Why Passing SPF, DKIM and DMARC Is No Longer Enough
For most Amazon SES senders, the deliverability checklist looks like this: configure SPF, publish a DKIM key, set DMARC to at least a monitoring policy, keep hard bounces below two percent, and watch complaint rates. That work matters. Without it, mail gets rejected outright. But in 2026, passing every authentication check and staying within SES's sending limits is no guarantee of inbox placement. Authenticated mail lands in spam every day, and the primary reason is almost always the same: poor engagement history.
Mailbox providers have separated two decisions that many senders treat as one. The first is whether to accept a message at all. Authentication and bounce rates govern that gate. The second is where to place the message once accepted. Passing DMARC gets you through the door; the actual placement decision happens downstream, based on how recipients behave when your mail arrives. A fully authenticated sender with a disengaged list will still miss the inbox consistently. Completing the technical setup and assuming the work is done is exactly where most SES senders stall.
How Mailbox Providers Score Sender Reputation
Sender reputation in 2026 is a two-layer problem. The first layer is technical hygiene: valid authentication records, stable infrastructure, consistent sending cadence, and complaint and bounce rates held within acceptable thresholds. This layer is well documented and most serious senders have it covered.
The second layer is behavioural. Gmail's filtering infrastructure and Microsoft's Junk Mail filter have both moved towards AI-driven models that weight dynamic behavioural signals over static sender reputation. IP reputation and domain age still matter, but their weight as standalone signals has declined. What filters now emphasise is how recipients actually interact with mail: reply rates, deletion-without-open patterns, spam complaint trajectory, and move-to-inbox actions from users who first received mail in the spam folder.
Domain reputation has also become more influential than IP reputation for the majority of senders. On shared SES infrastructure, your sending domain is your primary reputation asset. Every engagement event recorded against your domain contributes to a rolling picture that providers update continuously. Reputation builds gradually but deteriorates quickly, and a single campaign that generates poor engagement signals can influence how subsequent sends are treated for weeks.
The Five Engagement Signals Providers Weigh
Mailbox providers gather engagement data through their own clients and fold the resulting signals into their reputation models. Understanding what each signal communicates is essential to managing it.
Opens
An open signals that the subject line was compelling enough to prompt interaction and that the address is active. Providers treat a consistent pattern of opens as a positive trust indicator. However, as discussed below, raw open counts have become significantly less reliable as a measure of genuine human engagement.
Clicks
A click is the strongest positive engagement signal available to most senders. Unlike an open, a click requires deliberate user action that automated systems cannot replicate in the same way. Providers treat click activity as strong evidence that a recipient finds the mail valuable. From a reputation perspective, click rate is the metric that matters most.
Replies
Direct replies carry significant weight, particularly at Gmail and Outlook. A reply places your domain in the recipient's sent history and signals an established relationship. Transactional senders rarely see replies, but if you send newsletters or onboarding sequences, reply rate is a meaningful signal worth monitoring and actively encouraging through copy that invites a response.
Deletions Without Reading
When a recipient deletes a message without opening it, that is a mild negative signal. When it happens repeatedly across a large proportion of your list, it tells providers that recipients recognise your sender name and choose to discard the mail without engaging. This pattern accumulates quietly and is one of the harder signals to detect without proper event tracking.
Spam Reports
Spam complaints are the most damaging engagement signal of all. Google's published target is to keep your spam rate below 0.1%, with 0.3% representing the hard ceiling at which Gmail begins active blocking. Deliverability practitioners generally treat 0.08% as the true safe operating threshold, giving a meaningful buffer before automated enforcement begins. A spike in complaints teaches a provider's model that future mail from your domain is risky, and that assumption does not reset automatically. SES forwards complaint notifications via the Amazon SNS feedback mechanism, and monitoring them in near real time is non-negotiable.
Why Raw Open Rates Have Become Dangerously Misleading
Apple's Mail Privacy Protection, introduced with iOS 15 in September 2021, has fundamentally changed what an open event represents. When a subscriber has MPP enabled, Apple's servers pre-fetch all remote content in an email, including tracking pixels, at the point of delivery rather than when the user reads the message. Every email delivered to an Apple Mail user with MPP active registers as opened regardless of whether the recipient ever looks at it. MPP is an opt-in feature, but adoption is high enough that its effect on aggregate open data is substantial.
The scale of the distortion is significant. Apple Mail accounts for approximately half to nearly three-fifths of global email-client usage depending on the audience, and a large proportion of those users have MPP active. Campaigns that tracked open rates in the high twenties before MPP now routinely show rates in the low fifties with identical click rates and conversions. Nothing has changed except the number on the dashboard.
