Skip to content
Glassjaw
Industry notes 7 min read

Apple's Mail categories and the promotions problem

Mail on iPhone now sorts messages into Primary, Transactions, Updates and Promotions. What tends to land outside Promotions, and what it means for how you measure email.

Abstract cover art: concentric arcs around a bright focal point on a dark blue field

A client's head of ecommerce forwarded us a screenshot of her own phone last month. Her order confirmation from a competitor sat in a tab called Transactions. The competitor's sale email sat in Promotions, four screens away. Her question was the one every CRM team has been asking since: which of ours goes where, and can we influence it?

During the iOS 18 cycle Apple added category tabs to Mail on iPhone, sorting incoming messages into Primary, Transactions, Updates and Promotions. It is not a new idea. Gmail has sorted mail into tabs for more than a decade, and a generation of email marketers has already been through the panic, the workarounds and the eventual acceptance.

The useful conversation is not about escaping Promotions. It is about understanding what actually falls outside it, and about accepting that another piece of your reporting has just become harder to read.

Categorisation is sorting, not filtering

The first thing to be clear about with anyone senior: a category tab is not the spam folder. A message placed in Promotions has been delivered to the inbox. It is retrievable, searchable, and visible to anyone who taps that tab.

What changes is timing and attention. Messages in a secondary tab tend to be read in a batch, later, when someone chooses to look. That shifts your open and click curve rather than removing it, and it tends to hit time-limited offers harder than anything else. A one-day promotion read in an evening sweep is a promotion that expired unread.

It also changes the competitive set. In a single undifferentiated inbox, your sale email competes with a colleague's meeting request. In a Promotions tab it competes with every other retailer, all of whom send at similar times of day.

Gmail already ran this experiment

When Gmail introduced tabs, the widely predicted collapse in marketing performance did not happen. Reported opens dipped for many senders and then recovered as people adapted their habits. The brands that suffered most were the ones whose whole model depended on being mistaken for personal mail.

Two lessons carried over, and both apply now.

The first is that trying to force your way into the primary tab is a losing strategy. The techniques that work briefly are the ones that look deceptive to a filter and to a human. Stripped-down templates disguised as plain text. Subject lines pretending to be a reply. Promotional content dressed up as a service notice. They get reclassified, and the reclassification is not the worst outcome. A higher complaint rate is, because complaints affect whether you reach the inbox at all.

The second is that categorisation rewards honesty about what a message is. A receipt that is only a receipt behaves like a receipt. A newsletter that is genuinely a newsletter behaves consistently over time, and consistency is what these systems reward.

What tends to land outside Promotions

Apple has not published the signals behind its categories, so anyone telling you they can guarantee placement is guessing. What we can say is what the two categories are described as covering, and what we consistently observe in our own testing on real devices.

Transactional and operational messages tend to sort as transactional when they are structurally what they claim to be:

  • One purpose per message. An order confirmation that confirms an order, with no cross-sell carousel and no promotional banner competing for the tap.
  • A sending identity that matches the purpose. Receipts from a transactional subdomain, marketing from a marketing subdomain.
  • Content that a filter can recognise: order numbers, amounts, dates, delivery detail, account actions.
  • Steady, predictable sending behaviour rather than a monthly spike.

Updates tends to collect account notices, policy changes, statements and service announcements. Promotions collects what it says on the label, and for most marketing programmes that is the correct home.

The practical consequence is about hygiene rather than gaming. If you have been stuffing promotional blocks into transactional email to borrow its open rate, that habit now has a cost. It blurs the signal on your most reliably read messages, and it risks pulling a genuinely useful receipt into a tab where nobody looks for receipts.

Your measurement just got harder again

Since Mail Privacy Protection arrived, the open rate for Apple Mail recipients has been a measure of Apple's pre-fetching rather than of human attention. Categories add a second distortion on top: a shift in when people read, which moves clicks later without telling you why.

If your weekly report shows a change in Apple Mail engagement, you now have at least three candidate explanations and no way to separate them from the ESP dashboard alone. That is not a reason to panic. It is a reason to finish the migration to metrics that are yours.

The approach we described in our deliverability review for the year ahead applies directly here. Report on clicks, sessions, orders and revenue per delivered message. Segment engagement by mailbox provider so that Apple, Gmail and Microsoft are never blended into a single misleading average. Use a seed set and real devices to observe placement rather than inferring it from aggregate numbers.

One specific habit is worth adopting now: check placement on a real iPhone, on your own domain, for each of your main message types, and record the result monthly. It takes ten minutes and it is the only direct evidence you will get.

The fundamentals did not change

Everything that made you reachable before still makes you reachable now. Authentication, a clean list, a defined inactivity policy, separated streams and a complaint rate you actually watch.

Authentication deserves the specific mention, because a consistent, authenticated sending identity is how any classifier learns who you are. If you have not finished the work of moving your DMARC policy to reject, that is a higher priority than anything in this post. A domain that anyone can spoof cannot build a stable reputation, and a sender without a stable reputation has no reliable placement anywhere, tabs or no tabs.

Beyond that, the honest answer to the screenshot question is that you cannot control the tab. You can control what the message actually is, which stream it comes from, and how consistently you behave. Those three things are what the filter is reading.

So separate your streams properly, stop borrowing the receipt's inbox position for promotional content, and change the report before somebody uses last year's open rate to make a decision this quarter.

Ready to transform your marketing?

Let's build something great.

By sending this form you agree to our privacy policy.