Available Datasets

Each client's Customer Insights has their 1st party data flowing to the dedicated Amazon Quicksight account, a BI tool provided by AWS. The analyses available in the Amazon Quicksight account are an example of how to leverage the data Kevel Audience collects and processes. If clients desire access to the datasets for constructing their own reports, they should contact their Customer Success Manager for assistance.

The data is first pre-processed and cleaned, resulting in the following datasets:

• User Data
• Orders Data
• Line Items Data
• Attribution Data
• Attribution Conversion Data
• Churn Data
• Session Data

Please note that all datasets are refreshed twice a week (on Mondays and Wednesday, 00h05AM - WET).

Users

Each entry of this dataset pertains to a user and their attributes, and is a snapshot of the user at the time the data was refreshed. The default settings of this dataset include unsampled data on customers and a 1% sample of website users. By using this dataset, analysts can explore questions such as:

• "After how many orders does the % of churn decrease significantly?"

The users dataset's fields are based on:

• User Attributes, e.g. each user's likelihood to buy in the next 7 days, their RFM scores, order data, and so on.
• Events performed by a user, e.g. how many add to cart events were trigged by a given user in the last 28 days.
• User Identifiers, e.g. this helps you investigate how many users have a given ID type.
• New fields related to the dataset itself, aggregations based on order data (like category) and default segments (High Spenders, Active Shoppers). Refer to the table below.
• Each client's users dataset will also have their own unique user attributes, related to the segments created in Kevel Audience, e.g. "users that bought <category> in the last 90 days".

Other available fields

Field	Type	Description
export.firstOrderValue	decimal	Total of first order placed.
export.orders.netValueExcludingFirst	decimal	Subsequent total spend by a customer, excluding the net value of their first order.
orderData.productCategories	string	Product categories purchased by a user, since thir first order.
orderData.totalProducts	integer	Product quantity purchased by a user, since thir first order.
orderData.totalUniqueProducts	integer	Unique products purchased by a user, since thir first order.
predictions.purchaseCycle	integer	The average number of milliseconds between a user's consecutive purchases.
predictions.purchaseCycle.days	decimal	The average number of days between a user's consecutive purchases.
predictions.purchaseCycle.<category>	integer	The average number of milliseconds between a user's consecutive purchases within the same category. Category-specific calculations are disabled by default. Contact your Customer Success Manager if you want to have specific categories taken into account here.
predictions.purchaseCycle.days.<category>	decimal	The average number of days between a user's consecutive purchases within the same category. Category-specific calculations are disabled by default. Contact your Customer Success Manager if you want to have specific categories taken into account here.
sampleSize	decimal	Sample size will be 1 (100%) for customers. For user profiles without orders, a 1% sample will be used in the dataset.
snapshotDate	datetime	Time of the latest dataset updated.
segment_name	boolean	Name of the segment a user belongs to at the time of the snapshot.

Orders

Each entry of the Orders Data represents an order. It can be used to answer questions, such as:

• "Which payment methods are used more often?"

The orders dataset's fields are based on:

• Events performed by any user who has made an order event, e.g. the timestamp of their first refund.

Other available fields

Field	Type	Description
order.orderIndex	integer	The sequential position of an order place within a series of transactions, indicating the order in which the purchases occurred.
userHash	string	User ID associated with an order ID. Different orders by the same user will have the same user hash.

Line Items

It's similar to the Orders Data dataset, but with a more granular level: each entry is a lineitem of an order. By using this dataset, you can explore questions such as:

• "What brands are most effective at acquiring new customers?"

The lineitems dataset's fields are based on:

• Events performed by any user who has made an order event, e.g. the timestamp of their first refund.
• Lineitem details coming from the website's order place event or offline order imports.

By leveraging joins in Amazon Quicksight, we derived two more datasets from the original Lineitems data dataset: one to explore category combinations within the same order, the other to explore category combinations by the same buyer (same or different orders). This type of analysis is aimed at exploring cross-selling opportunities.

Attribution

Each entry of this dataset is an attributable event (or session), that relates to a conversion event. There are some details that you need to consider:

• attribution model: linear. This means that every attributable event in the journey of conversion is accounted for, and have the exact same value;
• lookback window: 30 days prior to the conversion. This means that, for each conversion, only attributable events in the last 30 days are considered;
• available period: 90 days of conversion data. Because the datasets aren't refreshed everyday, we're not considering the most recent 90 days all the time, but close to that.

For additional details regarding attribution, we encourage the Customer Insights users to explore the documentation available.

By using this dataset, we can help answering questions such as:

• "Which channels were the most successful in attracting new customers?"

The Attribution Data dataset fields are based on:

• User Attributes on the users that converted within the analysed conversion window, such as when was their first order, their lifetime value, and so on.
• Fields exclusively related to attribution. See table below.

