Skip to main content
Adrià García
Menu
← Back to the simulators

Customer Journey Analytics · Data views

Same data, different reports

Laura arrives through paid search, returns through an email link and buys later. Sales sees one purchase. Marketing compares conversion and channels. Change how the same events are read and see why their reports can differ.

Scenario

Try this change

Extend the session to 60 minutes. The rate changes without a new purchase. Then try changing which channel gets the revenue.

Inactivity that starts another session

Laura has a 45-minute gap between events. The data stays the same; the session boundary changes.

Channel retained by the dimension

Persistence in this example expires at the end of the session. An empty field does not replace the retained value.

How to assign revenue to channels

Explicit models consider all this person's touchpoints in the displayed reporting window, not just the purchase session.

See the full result ↓
Illustrative scenario ↑ Back to the controls
Example assumptions

Example: one person, six events and a €120 purchase. No stitching, filters or extra session rules. Session persistence; No Value does not overwrite. Person attribution over 09:00–10:10, one dimension. Not a full CJA replica.

The reading changes; the purchase does not

Reference versus your settings

A stays fixed. B uses your controls. The rate is purchasing sessions divided by sessions, not people who purchase.

Evidence behind the calculation

The same events from Laura

The channel is collected only on the two campaign touchpoints. An empty value does not mean direct traffic. Times are from the same morning.

The recommendation

Agree what the report measures before comparing numbers

A data view defines a reading of the data. Two configurations can answer different questions without Laura's behaviour changing.

  1. 01

    Set the session definition

    The gap between events determines whether the purchase falls in a new session. Document the timeout and any other session-start rules before comparing rates.

  2. 02

    Separate missing values from persistence

    Without persistence, the channel applies only to its own event. With persistence, an earlier value can accompany the purchase. The original event is not rewritten.

  3. 03

    Name the model and window

    In this single-dimension report, explicit attribution overrides its allocation. We use person scope and the displayed reporting window. Do not extend this calculation to multi-dimension reports.

  4. 04

    Do not confuse credit with impact

    Splitting revenue across channels does not establish which one caused the purchase. Attribution describes a model; estimating incremental impact needs a different comparison.

Try another decision Who is one person to CJA? → Who is "one person" to CJA, depending on the field you choose as person ID?

Explore this decision

Adobe documentation

CJA