Metric and Dimension Definitions (Dressipi)

Attributed Revenue – Total, Direct, and Indirect

Total order revenue, including tax and shipping when available, for orders placed in the requested Date range if any item in the order was purchased within 24 hours after an interaction with a Dressipi widget. If the item purchased was clicked on in the widget, then the revenue is considered “direct attributed revenue”; in the case that the widget was on an item PDP and was displaying Outfits (Widget Type Outfits) and the item purchased was the PDP item, then the revenue is considered “indirect attributed revenue”. Note this means that Widget Type Similar Items can only produce direct attributed revenue, while Widget Type Outfits can produce both direct and indirect attributed revenue.

An order can contribute to both direct and indirect attributed revenue, but the order’s revenue will count at most once for direct and once for indirect revenue, regardless of how many items in it met the criteria above. If the order contributes to both direct and indirect attributed revenue, it will NOT be double-counted when reporting total attributed revenue.

Date filtering refers to the order date.

If the metric is split by Device Class + Platform (desktop, smartphone, mobile app, or other), the Device Class + Platform refers to the Device Class + Platform where the transaction occurred. (It is possible for a purchase to be made on a different Device Class + Platform than where the widget interaction occurred.) Each order is only associated with one Device Class + Platform, so there is no risk of double-counting across Device Class + Platforms.

If the metric is split by Department, the Department refers to the Department of the first seed item used to generate the recommendations that produced the attributed revenue (currently, this is generally the PDP item, and generally, only 1 item is used as a seed, but there may be other Placements and other situations where more than 1 seed item is used). Note this means that with multiple items in the order from different Departments, each Department will have the whole order revenue associated with it, and when the metric is split by Department the total across Departments can be greater than total attributed revenue. This metric assumes that currently, each item only has one Department associated with it.

If the metric is split by Widget Type (currently Similar Items or Outfits), the Widget Type refers to the widget that produced the attributed revenue. Note that with multiple items in the order contributing attributed revenue from different Widget Types, each Widget Type will have the whole order revenue associated with it, and when the metric is split by Widget Type, the total across Widget Types can be greater than the total attributed revenue.

If the metric is split by Placement (currently only PDP and Basket), the Placement refers to the Placement that produced the attributed revenue. Note that with multiple items in the order contributing attributed revenue from different Placements, each Placement will have the whole order revenue associated with it, and when the metric is split by Placement, the total across Placements can be greater than the total attributed revenue.

It can be split by Date, Device Class + Platform, Department, Widget Type, and Placement.

Attributed Revenue % of Total Site Revenue, Direct and/or Indirect

Direct, Indirect, or Total Attributed Revenue divided by Total Revenue, with the same splits.

Can be split by Date, Device Class + Platform

Average Order Value (AOV)

Total Revenue / Orders.

If the metric is split by Department, the Department refers to the Department(s) of the item(s) in the order. Note that with multiple items in the order from different Departments, each Department will have the whole order revenue associated with it, and when the metric is split by Department, the results may not be consistent with the total metric.

Can be split by Date, Device Class + Platform, Department

Canonical User ID

Internal Dressipi identifier for a user. Stitches together information to collect all activity (notably cross-device attribution) related to what we believe is a single user.

Can be split by: N/A

Click Through Rate (CTR)

Recommendation Clicks / Recommendation Served Impressions

Can be split by Date, Device Class + Platform, Department, Widget Type, Placement

Clickers

Count of distinct Canonical User IDs who have clicked on an item in a Dressipi widget.

Date filtering is based on the date of interaction. If the same Canonical User ID clicks on different days, they are counted as one clicker for each day but not as a single clicker over the entire period. This means that clickers are counted on a daily basis, not cumulatively over the selected time period.

If split by Device Class + Platform (desktop, smartphone, mobile app, or other), Device Class + Platform refers to the Device Class + Platform where the click occurred. The same user can count toward clickers on multiple Device Class + Platforms, so the sum of this metric, when split over Device Class + Platform, may not agree with the count without splits.

