The Deliverability Rules Changed in February
In early 2026, Apple quietly rewrote the rules for anyone sending iMessages at any kind of volume. With iOS 26, the Messages app ships with a built-in spam filter — and it’s on by default. Unknown senders get auto-classified. Suspected spam gets routed to a separate section with no notification, no badge, no sound. If your recipient never thinks to open their “Unknown Senders” folder — and most people never will — your message is dead on arrival.
For normal users, this is a relief. For anyone using iMessage as an outreach channel, it’s a step change.
Before iOS 26, the rule was simple: send carefully enough that Apple doesn’t flag your account. Too much volume, too fast, too uniform — your Apple ID would hit a soft wall and stop delivering. Annoying, but recoverable.
After iOS 26, there are now two walls, and they work independently:
- The account wall. Apple’s backend anti-abuse system throttles or suspends the sender. This is the one everyone in the iMessage automation space has been talking about for years.
- The inbox wall. A new on-device classifier on every iOS 26 device decides whether this specific message, to this specific person, looks like spam — and routes it out of the main inbox even if your sending account is perfectly healthy.
A message that sails through Apple’s servers can now silently disappear the moment it lands on the recipient’s phone. That’s new, and it’s the thing most “how not to get blocked” posts haven’t caught up with yet.
This is the updated field manual. What triggers each wall, what the real thresholds look like for a solo operator (not a Mac farm), and how BlueDrip’s architecture changes which levers actually matter.
”Don’t Send Too Many” Is Not an Answer
Every article about iMessage scale tells you to “respect rate limits.” True, and useless. Without numbers, it’s vibes.
Here are real ones. Across operators running serious iMessage outreach through a single Apple ID — not a Mac farm, not a rotating pool, one human’s real primary ID — the safe ceiling is roughly 80 to 120 outbound messages per day. Some accounts handle more. Some get flagged at less. The variance tracks reputation: how long the Apple ID has existed, how much genuine two-way conversation it has on its books, and how human the sending patterns look to the classifier.
Above 150/day on a single ID, you’re running hot. Above 250/day, you’re gambling. Above 500/day, you’re not sending iMessages for very long.
The commercial cloud iMessage APIs — Sendblue, Linq, and the rest — handle big volume by spreading it across fleets of dedicated Apple IDs on Mac Minis in datacenters. If you need 1,000 messages a day, they spin it across 10+ IDs. Each ID gets a fresh, brand-new phone number your customer has never seen. None of them has reputation. None of them has history. Every send is a cold introduction from a number you’ve never texted before.
That’s the worst possible profile to have in an iOS 26 world. A brand-new number with zero conversation history, messaging strangers, is the textbook shape of the spam the new filter was trained to catch. The Mac farm model worked when Apple was only policing sender accounts. It is a much harder sell now that the inbox itself is making its own on-device judgment calls.
BlueDrip’s model is the opposite. You send from your primary Mac, from your existing Apple ID, from your real number — the one you’ve had for years, that has real friends and family in the thread history, that has sent and received thousands of casual messages, that has receipts and verification codes and birthday wishes mixed in with the business conversations. You are, by every signal Apple uses, a real human who occasionally sends more messages than usual. Your reputation isn’t a disadvantage to work around. It’s your moat.
The Three Triggers That Actually Matter
Trigger 1: Volume — and its shape
Respecting the daily ceiling is the easy part; you just count. The subtler mistake is volume shape. A real human doesn’t send 95 messages every single day like clockwork. A real human sends 8 on Monday, 40 on Tuesday because there was an event, 12 on Wednesday, and 0 on Sunday because it’s Sunday.
Perfectly flat daily volume is almost as spammy as high daily volume. Let the numbers vary. If you’re running a sequence that fires every weekday at 9 AM sharp and stops at 5 PM sharp with the same message count every day, you’re leaving a fingerprint that’s trivial for any pattern-based classifier to recognize.
Trigger 2: Report rate
This is the trigger that kills accounts fastest, and it operates on a different timescale than volume throttling. Every iMessage from an unknown sender shows a “Report Junk” link directly below the bubble. One tap and the sender goes into Apple’s pipeline.
Unlike volume — where you hit a soft wall and can back off — report rate can flip an Apple ID from fine to dead in hours. And Apple tells you nothing. No dashboard, no webhook, no email, no warning. The messages just stop going through.
A 1% report rate on a list of 100 is one person, and a single report is enough to start causing trouble. Three to five reports in a short window can take a weak account down. The math is that brutal, and there is no recovery path for reputation damage — only waiting.
Keeping reports near zero isn’t about clever sending. It’s about who you’re messaging and what you’re messaging them about. Which brings us to the third trigger.
Trigger 3: Pattern
Apple’s anti-abuse systems — both the backend one and the new on-device iOS 26 classifier — look for shapes that don’t match human behavior:
- Messages spaced at too-precise intervals (every 10.0 seconds is not how humans type).
- Identical text sent to multiple recipients back-to-back.
- Sudden bursts of outbound sends to numbers you’ve never communicated with.
