reputation drops, increased spam filtering — become visible in mailbox provider tools.

Unified Engagement Intelligence Across All Your ESPs

Boxset's Seltra Score aggregates engagement data from SendGrid, Mailgun, Amazon SES, and more into a single health metric — with alerts that catch problems before they reach your inbox placement.

See Your Seltra Score

Cross-ESP engagement tracking also reveals a critical insight that siloed monitoring misses: per-contact engagement divergence. The same subscriber can receive emails from your marketing platform and your transactional platform. If they engage with transactional emails (password resets, order confirmations) but ignore marketing emails, that divergence is a signal that they want your service but not your marketing content. Boxset surfaces these patterns so you can make smarter segmentation decisions — perhaps moving that subscriber to a transactional-only cadence rather than continuing to send marketing emails they will never open.

Frequently Asked Questions

Discover Your Seltra Score

Get a single number that reflects the health of your entire email operation. Boxset's Seltra AI analyzes data from all your ESPs to calculate your real deliverability score.

Check Your Score Free

More in Email Engagement