If split by Department, Department refers to the Department of the seed item used to generate the recommendations in the widget that was clicked. The same user can count toward clickers in multiple Departments, so the sum of this metric, when split over Departments, may not agree with the count without splits.

Can be split by Date, Device Class + Platform, Department, Widget Type, Placement

Conversion Rate

Converters / Interactors, i.e., count of distinct Canonical User IDs who made any purchase within 24 hours of interacting with a Dressipi widget divided by count of distinct visitors who interacted with a Dressipi widget. More specifically, Interactors are all users who interact with a Dressipi widget in a way that can be tracked to a conversion. For the Similar Items widget, Interactors are those who click on an item in the widget. For Outfits, Interactors include anyone who engages in any type of interaction with the Dressipi widget.

Date filtering is based on the date of interaction.

If split by Device Class + Platform (desktop, smartphone, mobile app, or other), Device Class + Platform refers to the Device Class + Platform of the interaction. The same user can count toward visitors or conversions on multiple Device Class + Platforms, so the sum of these metrics, when split over Device Class + Platform, may not agree with the count without splits.

If split by Department, Department refers to the Department of the seed item used to generate the recommendations for the widget where the interaction occurred, even if the purchase was from a different Department. The same user can count toward interactors or converters in multiple Departments, so the sum of these metrics, when split over Departments, may not agree with the count without splits.

Suppose the metric is split by Widget Type or Placement. In that case, the same user can count toward converters or interactors for multiple Widget Types or Placements, so the sum of these metrics when split over Departments may not agree with the count without splits.

Can be split by Date, Device Class + Platform, Department, Widget Type, Placement

Device Class + Platform

When the platform is a mobile app, then the mobile app, otherwise device class, grouped into Desktop, Smartphone, or Other.

Can be split by: N/A

Interaction Rate

Recommendation Interactions / Recommendation Served Impressions

Can be split by Date, Device Class + Platform, Department, Widget Type, Placement

Converters

Count of distinct Canonical User IDs who made any purchase within 24 hours of interacting with a Dressipi widget.

Date filtering is based on the date of interaction. If the same Canonical User ID converts on different days, they are counted as one converter for each day but not as a single converter over the entire period. This means that converters are counted on a daily basis, not cumulatively over the selected time period.

If split by Device Class + Platform (desktop, smartphone, mobile app, or other), Device Class + Platform refers to the Device Class + Platform of the interaction. The same user can count toward converters on multiple Device Class + Platforms, so the sum of this metric, when split over Device Class + Platform, may not agree with the count without splits.

If split by Department, Department refers to the Department of the seed item used to generate the recommendations for the widget where the interaction occurred, even if the purchase was from a different Department. The same user can count toward converters in multiple Departments, so the sum of this metric, when split over Departments, may not agree with the count without splits.

Suppose the metric is split by Widget Type or Placement. In that case, the same user can count toward converters for multiple Widget Types or Placements, so the sum of this metric when split over Widget Types or Placements may not agree with the count without splits.

Interactors

Count of distinct Canonical User IDs who interacted with a Dressipi widget.

Date filtering is based on the date of interaction. If the same Canonical User ID interacts on different days, they are counted as one interactor for each day but not as a single interactor over the entire period. This means that interactors are counted daily, not cumulatively, over the selected period.

If split by Device Class + Platform (desktop, smartphone, mobile app, or other), Device Class + Platform refers to the Device Class + Platform of the interaction. The same user can count toward interactors on multiple Device Class + Platforms, so the sum of these metrics, when split over Device Class + Platform, may not agree with the count without splits.

If split by Department, Department refers to the Department of the seed item used to generate the recommendations for the widget where the interaction occurred. The same user can count toward interactors in multiple Departments, so the sum of this metric, when split over Departments, may not agree with the count without splits.

Suppose the metric is split by Widget Type or Placement. In that case, the same user can count toward converters for multiple Widget Types or Placements, so the sum of this metric, when split over Widget Types or Placements, may not agree with the count without splits.

