Typical transactional alerts
A transactional alert tells a customer that something just happened to their order, money or booking. Common examples are order confirmation, dispatch and out-for-delivery messages, payment received or failed notices, booking and ticket confirmations, balance and debit alerts, and account-change confirmations. The customer expects them, often within seconds, and relies on them as a record.
Triggering alerts from your system
The best alerts are event-driven: your order, payment or booking system emits an event and a small worker calls the SMS API with the template and its variables. Include your own reference in every request and use it as an idempotency key, so a retry after a timeout never sends the same alert twice. Store the message ID that comes back and attach the delivery report when it arrives, so support can answer “did the customer get it?” without guesswork. Keep variables short so the message stays within one segment.
Keeping alerts genuinely transactional
An alert loses its category the moment it starts selling. A shipping notice that ends with a discount code is promotional in the eyes of the operator, and it risks rejection or filtering. Register one template for each alert type, keep offers out of them, and send marketing as a separate promotional message to customers who have opted in. Our guide to promotional and transactional SMS shows how to classify the borderline cases.
Handling failures and retries
Not every message is delivered the first time. Read the delivery report, retry only the failures that are worth retrying, and stop after a sensible limit. A payment alert that arrives hours late can confuse more than it helps, so give time-sensitive alerts an expiry. For critical messages, fall back to another channel, such as a WhatsApp message or a voice call, and record which channel finally reached the customer.
Coping with volume spikes
Alert traffic rises sharply on sale days, salary dates and month-end. Keep alerts on their own lane so a large promotional send never queues ahead of them, queue and throttle on your side if a burst exceeds your agreed throughput, and watch queue depth and delivery times during peaks. The delivery reports guide explains how to read what the network tells you back.