Other available fields

Field	Type	Description
attributionRow.attributableEventId	string	The attributable event identifier.
attributionRow.campaign	string	UTM campaign in the URL that initiated the session of the attributable event.
attributionRow.city	string	User location at the moment of the conversion, grouped by city.
attributionRow.clv	decimal	12 months future CLV of the acquired customers, distributed across the attributable events.
attributionRow.conversion.amount	decimal	Total conversion value.
attributionRow.conversion.productCategories	string	Product categories present in the conversion event.
attributionRow.conversion.siteId	string	The client's site identifier.
attributionRow.conversionEventId	string	The conversion event identifier.
attributionRow.country	string	User location at the moment of the conversion, grouped by country.
attributionRow.isEmpty	string	The conversion event came from an offline import, and did not match any conversion event in the system. Consequently, that conversion has no attributable event.
attributionRow.isFirstConversion	boolean	Whether the attributable event is related to a user with no previous orders.
attributionRow.medium	string	UTM medium in the URL that initiated the session of the attributable event.
attributionRow.model	string	Attribution model. Currently we only support linear attribution.
attributionRow.source	string	UTM source in the URL that initiated the session of the attributable event.
attributionRow.timestamp	datetime	Time of the attributable event.
attributionRow.value	decimal	Attributed value of the attributable event. It will depend on the attribution model.
conversionEventId	string	The conversion event identifier.
numberOfAttributables	integer	Number of attributable events for a conversion event.
attributionRow.model	string	Attribution model. Currently we only support linear attribution.
attributionRow.source	string	UTM source in the URL that initiated the session of the attributable event.
attributionRow.timestamp	datetime	Time of the attributable event.
attributionRow.value	decimal	Attributed value of the attributable event. It will depend on the attribution model.
conversionEventId	string	The conversion event identifier.
numberOfAttributables	integer	Number of attributable events for a conversion event.

Attribution Conversion

Each entry of this dataset is a conversion event, and it was built with the aim of answering questions about touchpoints of a customer journey, for example:

-"How many touchpoints does a user need to finally convert?"

Available fields

Field	Type	Description
campaignPaths	string	List of all UTM campaign that participated in a given conversion, sorted by the first to the latest. Empty values mean that there was an attributable event without a UTM campaign defined (e.g. Direct traffic).
contentPaths	string	List of all UTM content that participated in a given conversion, sorted by the first to the latest. Empty values mean that there was an attributable event without a UTM content defined (e.g. Direct traffic).
conversionEventId	string	The identifier of the conversion.
conversionValue	decimal	The value of the conversion.
firstCampaign	string	For a given conversion, the UTM campaign of its first attributable event.
firstContent	string	For a given conversion, the UTM content of its first attributable event.
firstMedium	string	For a given conversion, the UTM medium of its first attributable event.
firstSource	string	For a given conversion, the UTM source of its first attributable event.
lastCampaign	string	For a given conversion, the UTM campaign of its most recent attributable event.
lastContent	string	For a given conversion, the UTM content of its most recent event.
lastMedium	string	For a given conversion, the UTM medium of its most recent attributable event.
lastSource	string	For a given conversion, the UTM source of its most recent attributable event.
mediumPaths	string	List of all UTM medium that participated in a given conversion, sorted by the first to the latest. Empty values mean that there was an attributable event without a UTM medium defined (e.g. Direct traffic).
numberOfAttributables	integer	Number of attributable events for a conversion event.
productCategories	string	List of categories purchased in a conversion event.
sourcePaths	string	List of all UTM source that participated in a given conversion, sorted by the first to the latest.
timestamp	datetime	The date when the conversion event occurred.

Churn

Each entry of the dataset represents one month of data.

Please note that churn definition here isn't the same as in the RFM Map: it’s a slightly simplified definition, in which we determine the global median period between purchases and consider a user to be churned if they have passed that period in their purchase cycle.

Available fields

Field	Type	Description
active	integer	Number of active customers at the begining of the period. Active customers are customers whose next order is in the future.
churned	integer	Number of active customers that had become churned at the end of the period. Churned customers have exceeded their expected next order date.
cumulativeChurned	integer	Cummulative sum of customers who became churned.
sampleSize	decimal	The default is 100%, the whole customer base is considered.
start	datetime	Period to which active and churned are attached. They represent a month / year.

Session

Each entry of this dataset is a session, based on a 1% sample of users with activity in the last 90 days. This dataset provides insight on the percentage of sessions that reached a given Funnel Location.

Available Fields

Field	Type	Description
addToCarts	integer	Count of add to cart events within a user session.
checkoutSteps	integer	Count of check out step events within a user session.
finish	datetime	The time when the session ended.
orderPlaces	integer	Count of order place events within a user session.
pageViews	integer	Count of pageview events within a user session.
productCustomizations	integer	Count of product customization events within a user session.
productImpressions	integer	Count of product impression events within a user session.
productViews	integer	Count of product views events within a user session.
sampleSize	decimal	The default is 1% sample of users with activity in the last 90 days.
start	datetime	The time when the session started,
userIdIndex	integer	Index of the user in the User Database.

For guidance on how to leverage the datasets, please refer to the documentation on Getting Started with Customer Insights and Navigating the Dashboard guides.