Let SMPP bind traffic open Inbox conversations
By default, traffic that arrives on your SMPP bind (submit_sm from your ESME client) is recorded purely as machine traffic: every submit gets a per-channel Messages row, but it does not open an Inbox conversation and does not fire the inbox_new_conversation notification.
If your team genuinely answers bind-received free-text in the Inbox — a two-way aggregator workflow where agents reply to inbound SMS — turn the SMPP Inbox toggle on and pass-through bind submits open a conversation per send again, exactly as they did before the default flipped.
Everything on this page is a tenant-owned setting. The toggle flips one flag, inbox_conversations, on your organization’s smpp settings; it sends nothing and routes nothing.
1. What the toggle does
The flag gates the conversation creation step of bind ingest, and nothing else:- ON — a pass-through
submit_smreceived on your bind upserts an Inbox conversation per send and mints theinbox_new_conversationnotification on a fresh thread. Agents see and can answer the message in the Inbox. - OFF (the default) — the same submit records a full Messages row with its normal status trail, but no conversation is created and no new-conversation bell or email fires. Bind ingest stays machine-side.
submit_sm itself is gated.
2. Enable or disable it
In the dashboard
Go to Settings → Channels → SMPP Inbox. Read the current state and flip the toggle; the change takes effect on the next bind submit.Over the API
Read the current state withGET /messages/smpp-bind-inbox:
PUT /messages/smpp-bind-inbox — send true to let bind traffic open conversations, false to keep bind ingest machine-side:
inbox_conversations — anything other than a literal boolean is rejected.
3. When to turn it on — and when to leave it off
Turn it ON when bind-received free-text is genuinely answered by humans in the Inbox. Two-way aggregator workflows are the canonical case: recipients text the aggregator’s short code, and your agents reply from the Inbox. Leave it OFF when your bind carries OTP, notification, or other one-way traffic. A bind that mostly submits one-time codes would otherwise open one conversation per code and ring one bell per destination — queue noise, not engagement. If you also terminate OTP submits from an absorbed bind onto a delivery channel, keep this toggle on false and pin the no-inbox behaviour on the termination rule instead — see transactional no-inbox mode. Most tenants leave it off. The setting exists because the previous default (on) flooded the queue with synthetic conversations for anyone running high-volume bind traffic.4. How it interacts with the other inbox gates
Three separate mechanisms decide whether a send touches the Inbox. They are independent and compose:- This flag — the tenant-wide default for messages received on your SMPP bind. It says, for all bind submits at once, whether the conversation step runs.
inboxstamps on termination-rule hops — per-hop, per-send overrides on the outbound side. Adeliverhop declaringinbox: 'none'orinbox: 'thread'pins the behaviour for traffic the rule matches, whatever the bind default is. See transactional no-inbox mode.- Template-category inference — a WhatsApp template of category AUTHENTICATION (Meta’s OTP category) is treated as transactional and skips the conversation upsert even with no rule saying so.
inbox: 'none', the hop wins — flip the hop, not this toggle.
5. Troubleshooting
“Replies on my bind do not appear in the Inbox.” Check the toggle first — with it on the defaultfalse, pass-through bind submits record Messages rows only, so there is nothing for an agent to open. Flip Settings → Channels → SMPP Inbox on, or PUT inbox_conversations: true. Note the toggle only affects new submits; it does not retro-open conversations for traffic recorded while it was off.
“A conversation opens but no new-conversation notification fires.” Conversation creation and the notification are two steps; once the conversation exists, the inbox_new_conversation notification still depends on your notification routing — check that your role/team rules route Inbox events to the agents you expect (Settings → Team → notifications).
“I turned it on but OTP traffic now floods the queue.” That is the toggle doing exactly what it says. Flip it back off and pin no-inbox on the termination rule for your OTP binds instead (see the transactional no-inbox guide); the per-hop route keeps machine traffic out while leaving two-way paths free to thread.
See also
- Connect via SMPP — bind credentials, ports, and the wire protocol
- Transactional no-inbox mode — the per-hop alternative when only some bind traffic should skip the Inbox
- SMPP edge model — how the relay underneath your bind is organized