Channels and delivery
How messages reach customers through channels and providers, and every check that can hold a message back.
A channel is a way of reaching a customer. A provider is the service that actually carries the message on that channel, such as your email service or SMS gateway. A template holds what to say on each channel. A campaign or journey decides who gets it and when. This page explains how those fit together, and what happens between "send" and "delivered".
Channels
| Kind | Channels |
|---|---|
| Message channels | Email, SMS, RCS, WhatsApp, Viber, Push, Webhook |
| On-site channels | Web banner, Web personalization, Mobile in-app |
Message channels are sent to a person. On-site channels show content when the person is already visiting your website or app, so the timing rules for messages, such as silent hours, do not apply to them. A campaign can also choose Auto (best channel), which picks the channel per profile at send time among the channels that have a template and a provider, favouring the channel the customer is most likely to respond to.
Your organisation can also define its own channels. They appear next to the built-in ones.
Providers
Providers are set up under Settings → Integrations → Channels. For each channel you can add several providers, mark one as the Default, and select Test connection. A campaign or journey that does not name a provider uses the channel's default. A campaign can pin one provider instead.
Credentials are stored encrypted, can be replaced but never read back, and the channel and provider type cannot be changed after creation. Create a new provider instead. Deleting a provider that campaigns are pinned to makes them fall back to the channel default; if it was the default, the oldest remaining provider takes over.
Some providers report delivery receipts: confirmations that a message reached the phone or inbox. Where a provider reports none, a send is recorded as sent once the provider accepts it.
Finding the contact address
Optima finds where to send from the profile's identifiers:
| Channel | Identifier used |
|---|---|
email | |
| SMS | phone, phone_number or msisdn (the first present) |
| Push | The device token your app registered, as push_token_fcm or push_token_apns |
| Webhook | The address supplied with the send |
A profile with no suitable identifier cannot be contacted on that channel. There is no separate device list: push tokens are ordinary identifiers.
From decision to delivery
When a campaign or journey decides to contact a profile, the message goes through a final set of checks immediately before it is transmitted. The checks below are the main ones. A message that fails a check is not sent, and the reason is recorded.
- Consent for the campaign's purpose is checked again, live. Consent given earlier is never assumed.
- Suppression. Customers who unsubscribed are skipped.
- Contact policies, described below.
- Contact and device. The profile must have a valid address or registered device for the channel.
- Rendering. The template is filled in for this customer. A message that cannot be rendered is not sent.
- Control group. Customers deliberately held out to measure impact are not contacted.
- Silent hours may delay the message.
- Transmission through the provider.
The campaign's Test tab can run these checks against your audience in advance, under the heading Who will get this, without sending anything.
What gets recorded
Every outcome is recorded against the profile as an event: queued, sent, delivered, opened, clicked, bounced or failed, with a reason when a message was dropped. These events are the single source for campaign statistics, frequency caps, fatigue limits and attribution. Open and click tracking follow the template's tracking settings.
Webhook sends
A webhook send posts the template's body to your address. When you set an HMAC secret on the provider, each request carries an X-Optima-Signature header with the value sha256=<hex> (HMAC-SHA256) that your receiver can verify. Only requests to the fixed address saved in the provider are signed; an address built per send or from a template is sent unsigned. Optima refuses to call addresses inside private networks unless your administrator has allowed them.
Contact policies
Contact policies protect customers from too much contact. They are screens under Contact policies in the Engage group.
| Policy | What it does |
|---|---|
| Frequency cap (per campaign) | Limits how often one campaign contacts one profile within a window of days: a total, a per-channel maximum, or both. |
| Fatigue policies | A global guardrail across all campaigns: no more than a set number of touches per profile in a rolling window. |
| Silent hours | Times when campaign messages must not be delivered. |
| Response policies | Stop offering anything to a customer who has answered enough offers overall. |
| Control groups | A slice of the audience deliberately not contacted, so you can measure what a campaign caused. |
Fatigue policies
A fatigue policy has a Name, a Scope (Single profile, or Profile hierarchy to cap a whole linked tree such as a household together), an optional Channel (all channels if empty), Max touches, Window (days) and optional Campaign types. Where the channel is not yet known, as in a next-best-action decision or an Auto send, channel-scoped policies cannot apply yet, and all-channel policies do. Channel-scoped policies take effect at the final delivery check once the channel is known.
Every active policy is enforced at three points:
- Decisions. A profile that has been touched too often is excluded from offers with the reason Fatigue.
- Journey send steps. The step's On fatigue setting decides: Defer parks the send and retries hourly until the window clears, Skip drops this send and continues the journey, Exit leaves the journey. The default is Defer.
- Delivery. The last check before a message leaves. A violation drops the send and records it as failed with the reason fatigue.
A fatigue policy sits beside the campaign's own frequency cap. The cap protects within one campaign; the policy protects across all of them.
Silent hours
Silent hours are set once for the environment and can be overridden per channel, per campaign or per journey step. A more specific setting replaces the less specific one wholesale. The windows are read on a Timezone you choose, or on each recipient's own clock if you point to a profile attribute holding it. A message that reaches the front of the queue during a silent window waits until the next window opens, and consent and campaign state are checked again each time. Max wait (hours) bounds how long it may wait. On-site channels and system emails such as password resets are not affected.
Control groups and response policies
A control group keeps a stable slice of each entity type from being contacted. The same profile always lands in the same group, and raising the percentage only adds people. It is applied at the last moment before sending, after the consent check, so the control group is exactly the people who would otherwise have been contacted. That is what makes lift measurable on the campaign's Measurement tab.
A response policy acts on decisions: when a customer has accepted, rejected or postponed enough offers within a period, every candidate offer is excluded and the decision comes back empty. It does not recall messages already queued.