Metrics and attributes for Zendesk Answer Bot

Note: To use these metrics and attributes in custom reporting, you need a Zendesk Explore Professional or Enterprise plan. Accounts on the Explore Lite plan have access to a subset of these metrics and attributes through prebuilt dashboards only.

Note: Zendesk has renamed our bot capabilities. Answer Bot is now Zendesk bots. Flow Builder is bot builder, and Article Recommendations are autoreplies. For more information on this change, see this announcement. In the near future, the Explore dataset described in this article will be updated with the new names.

Use this article to discover the metrics and attributes you can use to build Explore reports based on your usage of Zendesk Answer Bot. These datasets are also used for the Answer Bot prebuilt dashboards (see Overview of the Answer Bot dashboard).

For more information about how to create reports with Explore, see Creating reports.

This article contains the following topics:

• Article Recommendations dataset
• Flow Builder dataset

Article Recommendations dataset

This section contains the following topics:

• Article Recommendations dataset schema
• Article Recommendations metrics
• Article Recommendations attributes

Article Recommendations schema

Use this diagram to help you understand the elements of the Article Recommendations dataset and their relationships.

Article Recommendations metrics

This section lists and defines all metrics available in the Article Recommendations dataset.

Table 1. Article Recommendations metrics

Metric	Definition	Explore formula
Attempts	All instances in tickets and chats where Article Recommendations was available to recommend articles, regardless of whether the bot identified relevant articles to suggest.	[Answer ID]
Answers	Instances in tickets and chats where the bot identified articles relevant to a customer’s query and suggested them to the customer.	IF ([Answer status - Unsorted]="Offered" OR [Answer status - Unsorted]="Clicked" OR [Answer status - Unsorted]="Resolved") THEN [Answer ID] ENDIF
Unsuccessful attempts	Instances in tickets and chats where the bot could not identify any articles relevant to a customer’s query and did not make a suggestion to the customer.	IF ([Answer status - Unsorted]="Unoffered" ) THEN [Answer ID] ENDIF
Clicks	Counts successful Answer Bot attempts where at least one generated suggestion was clicked.	IF ([Answer status - Unsorted]="Clicked" OR [Answer status - Unsorted]="Resolved") THEN [Answer ID] ENDIF
Resolutions	Count of Answer Bot answers that resolved the end user’s request.	IF ([Answer status - Unsorted]="Resolved") THEN [Answer ID] ENDIF
Indirect resolutions	The number of Answer Bot answers that indirectly resolved the end user’s request. In some situations, articles suggested by Answer Bot don’t resolve the customer’s enquiry directly but instead point them in the right direction in the help center to find the answer to their question. These resolutions are known as indirect resolutions. Indirect resolutions are reflected in Resolutions, Indirect resolutions and Answer status fields.	D_COUNT(Resolutions)-COUNT(Resolution articles)
% Suggestion rate	The percentage of enquiries where Answer Bot offered suggestions from the total number of enquiries it attempted to answer.	D_COUNT(Offered answers)/D_COUNT(Attempted answers)
% Click-through rate	The percentage of answers clicked by end users from the total answers offered by Answer Bot.	D_COUNT(Clicked answers)/D_COUNT(Offered answers)
% Resolution rate	The percentage of enquiries resolved by Answer Bot from the total enquiries where it offered suggestions.	COUNT(Resolutions)/D_COUNT(Offered answers)
Suggested articles	Count of the suggestions made by Answer Bot.	[Suggestion ID]
Clicked articles	The number of Answer Bot suggestions that have been clicked by a user.	IF ([Suggestion clicked]="true" ) THEN [Suggestion ID] ENDIF
Resolution articles	Count of solved tickets solved through a suggestion from Answer Bot.	IF ([Suggestion resolved]="true" ) THEN [Suggestion ID] ENDIF
Rejected articles	The number of Answer Bot suggestions that have been marked as unhelpful by end users.	IF ([Suggested article rejected]=TRUE) THEN [Suggestion ID] ENDIF
% Article click-through rate	The percentage of Answer Bot suggestions that have been clicked by an end-user.	COUNT(Clicked suggestions)/COUNT(Suggestions)
% Rejection rate	The percentage of Answer Bot suggestions that have been marked as unhelpful by end users.	COUNT(Rejected articles)/COUNT(Suggested articles)
Answered tickets	Count of tickets where an Answer Bot response was generated, a suggestion was clicked, and this resulted in the ticket being solved.	IF ([Answer status - Unsorted]="Offered" OR [Answer status - Unsorted]="Clicked" OR [Answer status - Unsorted]="Resolved" ) THEN [Answer ticket ID] ENDIF
Unanswered tickets	Count of tickets where an Answer Bot response was not given.	IF ([Answer status - Unsorted]="Unoffered" ) THEN [Answer ticket ID] ENDIF
Answer clicked tickets	Count of tickets where an Answer Bot response was generated, which resulted in the ticket being resolved.	IF ([Answer Status - Unsorted]="Clicked" OR [Answer status - Unsorted]="Resolved" ) THEN [Answer ticket ID] ENDIF
Solved tickets	Count of tickets with a status of Solved.	IF ([Answer status - Unsorted]="Resolved") THEN [Answer ticket ID] ENDIF
Non-reopened solved tickets	Count of tickets with a status of Solved, that have not been re-opened.	IF ([Answer status - Unsorted]="Resolved" AND VALUE(Reopens)<1) THEN [Answer ticket ID] ENDIF
Reopened solved tickets	Count of tickets that were set to a status of Solved with Answer Bot, and were subsequently re-opened.	IF ([Answer status - Unsorted]="Resolved" AND VALUE(Reopens)>0) THEN [Answer ticket ID] ENDIF
Agent unassisted solved tickets	Count of tickets with no Assignee set, and with a status of Solved through an Answer Bot suggestion.	IF ([Answer status - Unsorted]="Resolved" AND VALUE(Agent replies)<1) THEN [Answer ticket ID] ENDIF
Agent assisted solved tickets	Count of tickets that are assigned to an agent, where the agent has replied, and the ticket status is Resolved using an Answer Bot suggestion.	IF ([Answer status - Unsorted]="Resolved" AND VALUE(Agent replies)>0) THEN [Answer ticket ID] ENDIF
% Ticket usage rate	The rate of tickets that were assisted by Answer Bot suggestions in comparison to total number of tickets.	DCOUNT_VALUES([Answer ticket ID])/DCOUNT_VALUES([Ticket ID])
% Ticket resolution rate	The percentage of tickets resolved from Answer Bot suggestions compared to the total solved ticket volume where an Answer Bot response was given.	D_COUNT(Solved tickets)/D_COUNT(Answered tickets)
% Ticket non-reopened resolution rate	The rate of ticket resolutions provided by Answer Bot suggestions in comparison to total tickets where an Answer Bot answer was generated.	D_COUNT(Non-reopened solved tickets)/D_COUNT(Answered tickets)
% Ticket unassisted resolution rate	The rate of ticket resolutions that have not been re-opened provided by Answer Bot suggestions in comparison to total tickets where an Answer Bot answer was generated.	D_COUNT(Agent unassisted solved tickets)/D_COUNT(Answered tickets)
Click time (min)	The duration in minutes between a suggestion being generated in Answer Bot, and this suggestion being clicked.	(Click time (min))
Resolution time (min)	The duration in minutes between a suggestion being generated in Answer Bot, and this suggestion solving the ticket.	(Resolution time (min))
Click time (hrs)	The duration in hours between a suggestion being generated in Answer Bot, and this suggestion being clicked.	VALUE(Click time (min))/60
Resolution time (hrs)	The duration in hours between a suggestion being generated in Answer Bot, and this suggestion solving the ticket.	VALUE(Resolution time (min))/60
Agent replies	Counts the total number of agent replies to a ticket.	(Agent replies)
Reopens	The number of times a ticket status changed from Solved to Open.	(Reopens)