- Keyword patterns common in scam templates: urgency language, crypto, “claim now,” link-followed-by-nothing-else.
- A sender with no recent inbound messages. An account that only talks and never listens is a screaming signal. Real iMessage users receive messages as well as send them.
Note that last one carefully. BlueDrip’s sending pattern blends into your existing thread history — that’s the whole point of the architecture — but if you install BlueDrip on a fresh Mac with a fresh Apple ID and start blasting sequences the next day, you’ll get caught faster than on your primary ID. Your primary ID is the one with history. Use it.
The Warm-Up Curve
If you’re starting from a well-aged Apple ID with real conversation history, your warm-up can be aggressive. Something like:
- Week 1: 25 messages/day
- Week 2: 50 messages/day
- Week 3: 75 messages/day
- Week 4+: 100 messages/day, holding there indefinitely
If you’re starting from a newer Apple ID — under a year old, with limited history — halve those numbers and add two weeks. Let the account build up real receive-side traffic before you push outbound volume. Have actual conversations from it. Join a few group threads. Text your mom. Make it look like what it actually should be: a person’s phone.
The fastest way to torch a new Apple ID is to install BlueDrip on day 1 and run a 100-contact sequence on day 2. Don’t.
Personalization Is a Defense, Not a Conversion Trick
Every marketing blog post will tell you to personalize your messages because it helps response rates. True, but that’s not the reason you should care in 2026.
Personalization is your primary defense against the iOS 26 on-device classifier. Identical messages sent to multiple recipients are the single clearest automated-spam signal that exists. Even small variations per message — the recipient’s first name, a line about where you met, a reference to something they mentioned — push each send into a different linguistic shape. Apple’s classifier compares messages across its corpus of reported spam. Generic template text that looks like ten thousand other generic template sends is exactly what it’s trained on.
Use per-recipient variables in every sequence template. Use custom fields for anything you know about the contact. If your message could have been written for any of the 500 people in your list without changing a word, rewrite it until it couldn’t.
BlueDrip supports this with LLM-driven templates: bring your own API key — OpenAI, Anthropic, or any OpenRouter model — and write the intent of the message instead of the message itself. The model generates the actual text per recipient at send time, so every outgoing message is a genuinely distinct sentence rather than a template with three variables swapped in.
List Hygiene: One Rule
Do not message people you don’t know exist.
That sounds obvious. In practice it means: no purchased lists, no scraped numbers, no “I found their number on a directory site” contacts with no prior relationship. Cold purchased lists will generate report-rate spikes no amount of clever pacing can absorb. They also poison your delivery data — messages to iMessage-capable contacts that unexpectedly fall through to SMS, or fail outright, are their own negative signal on your sending account.
BlueDrip’s sequences are built for people who should hear from you: existing leads who have already engaged, clients with a prior relationship, attendees who opted into an event, customers who bought something. Used that way, your report rate stays near zero by construction — not because you’re being clever, but because the people on the other end aren’t surprised to hear from you.
What to Do When You Feel the Wall
Signs you’re being throttled, in roughly the order they appear:
- Messages stuck on “Sending…” in the sequence log longer than usual.
- Contacts you know have iMessage starting to fall through to SMS (green bubbles where you used to see blue). SMS fallback is normal and expected for Android recipients — BlueDrip routes there automatically. The warning sign here is the change: previously-blue contacts suddenly dropping to green.
- Outright send failures with no obvious cause.
- Nothing sending at all.
If you hit any of these, stop sending immediately. Continuing to push extends the throttle. Then:
- Wait a full 24 hours before touching the account again. Not four hours. Not overnight. Twenty-four.
- When you resume, halve your daily ceiling for at least a week.
- Audit the last 48 hours of sent content for anything that looks more aggressive than usual — a new template, a new list segment, a new type of offer. Whatever changed most recently is usually the culprit.
- Revisit the list. If there’s any chance you messaged people who didn’t know who you were, cut them out of rotation entirely.
The biggest mistake operators make after getting throttled is impatience. You do not save the account by pushing through. You save it by backing off further than you think is necessary.
The Long Game
iMessage is the best messaging channel on the planet right now. It has 98% open rates, three-minute average read times, and a built-in “this is a real person” trust signal that SMS simply cannot replicate. That’s worth protecting.
But it’s a borrowed channel. Apple tolerates legitimate business use because the volume is modest and the experience isn’t polluted. When that stops being true, they lock it down harder — and iOS 26’s on-by-default spam filter is the early edge of exactly that. Every operator sending through iMessage loses when the channel gets locked down. That’s not a hypothetical; it’s the entire history of every marketing channel since email.
The operators who are still sending iMessages in 2028 will be the ones who treat every send like it’s going to their own phone. They won’t be trying to squeeze maximum volume out of a single Apple ID. They’ll be running a sending practice that could operate at 80 messages a day forever, on a real Mac, from a real number, to real people who should be hearing from them.
That’s the version of iMessage outreach that survives any new filter Apple ships next.
It’s also, for what it’s worth, the version BlueDrip was built for.