Other providers add further noise. Google routes all image requests, including tracking pixels, through its own proxy servers, which obscures IP addresses, device types and geolocation. This proxy layer does not generate machine opens in the same systemic way that MPP does, but it does introduce timing inconsistencies and attribution gaps that reduce the reliability of open data. Yahoo and Microsoft have adopted comparable image-caching behaviours.
The practical consequence for SES senders is this: if your engagement strategy relies on open rates to identify active subscribers and suppress disengaged ones, you are almost certainly retaining a large number of recipients who have not read your mail in months. You are mailing a list that appears healthy by one metric while slowly degrading your domain reputation through the accumulation of non-engagement signals that providers can see even when you cannot.
Building a Trustworthy Engagement Picture From SES Event Data
Because opens are unreliable, clicks become the ground truth for engagement decisions. Unlike open events, click events require a deliberate action from the recipient. Amazon SES tracks clicks by replacing links in the email body with redirect URLs hosted on an SES-operated domain. When a recipient follows a link, SES records the event and immediately forwards the recipient to the intended destination. That click event is then published to whichever event destination you have configured, whether Amazon SNS, Amazon CloudWatch, Amazon Kinesis Data Firehose or a third-party tool.
To capture open and click events in SES, you create a configuration set and add an event destination specifying the event types you want to publish. Within the event types panel you can select opens and clicks individually. For click tracking, SES modifies each link in the email body at send time, so no changes to the email template are required beyond ensuring the configuration set is attached to each send. You can optionally configure a custom subdomain for tracking links, which removes SES-branded redirect URLs from the message body and creates a more consistent experience for recipients.
With click events flowing into your event destination, you can build an engagement record keyed to each recipient address. The fields that matter for reputation management are the date of the most recent click, the total number of clicks across a defined window, and the ratio of sends to clicks over that window. These three data points, combined with complaint and bounce events from the same stream, give you everything you need to classify your list by engagement tier.
SES Monitor aggregates this event stream automatically, presenting open and click activity per recipient alongside complaint and bounce data, so that engagement tiers are visible without requiring you to build and maintain your own event processing pipeline.
The Engagement Decay Curve
Disengaged subscribers do not damage your reputation the moment they stop clicking. The harm accumulates over time as the proportion of your send volume that generates zero positive signals grows. Providers observe engagement across rolling windows, weighting recent behaviour more heavily than historical activity.
A useful mental model is to think of your list as having a biological decay curve. Subscribers who were once active gradually transition through recognisable stages: they open less frequently, then stop opening entirely, then begin deleting without reading, and eventually the address becomes abandoned, at which point it may be recycled into a spam trap. Each stage carries higher reputation risk than the last. Without click-based tracking, a sender relying on open rates will not notice this decay until deletions and spam traps appear, by which point the reputation damage is already significant.
The consensus among deliverability practitioners is that a window of 90 to 180 days without any click or confirmed engagement event is the threshold at which a subscriber should be moved to a reduced-frequency or re-engagement track. Allowing disengagement to persist beyond six months without intervention risks meaningful domain reputation degradation. Email lists decay by approximately 20 to 30 percent each year through natural attrition, meaning that without active hygiene, dead weight accumulates faster than most senders realise.
Segmenting by Engagement Tier in Amazon SES
A practical engagement segmentation model for SES senders uses four tiers, applied to the click-based engagement record described above.
Active
Clicked within the last 60 days. Send at full cadence. This cohort is your reputation asset. Prioritise sending to this segment first whenever you are establishing a new sending pattern or recovering from a deliverability dip, as strong early engagement signals improve placement for subsequent waves.
Cooling
No click in 61 to 120 days. Reduce frequency by approximately half. Review subject lines and content relevance for this cohort. A subscriber who has cooled may still be reading but not clicking, or may have changed their behaviour due to inbox overload rather than genuine disinterest.
Dormant
No click in 121 to 180 days. Move to a structured re-engagement campaign and suspend regular sends. This cohort represents meaningful reputation risk if mailed at full volume.
Dead
No click in over 180 days, or no response to a re-engagement attempt. Suppress from ongoing sends. Move to the SES account-level suppression list or maintain a dedicated suppression table in your application database. Do not delete the record entirely; keep it flagged so that any future re-subscription goes through a confirmed opt-in process before regular sending resumes.
Re-engagement Campaigns: Structure, Timing and the Decision to Suppress
A sunset policy is a systematic process for managing subscribers whose engagement has diminished over time. It is distinct from one-time list cleaning. Where list cleaning removes bad addresses, a sunset policy gradually reduces sending to disengaged contacts, attempts re-engagement, and only then suppresses non-responders. The order matters. Removal is the final step, not the first. Jumping straight to deletion discards contacts who were winnable, while never sunsetting at all lets dead weight accumulate until it damages your entire programme.
