Skip to main content

Conversion attribution

Conversion attribution is the process by which the Frosmo Platform:

  1. Determines the web content responsible for a conversion (either a transaction or a non-purchase conversion).

  2. Registers the conversion for the content.

note

Conversion attribution is a prerequisite for generating conversion-based performance statistics for content.

Conversion attribution requires that your site uses conversion tracking or transaction tracking.

The Frosmo Platform attributes conversions to the following content:

Conversion attribution for modifications

Modification-based conversion attribution is the process by which the Frosmo Platform determines the modification responsible for a conversion (either a transaction or a non-purchase conversion) and registers the conversion for that modification. The more conversions a modification gets attributed, the more successful the modification. You can view modification conversion statistics in the Frosmo Control Panel.

The platform uses basic modification events as the basis for determining which modification was responsible for a conversion.

Modification-based conversion attribution works as follows:

  1. When a visitor completes a conversion, the conversion is attributed to the last modification the visitor clicked.

  2. If the visitor did not click any modification, the conversion is attributed to the last modification that got a true display in the visitor's browser.

  3. If no modification got a true display, the conversion is attributed to the last modification that got a display in the visitor's browser.

A conversion is always attributed to a single modification. The same conversion cannot be attributed to multiple modifications.

By default, the platform automatically attributes conversions to modifications. You can enable and disable automatic conversion attribution for a modification from the modification's advanced settings.

Conversion attribution time limit for modifications

If conversion or transaction tracking on a site is session-based, the platform attributes conversions only to modifications that are displayed or clicked during the visitor's current browser session. Session-based tracking is the default.

If conversion or transaction tracking on a site is not session-based, the platform attributes a conversion to a modification within up to 1 day after the modification was displayed or clicked. While 1 day is the default, the attribution period can be configured to any value for a site.

To check whether the tracking on your site is session-based, contact Frosmo support.

Examples: Attributing conversions to modifications

The following examples illustrate the logic behind attributing conversions to modifications.

Modification example 1

If a visitor views two modifications, resulting in a true display for both, and then completes a conversion, the conversion is attributed to the modification last viewed by the visitor. Conversion attribution prioritizes the most recent modification event over earlier events of the same type (click, display, or true display).

Conversion attributed to the modification last viewed by the visitor
Figure: Conversion attributed to the modification last viewed by the visitor

Modification example 2

If a visitor first views modification 1 and clicks it, then views modification 2, which gets a true display, and finally completes a conversion, the conversion is attributed to modification 1. Conversion attribution prioritizes a modification that gets a click over one that gets a true display.

Conversion attributed to the modification last clicked by the visitor
Figure: Conversion attributed to the modification last clicked by the visitor

Modification example 3

If a visitor first views modification 1, which gets a true display, and then views modification 2, which only gets a display, the conversion is attributed to modification 1. Conversion attribution prioritizes a modification that gets a true display over one that gets a display.

Conversion attributed to a true-displayed modification over a displayed modification
Figure: Conversion attributed to a true-displayed modification over a displayed modification

Conversion attribution for recommendations

Recommendation-based conversion attribution is the process by which the Frosmo Platform registers a transaction for a recommended item that a visitor clicked in a recommendation and then purchased. The platform also registers the purchase as a transaction for the recommendation as a whole. If the visitor clicked the same item in multiple recommendations, the platform registers the transaction for the recommendation the visitor clicked last. (The platform nonetheless registers the click for each of the recommendations where the visitor clicked the item.)

The platform uses recommendation-based conversion attribution solely for generating recommendation statistics.

note

Recommendation-based conversion attribution only works with recommendation modifications that use a standard recommendation template from a basic feature setup.

The Frosmo Platform does not attribute non-transaction conversions to recommended items.

Transaction attribution time limit for recommendations

By default, the platform attributes a transaction to a recommended item and, by extension, to the corresponding recommendation within up to 30 days after the item was clicked.

For example, if a visitor clicks an item in a recommendation on January 1 and then buys the item on January 31, the platform registers the purchase as a transaction for the item and recommendation, and the transaction shows up in recommendation statistics. If the visitor instead buys the item on February 1, the platform does not register the purchase as a transaction for the item or recommendation. (The platform still registers a transaction for the item itself, but the transaction only shows up in product statistics.)

note

The Frosmo Control Panel always displays recommendation statistics for a specific time range. Since the transaction attribution period is 30 days, the click responsible for a transaction may fall outside the selected time range, meaning the statistics for the time range will include the transaction but not the contributing click. For an example, see Recommendation example 1.

Examples: Attributing transactions to recommendations

The following examples illustrate the logic behind attributing transactions to recommendations.

Recommendation example 1

A visitor clicks an item in a recommendation on May 31 and then buys the item on June 10. Since the purchase happened within 30 days of the click, the platform registers a transaction for the recommended item and for the recommendation. (The platform also separately registers a transaction for the item in product statistics.)

If you view the recommendation statistics for the last 14 days on June 13, both the click and the transaction show up in the statistics, as both are inside the 14-day range. However, if you view the 14-day statistics again on June 19, only the transaction shows up in the statistics, since May 31 is now outside the 14-day range. (In addition to the number of clicks during the selected time range, the summary recommendation statistics also show the number of clicks from the 30 days preceding the selected range, so the May 31 click is included in this latter number.)

Transaction attributed to the clicked recommendation
Figure: Transaction attributed to the clicked recommendation

Recommendation example 2

A visitor clicks an item in a recommendation on May 31 and then buys the item on July 4. Since the purchase did not happen within 30 days of the click, the platform does not register a transaction for the recommended item or for the recommendation. (The platform nonetheless registers a transaction for the item in product statistics.)

Transaction no longer attributable to the clicked recommendation
Figure: Transaction no longer attributable to the clicked recommendation

Recommendation example 3

A visitor clicks an item in recommendation A on June 1, they click the same item in recommendation B on June 8, and they then buy the item on June 18. The platform registers a transaction for the recommended item in recommendation B. The statistics for recommendation B will show both the click and the transaction, while the statistics for recommendation A will only show the click.

Transaction attributed to the recommendation last clicked by the visitor
Figure: Transaction attributed to the recommendation last clicked by the visitor