Explore SMS message filters

Highlight

iOS 16 adds 12 new sub-categorization capabilities to the SMS message filtering extension. Developers can automatically classify text messages from unknown senders into sub-folders of Transactional and Promotional, instead of simply marking them as spam.

Core Content

SMS inbox clutter

（01:08）

In many countries, SMS has become the main channel for companies to notify customers - bank transaction reminders, marketing activities, express delivery notifications, and appointment reminders. The messages, which come from short codes, alphanumeric codes and regular phone numbers, are mixed into inboxes, making it difficult for users to find personal messages from family and friends.

iOS offers a “Filter Unknown Senders” feature, but even filtered folders can quickly pile up if multiple messages are received every day.

How the Message Filter Extension works

（02:04）

iOS provides an extended sandbox-based model for classifying messages from unknown senders. After users install the filtering app from the App Store, they turn on filtering in Settings > Messages > Unknown & Spam and select a filter. Note that only one filter can be active at the same time.

Starting with iOS 14, the Messages app displays three folders: Transactions, Promotions, and Junk. This classification structure remains consistent regardless of which filter is used.

Subcategory enhancements for iOS 16

（03:37）

Prior to iOS 16, developers could only sort messages into three top-level categories. iOS 16 adds 12 new subcategories to make classification more refined:

Transactional Subcategory:

Finance
Orders
Health
Public Services
Weather
Reminders -Rewards

Promotional subcategory:

Coupons (coupons)
Offers

Taking the Indian market as an example, users often receive a large number of financial transaction text messages. These messages can now be organized into the Finance subfolder under Transactions.

Two-stage filtering process

（04:44）

The filtering process is divided into two stages:

Configuration Phase: When the user selects a filter in settings, the new iOS 16 API requests the capabilities supported by the filter. The filter returns a list of supported categories and subcategories, and updates the Messages app’s inbox folder after iOS verification.

Runtime Classification Phase: Every time a text message is received from an unknown sender, iOS queries the filters to determine which category and subcategory the message belongs to. The filter must return capabilities declared during the configuration phase.

Detailed Content

Create Message Filter Extension

（06:32）

In Xcode, add the Message Filter Extension target to your project. After creating,MessageFilterExtension.swiftThe required function templates are automatically generated.

Declare supported subcategories

（07:02）

accomplishhandle(_capabilitiesRequest:context:completion:)Method declaration capabilities:

func handle(
    _ capabilitiesRequest: ILMessageFilterCapabilitiesQueryRequest,
    context: ILMessageFilterExtensionContext,
    completion: @escaping (ILMessageFilterCapabilitiesQueryResponse) -> Void
) {
    let response = ILMessageFilterCapabilitiesQueryResponse()
    // Select up to 5 subcategories
    response.transactionalSubActions = [
        .transactionalFinance,
        .transactionalOrders,
        .transactionalHealth
    ]
    response.promotionalSubActions = [
        .promotionalCoupons,
        .promotionalOffers
    ]
    completion(response)
}

Key points:

Declare up to 5 subcategories -transactionalSubActionsSet transaction category subcategory -promotionalSubActionsSet promotion subcategories
After declaration, the message application will display the corresponding subfolder

Message classification logic

（08:16）

accomplishhandle(_queryRequest:context:completion:)Method handles incoming messages:

func handle(
    _ queryRequest: ILMessageFilterQueryRequest,
    context: ILMessageFilterExtensionContext,
    completion: @escaping (ILMessageFilterQueryResponse) -> Void
) {
    guard let message = queryRequest.messageBody else { return }
    let response = ILMessageFilterQueryResponse()

    switch message {
    case _ where message.contains("debited"):
        response.filterAction = .transaction
        response.filterSubAction = .transactionalFinance
    case _ where message.contains("coupon"):
        response.filterAction = .promotion
        response.filterSubAction = .promotionalCoupons
    default:
        response.filterAction = .allow
    }

    completion(response)
}

Key points:

queryRequest.messageBodyGet message text -filterActionSet top-level categories (.transaction、.promotion、.junk、.allow）
filterSubActionSet subcategories
If the returned action and sub-action combinations do not match, iOS will ignore the sub-action and only use the action
Messages that cannot be classified are returned.allow, and let it go into your regular inbox

Classification result display

Once configured, the Messages app will display subfolders. For example:

Transactions > Finance: Bank SMS containing the “debited” keyword
Transactions > Orders: SMS containing order information
Promotions > Coupons: Marketing text messages containing the “coupon” keyword

Updates to Apple India SMS Filter

（11:03）

Apple’s SMS filters in India also make use of iOS 16’s new capabilities, with new Finance, Orders, and Reminders subfolders. Bank transactions go into Finance, takeout/delivery messages go into Orders, and important events and to-do items go into Reminders.

Core Takeaways

What: Develop localized SMS classification tools for specific markets (e.g. India, Southeast Asia)
Why it’s worth doing: These markets use a lot of text messages. Banks, e-commerce, and public services all rely on text message notifications, and there is a strong demand for classification.
How to start: Create Message Filter Extension and configure classification rules based on common local SMS template keywords
What to do: Build an intelligent financial assistant to automatically organize bank transaction text messages
Why it’s worth doing: The Finance subcategory gives bank text messages a dedicated place to further extract transaction data.
How to start: Identify the bank SMS format in the Extension, extract the amount, time, and merchant information, and store it in the App’s database to generate a consumption report
What to do: Develop promotional information aggregation tool
Why it’s worth it: Coupons and Offers subcategories let users never miss offers while keeping their inbox tidy
How to start: Classify promotional text messages into corresponding folders, and provide unified discount code management and expiration reminders in the App
What to do: Create an enterprise-grade SMS management solution
Why is it worth doing: Enterprise internal systems (OA, ERP, monitoring alarms) often send notifications through SMS, which need to be distinguished from marketing SMS.
How to start: Customize filters for the enterprise and classify internal system text messages into Reminders or Public Services to ensure that important notifications are not drowned

What’s new in Foundation — New features of the Foundation framework
What’s new in Swift — New features of Swift language
Create your Privacy Nutrition Label — Guide to creating a privacy nutrition label