Can be split by Date, Device Class + Platform, Department, Widget Type, Placement

Item Views

Count of item page view events.

Can be split by Date, Device Class + Platform, Department

Orders

Count of distinct order IDs.

If the metric is split by Department, the Department refers to the Department(s) of the item(s) in the order so that each order can be associated with multiple Departments. When the metric is split across Departments, the count of the orders from different Departments may not agree with the total without splits.

Can be split by Date, Device Class + Platform, Department

Placement

Location of a Dressipi widget. e.g., Item page, Basket, Email.

Can be split by: N/A

Purchasers

Count of distinct Canonical User IDs who have placed an order.

Date filtering is based on the date of order. If the same Canonical User ID makes a purchase on distinct calendar days, they are counted as separate purchasers.

If split by Device Class + Platform (desktop, smartphone, mobile app, or other), Device Class + Platform refers to the Device Class + Platform where the click occurred. The same user can count toward purchasers on multiple Device Class + Platforms, so the sum of this metric, when split over Device Class + Platform, may not agree with the count without splits.

If split by Department, Department refers to the Department of the purchased item. The same user can count toward purchasers in multiple Departments, so the sum of this metric, when split over Departments, may not agree with the count without splits.

Can be split by Date, Device Class + Platform, Department

Recommendation Clicks

Count of recommendation click events.

Can be split by Date, Device Class + Platform, Department, Widget Type, Placement

Recommendation Interactions

Count of recommendation interaction events.

Can be split by Date, Device Class + Platform, Department, Widget Type, Placement

Recommendation Served Impressions

Count of recommendation served impression requests.

Can be split by Date, Device Class + Platform, Department, Widget Type, Placement

Revenue Per Visitor (RPV)

Total Revenue / Visitors

If the metric is split by Department, the Department refers to the Department(s) of the item(s) in the order. Note this means that with multiple items in the order from different Departments, each Department will have the whole order revenue associated with it, and when the metric is split by Department the results may not be consistent with the total metric.

Can be split by Date, Device Class + Platform, Department

Session

All activity associated with a Canonical User ID on a particular Device Class + Platform. A session ends after 30 minutes of inactivity. Activity includes any page views on the site.

Can be split by: N/A

Total Revenue

Total site revenue, including tax and shipping. If the metric is split by Device Class + Platform (desktop, smartphone, mobile app, or other), the Device Class + Platform refers to the Device Class + Platform where the transaction occurred. (It is possible for a purchase to be made on a different Device Class + Platform than where an item view or widget interaction by the same Canonical User ID occurred.) Each order is only associated with one Device Class + Platform, so there is no risk of double-counting across Device Class + Platforms.

If the metric is split by Department, the Department refers to the Department(s) of the item(s) in the order. Note this means that with multiple items in the order from different Departments, each Department will have the whole order revenue associated with it, and when the metric is split by Department the total across Departments can be greater than the total attributed revenue. This metric assumes that currently, each item only has one Department associated with it.

Can be split by Date, Device Class + Platform, Department

Visitors

Count of distinct Canonical User IDs who have viewed a tracked page.

If for some reason a recommendations served impression is recorded without a corresponding page view, metrics related to interactions could be inconsistent with visitor metrics.

Date filtering is based on the date of page load. If the same Canonical User ID views a tracked page on distinct calendar days, they are counted as separate visitors.

The same user can count toward visitors on multiple Device Class + Platforms, so the sum of these metrics, when split over Device Class + Platform, may not agree with the count without splits.

The same user can count toward visitors in multiple Departments, so the sum of this metric, when split over Departments, may not agree with the count without splits.

Can be split by Date, Device Class + Platform, Department

Visits

Count of sessions.

If the metric is split by Department, it refers only to sessions that include tracked activity in the Department (i.e. page views of an item in that Department).

The same session can count toward visits in multiple Departments, so the sum of this metric, when split over Departments, may not agree with the count without splits.

Can be split by Date, Device Class + Platform, Department

Widget Type

Currently, Similar Items or Outfits

Can be split by: N/A