Where does the journey start?
A first-open event can be examined alongside available campaign context. Attribution depends on the agreed rules and the signals available to the implementation.
FIELD TO KEEP CONSISTENTcampaignGive acquisition data a place in the product journey. Map the relationship between a source, an app event and the outcome your team needs to understand.
Choose a sample event to see the fields behind it and the question it could help answer.
{
"event": "first_open",
"platform": "ios",
"campaign": "sample_launch",
"source": "example_network",
"schema_version": 1
}A first-open event can be examined alongside available campaign context. Attribution depends on the agreed rules and the signals available to the implementation.
FIELD TO KEEP CONSISTENTcampaignAn event contract describes what a payload means, when it should be produced and what a reader can safely assume.
| Field | Example | What to agree |
|---|---|---|
event | purchase | The action is completed, not merely attempted or displayed. |
event_id | sample_purchase_001 | A unique identifier for this occurrence, reused if the same event is retried. |
occurred_at | 2026-01-15T10:30:00Z | When the action happened, with a documented time-zone convention. |
value | 19.90 | The monetary basis: gross, net or another explicitly named definition. |
currency | USD | The currency associated with the value; do not combine unlike amounts silently. |
schema_version | 1 | The version of the contract that explains these fields. |
Retries are not new actions. If the same purchase is sent again, an agreed event identifier can help a receiving system recognise the duplicate.
A changed meaning is a changed contract. Document when a field changes and how old and new records should be compared.
Edit the sample, or load a broken one. This browser-only exercise checks a few explicit rules without sending the payload anywhere.
An MMP conversation works better when source definitions and product events are understood on both sides.
Compare the actions that follow an install, not just the install count. Establish the attribution assumptions before comparing campaigns.
Define the moment behind an event. A clear activation definition helps teams distinguish a completed action from a screen view.
Keep names, timestamps and observation windows aligned. Document differences before treating two numbers as a contradiction.
There is no universal event plan. Start with the app, the data sources and the decisions the team needs to make.
List the important actions, the properties that describe them and the owner of each definition.
Account for consent, platform restrictions and available source signals. Record what cannot be observed.
Review sample payloads, required fields and reporting windows. Agree on the checks that make the numbers usable.
A chart is the last part of a longer chain. Keep the important assumptions visible at each step.
Keep the action, timestamp and source context together. Separate missing fields from deliberately unavailable information so the distinction is not lost downstream.
event + time + contextHandle retries, define reporting dates and review incomplete records. A processing rule should be understandable without reconstructing a series of dashboard filters.
validate + reconcileName the cohort, event and observation window. Explain exclusions and data gaps alongside the number so another team can reproduce the reading.
definition + windowDecide whether reporting follows the occurrence time or receipt time, and how previously displayed periods are updated. A late event should not be interpreted as a new user action simply because processing happened later.
Yes. Agree on the authoritative source and how duplicate reports of one action are recognised. Combining two sources without reconciliation can make activity appear larger than it is.
No. Start with a specific purpose and collect only the fields needed for it, subject to applicable consent and platform requirements. The examples on this page do not require real customer identifiers.
No. The inspector and payload check are local demonstrations. SDK availability, collection endpoints, attribution rules, supported integrations and operational requirements must be confirmed separately.
A short overview of your app, source data and reporting gaps is enough to begin.