For dormant subscribers, send a sequence of two or three messages over ten to fourteen days. Keep the content honest and direct. Acknowledge that you have noticed they have not engaged recently, offer a clear value proposition for staying on the list, and make it straightforward to update preferences or unsubscribe. If a subscriber does not click anything across the re-engagement sequence, suppress them. A non-response is itself a signal, and continuing to send is a net negative for your domain reputation.
Time re-engagement sends carefully. Do not batch your entire dormant segment into a single send. Spreading sends across several days reduces the risk of a sudden engagement-rate dip affecting your active campaigns, which may share the same sending domain.
Configuring SES Event Destinations to Capture Engagement in Real Time
The foundation of everything above is a correctly configured event publishing setup. In the Amazon SES console, navigate to Configuration Sets and create or select the configuration set attached to your sending. Open the Event Destinations tab, choose Add Destination, and select the event types you want to publish. For engagement management, you need at minimum: opens, clicks, hard bounces and complaints. Delivery events are useful for diagnosing delivery delays but are lower priority for engagement work.
For the destination type, Amazon SNS is the most flexible choice if you want detailed per-event logs. CloudWatch suits teams that want aggregated counts and threshold alarms without building downstream processing. Kinesis Data Firehose is appropriate for high-volume senders who need to persist events to a data warehouse for longer-term analysis. A configuration set can have multiple event destinations simultaneously, so you can publish to both SNS and CloudWatch from the same configuration set if you need both detailed logs and aggregate monitoring.
SES Monitor connects directly to this event stream, surfacing open and click events per recipient in a dashboard built specifically for deliverability management. Rather than building and maintaining a custom SNS subscriber, Lambda function and database schema to record engagement history, SES Monitor handles the event ingestion and gives you the per-address engagement records you need to operate the four-tier segmentation model described above.
Practical Suppression Rules: Thresholds and Automation
Operationalising suppression requires clear thresholds and a mechanism to act on them without manual review of every address. A reasonable baseline ruleset for SES senders is as follows.
Hard bounce on any send: suppress immediately and permanently. SES adds these addresses to the account-level suppression list automatically, but ensure your application database reflects the suppression so that the address cannot be re-added through a web form without going through verified opt-in.
Complaint received via feedback loop: suppress immediately. Do not wait for a second complaint. The address has already generated a negative signal against your domain.
No click in 180 days following a re-engagement attempt: suppress from all regular sends. Mark the address as dormant in your database for record-keeping.
Avoid over-pruning by applying suppression windows to click activity rather than open activity. Because opens are inflated by MPP and proxy pre-fetching, a subscriber who appears to open every email but never clicks is more likely disengaged than active. Click-based thresholds produce suppression lists that reflect genuine behaviour rather than automated pixel fires.
Monitoring Inbox Placement After Engagement-Based List Hygiene
After suppressing a significant dormant cohort, expect to see your aggregate engagement rates improve within two to four weeks as the denominator of your sends shifts towards genuinely active recipients. Complaint rates typically drop in parallel. The more meaningful indicator, inbox placement rate, takes slightly longer to reflect the improvement because providers update their reputation models on rolling windows rather than instantly.
Use Google Postmaster Tools to monitor domain reputation band changes for Gmail traffic. Postmaster Tools reports reputation across four bands: Bad, Low, Medium and High. A move from Low to Medium following list hygiene is a measurable signal that the intervention is working. For Outlook and Yahoo traffic, Junk Mail Reporting Programme data and Smart Network Data Services signals serve a similar purpose, though they are less granular.
Within SES Monitor, track the ratio of click events to send events over time as your primary engagement health metric. A rising click-to-send ratio after suppression confirms that your active audience is proportionally larger and more responsive. Combine this with complaint rate trends from the SES console to maintain the two-layer view: technical hygiene plus behavioural health, monitored together.
The Engagement Reputation Audit Every SES Sender Should Run Quarterly
Authentication is the entry ticket. Engagement is what keeps you in the inbox. Most SES senders have invested seriously in the first and almost nothing in the second. Closing that gap is not technically complex, but it requires treating your list as a living reputation asset rather than a static dataset to send to.
Run the following audit every quarter. First, confirm that open and click events are flowing from all configuration sets into a destination you can query per recipient. Second, classify your list into the four engagement tiers using click timestamps. Third, pull your dormant and dead segments and run the suppression or re-engagement process described above. Fourth, check your complaint rate in the SES console and your domain reputation in Google Postmaster Tools. Fifth, review inbox placement trends across major providers.
Done consistently, this process shifts the composition of your active list towards recipients who genuinely engage, which is precisely the signal that mailbox providers use to decide whether your mail belongs in the inbox. Technical hygiene gets you through the door. Behavioural signals determine whether you stay there.