Idempotency & safe retries.

What the API gives you

Every accepted send returns an identifier. POST /sms/sendSms returns the message ID as the response body, POST /sms/sendbulk returns a bulkSessionId, and POST /whatsapp/message/sendbulk returns a batchId. Once you hold that identifier, the message was accepted: never send it again.

Store the identifier against your own operation ID (an order number, a login attempt, a campaign) so that a retry from anywhere in your stack can see that the work is already done.

// 1. Record the attempt first, keyed by your own operation ID.
await db.sends.insert({ opId: "order-8842-confirm", status: "pending" });

// 2. Send.
const res = await fetch("https://api.broadnet.me/sms/sendSms", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.BROADNET_TOKEN}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    senderId: "BroadNet",
    destinationNumber: "447700900123",
    message: "Your order 8842 is confirmed",
    messageType: "TEXT",
  }),
});

// 3. Keep the message ID. With it, the send is done.
if (res.ok) {
  const messageId = await res.text();
  await db.sends.update({ opId: "order-8842-confirm" }, { status: "accepted", messageId });
}

When a request times out

The risky case is a request that reached BroadNet but whose response never reached you. Before you send again, look for evidence that the first attempt was accepted: a delivery status for that number and text on your webhook, or, for bulk jobs, a new entry in GET /sms/list.

What each outcome means

Decide what to do from the outcome of the call, not from a fixed retry count.

Best practices

• Write the attempt to your database before you call the API, and the returned ID after.
• Treat a stored message ID or batch ID as proof of acceptance.
• Retry 5xx responses with capped exponential backoff; see Retries & backoff.
• After a timeout, check delivery status before sending the same message again.

Related documentation