Purchase and revenue goals
Count a store purchase as the conversion of a push, in-app message or journey, and compare revenue per person against the control group.
A purchase goal matches any purchase or only products you choose. Every experiment report also shows each group's revenue, whatever the goal. For variants, control groups and confidence, see Experiments.
Goals and revenue only see purchases Grovs records. Turn on revenue tracking and log purchases from your app first.
Set a purchase goal
In a push campaign or in-app message, open the Experiment step of the composer. In a journey, open its goals. For other goal kinds, see Experiment goals.
Under Count as a conversion, choose A purchase.
Under Which products, keep Any product, or choose Only these products and add up to 20 product IDs.
Under Conversion window, pick how long after seeing the message a purchase still counts: within 1 day, within 3 days, within 7 days or within 30 days. The 30-day window is offered only for purchase goals.
What counts as a purchase
- Every
buycounts, including subscription renewals. - With Only these products, only the listed product IDs count.
- A person converts if they buy within the window after first seeing the message.
- An app event named
purchaseis not a store purchase. It never reaches a purchase goal, but it matches a custom event goal namedpurchase.
Purchases in journeys
A purchase goal takes the person out of the journey as converted, like other journey goals.
- Only a purchase made after the person entered the journey counts.
- Journeys have no conversion window. A purchase counts any time while the person is in the journey.
- A purchase attributed to the person later still takes them out, if it was made after they entered.
Revenue on the report
Every experiment report shows revenue, with or without a purchase goal.
| Number | Meaning |
|---|---|
| Revenue | Revenue from the group's people within the window after they first saw the message |
| Per person | That revenue divided by the number of people in the group |
With a control group, the card also compares them, for example: "Variants brought about $1,240 more than the control (+18% per person)."
- Revenue is in US dollars and net of refunds. A reversed refund restores it.
- The window is the goal's conversion window. Without a goal it is 72 hours.
- The comparison pools all variants against the control, using exact per-person values. Only displayed numbers are rounded.
- The percentage is omitted when the control's revenue per person is zero.
- The comparison is omitted when there is no control group or either side has no exposed people.
- Revenue has no confidence figure. Confidence applies to conversions only.
- A journey with a control group and no goal still gets a report with revenue.
- When two visitors merge, purchases made before the merge stay with the old visitor. They can drop out of the revenue figures.
The goal format
A purchase goal is stored as:
{ "purchase": true, "products": ["premium_month", "premium_year"], "window_hours": 720 }| Field | Rules |
|---|---|
purchase | Must be true |
products | Optional. 1 to 20 product IDs of 1 to 100 characters. Omit for any product. Allowed only on a purchase goal |
window_hours | 24, 72, 168 or 720. 720 is allowed only on a purchase goal |
A goal has exactly one of event, screen, system or purchase.
Report groups carry revenue_cents and revenue_per_person_cents. The report carries revenue_lift and added_revenue_cents, which are null when there is nothing to compare.