How Inbox Filtering Actually Works in 2026

A decade ago, inbox filtering was a binary decision. The email arrived or it did not. Spam filters did most of the work and the inbox was a single, undifferentiated pile.

That is no longer how any major provider operates. Gmail, Outlook, and Apple Mail now make a routing decision on every email: not whether to deliver it, but where inside the inbox to put it. Primary or promotions. Focused or other. Inbox or a folder the subscriber rarely opens. The choice is made in milliseconds, on every subscriber, on every send.

Understanding how that routing works is no longer optional. It is the difference between a campaign that reaches engaged subscribers and one that quietly lands in a folder nobody checks.

This piece covers what each major provider weighs heaviest in 2026, the signals senders can actually influence, and where Mail Privacy Protection has shifted the conversation.

Gmail: engagement-led routing

Gmail handles the largest share of consumer email globally. Its routing model has, for several years now, been heavily engagement-led.

Gmail does not just look at whether a subscriber opened your last email. It looks at the pattern across many sends from you to that subscriber. Has this person opened? Clicked? Replied? Starred? Moved your emails to a folder? Marked them as important? Or have they consistently scrolled past, archived without opening, or, worst of all, marked them as spam?

These signals decide whether your email lands in the primary inbox, the promotions tab, or, in increasingly common cases, the spam folder despite passing every technical check.

Two specific Gmail behaviours are worth understanding:

  • Tabbed inbox routing. The primary versus promotions split is made per-sender and per-recipient. The same sender can land in primary for one subscriber and promotions for another, based entirely on how each one historically interacts.
  • Spam folder placement based on negative engagement. Gmail will route subsequent emails to spam if a meaningful pattern of negative engagement accumulates, even if no individual subscriber marks the email as spam. The threshold is opaque, but the pattern is consistent.

What influences Gmail routing positively: consistent open and click behaviour from a stable segment of subscribers, reply rates above zero, and senders who maintain a stable cadence rather than going dark for weeks and then sending three campaigns in one day.

Outlook: the focused inbox and reputation-led decisions

Outlook (Microsoft's consumer and business email products) operates differently from Gmail. The focused inbox is the primary routing decision: focused or other. The other tab is where most marketing emails land unless the subscriber has explicitly moved similar emails to focused.

Outlook weights three things heavily:

  • Sender reputation calculated across the wider Outlook user base, not just per-subscriber. A sender with poor reputation across millions of inboxes will land in other for new subscribers even if those subscribers have shown no negative behaviour individually.
  • SmartScreen-style content filtering, which scrutinises specific words, formatting choices, and link structures more aggressively than Gmail.
  • Authentication strictness, particularly on DMARC alignment. Outlook is less forgiving of misaligned authentication than Gmail.

The focused inbox is also slower to change its mind. A subscriber who has had a sender in "other" for months will not see that sender move to focused unless they manually move several emails first. This means rehabilitating a sender on Outlook takes longer than on Gmail. The compound effect of small improvements still works, but it requires patience.

Apple Mail and the Mail Privacy Protection shift

Apple Mail Privacy Protection (MPP) has been live since 2021 and now applies to the majority of Apple Mail users. Its effect on email marketing has been substantial.

MPP pre-fetches images in every email, regardless of whether the subscriber actually opens the message. From a tracking perspective, every MPP subscriber appears to have opened every email you sent them. Open rates from Apple Mail users are now essentially meaningless as an engagement signal.

Two consequences follow from this:

  • Open rate as a metric is degraded across the entire programme. The proportion of Apple Mail users in any given list will vary, but for B2C senders in many markets, MPP-affected opens can represent 50 to 60 percent of total opens. The number on your dashboard is not a number you can act on confidently any more.
  • Click-through rate, click-to-open rate, conversion rate, and reply rate are now the metrics worth tracking. These are not affected by MPP because they require an actual subscriber action.

The strategic implication is that Apple's routing decisions also lean more heavily on these post-open behaviours than on opens themselves. The signal Apple Mail uses to decide whether your email goes to the inbox or somewhere else is similar to Gmail's, but with opens largely discounted.

What every provider weighs the same way

Across all three providers, four signals carry consistent weight:

  • Spam complaint rate. Even one complaint per thousand emails is meaningful. Five complaints per thousand is a serious problem. The threshold above which deliverability declines sharply is somewhere between 0.1 and 0.3 percent depending on the provider.
  • Hard bounce rate. Sending to a list with a high proportion of invalid addresses tells every provider that the list was not collected carefully. A hard bounce rate above 2 percent on a single send is enough to trigger reputation impact.
  • Sending consistency. Long gaps followed by sudden volume look suspicious. Consistent cadence at the same volume earns trust over time.
  • List growth velocity. A list that doubles in a week without a clear reason looks like a list that was acquired, not built. Inbox providers can tell.

What senders can actually influence

Two things matter most, in this order:

1. Send to subscribers who engage

Every email sent to a subscriber who does not engage is a small negative signal. Every email sent to a subscriber who does engage is a small positive signal. The shape of your active sending list determines the shape of your sender reputation. There is no shortcut around this.

The fix is not better content (though better content helps). The fix is sending to fewer subscribers, more engaged ones. A list of 25,000 subscribers where 80 percent engage will outperform a list of 100,000 where 20 percent engage on every single metric inbox providers measure.

2. Make engagement easy

Emails that are easy to scan, easy to click, and easy to reply to generate more engagement signals. Plain text emails often outperform heavily designed HTML emails for engagement, partly because they look less like marketing. Single calls to action outperform multiple. Short emails outperform long ones for engagement rate (though longer emails can outperform on conversion when the content earns the length).

None of this is news to experienced senders. But the consequence is sometimes missed: every design and content decision that reduces friction increases engagement, which increases deliverability, which compounds into better results on every future send.

Where Mail Blaze fits

Mail Blaze's reporting tools focus on the engagement metrics that have not been degraded by MPP: click-through rate, click-to-open rate, conversion rate, and reply rate. The Subscriber Score also incorporates engagement consistency over time, which is a closer proxy to what inbox providers actually weigh than a single open rate.

Comparative filters let you isolate specific subscriber behaviours and see what is moving deliverability. For instance, comparing engagement from Gmail and Outlook subscribers separately is often illuminating: a deliverability problem in one provider but not the other points clearly to the source.

What to do this week

Look at your reporting from the last three months. Stop looking at the overall open rate. Look at click-through rate, click-to-open rate, and (if you can measure it) reply rate, broken out by provider.

Note which provider is driving the most engagement, and which one is underperforming. The underperforming one is where the deliverability work is needed. The overperforming one tells you something about what is working in your content and approach: replicate it deliberately for the underperforming segment.

Related reading

BACK