New Advanced Reports FAQ

In the previous data model, transactions were attributed to a Coveo service (search, listings, or recommendations) based on client-side events, using the value of the event’s originLevel1 property.

Transactions are now attributed to a Coveo service (search, listings, or recommendations) based on server-side events. These events are automatically sent when the Search API is queried. This ensures that all transactions are attributed to the correct service, even if the client-side events aren’t sent, or don’t contain the correct information.

You should notice an increase in transactions counts and in revenue attributed to the different Coveo services.

You’ll no longer see fractional transactions in your reports.

For example, numbers like 46.78 transactions attributed to the search service will no longer appear in your reports.

In the previous data model, foreign exchange rates were statically configured to align with the exchange values of June 1, 2021.
Additionally, certain exchange rates were unavailable and were artificially assigned a value of 1.

In Coveo’s Core model, foreign exchange rates are computed daily and cover all currencies.

Transactions that were made in a currency other than USD will now have more accurate conversion rates.
Revenue metrics will show different values, meaning that they can increase or decrease depending on how the rates have changed since June 1, 2021.

In the previous data model, when a customer clicked many products following the same query (for example, by using a product’s Quick view or opening a result in a new tab), only the click action on the product that appeared highest in the results was considered.

For example, if a customer searched for boots, and then, in a new tab, opened the products listed in positions 1 and 9, the average click rank would be 1.

In Coveo’s Core model, when a customer clicks multiple products (from the same query), the average click rank is computed based on the position of each product that was clicked.

For example, if a customer searched for boots, and then, in a new tab, opened the products listed in positions 1 and 9, the average click rank would be 5.

You should notice an increase in click ranks, as the average click rank is now computed based on the position of each product that was opened.

You should notice a decrease in conversion rates, as a visit that contains multiple transactions will now be counted as one converting visit.

This is to avoid cases where conversions were greater than the number of visits.

In the previous data model, the conversion rate attributed to the recommendations service was computed using the following formula:

Recommendation conversion rate = conversions attributed to recommendations / clicks on recommendations

This means that the recommendation conversion rate was computed based on the number of transactions attributed to the recommendations service, divided by the total number of click actions on recommendations.

In Coveo’s Core model, the conversion rate attributed to the recommendations service is computed using the following formula:

Recommendation conversion rate = conversions attributed to recommendations / visits with recommendations shown

The recommendation conversion rate is now calculated by dividing transactions attributed to the recommendations service by total visits with at least one recommendation event.

You should notice a decrease in conversion rates attributed to the recommendations service, as visits with recommendations shown represents a higher number than clicks on recommendations.

In the previous data model, transactions that were sent more than once were counted multiple times.

In Coveo’s Core model, transactions that are sent more than once are counted only once.

If you were sending purchase events more than once, you should notice a decrease in transaction counts.

In the previous data model, Coveo usage analytics (UA) employed a process of lowercasing product identifiers before associating a click event with a purchase event.

For example, both a product named product123 and another named PRODUCT123 were treated as identical. Consequently, if a customer initially clicked the first product (product123) and subsequently made a purchase of the second product (PRODUCT123), the credit for that purchase would be attributed to the initial click action on the first product.

In Coveo’s Core model, product identifiers are treated as explicitly defined.

For example, a product named product123 and another named PRODUCT123 are now recognized as distinct entities. Consequently, if a customer clicks the first product (product123) and later makes a purchase of the second product (PRODUCT123), the credit for the purchase is not linked to the initial click action on the first product but rather to the event that led the user to purchase the second product.

If you were using different casing for your product identifiers, you should notice a decrease in transactions counts and revenue attributed to the different Coveo services.