Skip to main content

Campaign Setup (Dashboard)

The SDK only renders what Engage returns. A fetch returns a message only when a campaign matches the requesting user and is authored for API-pull delivery. This page lists exactly what a pull campaign must look like. (The dashboard steps are the same for Android and iOS.)

Requirements checklist​

For a fetch to return a campaign, all of the following must hold:

  • ✅ Delivery mode = API pull — the campaign's notificationJson.deliveryMode is "api". This is the single most common miss: in-app campaigns default to FCM delivery, and an FCM campaign is invisible to the pull endpoint.
  • ✅ Status = LIVE — not draft, paused, or scheduled-for-later.
  • ✅ Type = In-App (notificationType = InApp).
  • ✅ Audience = Everyone or Audience — not action/trigger-based. Action-based campaigns are not pullable.
  • ✅ Same project — the campaign lives under the project/source that your SDK sdkKey resolves to (see Environment matching below).
  • ✅ Time window open — displayTime is in the past (or unset) and expireDate is in the future (or unset).
  • ✅ Screen target matches — if condition.target is set, the SDK's fetch must send the same target (see Screen targeting). Leave it empty for a campaign that may appear on any screen.
API-pull delivery is not "manual dispatch"

A pull campaign is delivered by the SDK polling — there is no "Send now" step. If the dashboard reports something like reason: API_DELIVERY / "manual dispatch is not used" when you try to dispatch it, that is expected and correct: the campaign is already live for the SDK to pull. Do not try to dispatch it manually.

Audience: Everyone vs Audience​

TypeWho receives itNotes
EveryoneEvery user that pollsNo cohort lookup. Best for your first test — it removes audience matching from the equation.
AudienceUsers in the targeted cohort(s)The user must be resolvable as a member. An anonymous user (not yet identify()-ed), or a user whose profile lives in a different environment than the in-app source, will not match.

For a first end-to-end test, start with Everyone. Once that renders on the device, switch to Audience to validate targeting.

Timing fields (condition)​

FieldUnitMeaning
displayTimeepoch millisecondsEarliest time the message may show. Unset = show immediately. The SDK also applies a 60s tolerance.
expireDateepoch millisecondsAfter this, the campaign is no longer returned / shown. Unset = never expires.
delayminutesExtra delay before rendering after it is fetched. 0 = render as soon as possible.
targetstringScreen name for screen targeting. Empty = any screen.
Units

displayTime / expireDate are epoch milliseconds, and delay is in minutes (a delay of 5 defers rendering by five minutes). For a quick test: no displayTime, expireDate far in the future, delay = 0.

Environment matching​

The sdkKey your app sends must resolve to the same project/source that owns the campaign, in the same Engage environment the app fetches from.

  1. Key ↔ project. The public sdkKey (your SDK apiKey, or engageInAppConfig.sdkKey) identifies a project. Create the campaign under that project — not a different one that happens to be open in the dashboard.
  2. App ↔ environment. The campaign lives in whatever Engage environment your fetchEndpoint points at (e.g. api-eu). Create the campaign in the dashboard for that same environment, not a different region/instance.

A mismatch here is the usual reason a correctly-authored, LIVE campaign still returns an empty messages array — see Troubleshooting.

Rendering content​

Author real in-app content (title, body, image, buttons, layout) for the campaign. A campaign that matches but has empty layout content will fetch successfully yet render an empty/blank message. Supported layouts include native banner, modal, fullscreen, and carousel.