Article Recommendations attributes

This section lists and defines all attributes available in the Article Recommendations dataset. You can use this diagram to help you understand the meaning of some of the attributes in the list:

Table 2. Article Recommendations attributes

Attribute	Definition
Answer status	The status of a suggestion provided by Answer Bot. The status is set depending on the end-user's response. Values include Unoffered, Offered, Clicked or Resolved.
Answer channel	The channel on which Answer Bot interacted with the end user or agent. Values include: Agents (when an agent uses the Knowledge Capture app or the knowledge section of the context panel from within a ticket) Conversation Bots (on messaging channels including Web Widget, mobile SDKs, and social messaging apps) API Email SDK (Classic mobile SDKs only) Slack Web form Web Widget (Classic only)
Answer brand	The brand of ticket corresponding to the brand of the knowledge base from which Answer Bot generated suggestions for an answer.
Answer enquiry	The end user’s question for which Answer Bot looked for suggestions. The attribute values are limited to the first 255 characters.
Answer ID	The ID of the notification with suggestions sent by Answer Bot.
Answer resolution article ID	The ID of the knowledge base article which resolved the end user’s enquiry.
Answer ticket ID	The ID of the ticket from which the Answer Bot response was triggered.
Suggested article clicked	The event of the suggestion generated by Answer Bot being clicked by an end-user. Values are True and False.
Suggested article resolved	The event of the suggested article being clicked from the answer generated by Answer Bot which led to the ticket being resolved. Values are True and False.
Suggested article rejected	The article suggestion generated by Answer Bot being marked as unhelpful by an end-user. Values are True and False.
Suggestion ID	The ID of the knowledge base article suggestion sent by Answer Bot.
Suggested article language	The language of the knowledge base article suggestion sent by Answer Bot.
Suggested article locale	The locale of the knowledge base article suggestion sent by Answer Bot. Values include EN-US, EN-GB, DE, FR, RU.
Article ID	The ID of the knowledge base article.
Article ID and locale	The ID and locale of the knowledge base article.
Article translation title	The title of the knowledge base article in a specific language.
Article translation URL	The URL of the knowledge base article.
Article author	The name of the user who originally created a knowledge base article.
Ticket ID	The ID number of the ticket.
Ticket status	The current status of the ticket.
Ticket group	The name of the group where the ticket was assigned.
Ticket assignee	The name of the user to who the ticket is assigned.
Ticket brand	The brand of the ticket.
Ticket channel	The channel a ticket was created from. For more information about the ticket channels Explore collects, see Understanding ticket channels in Explore.
Ticket form	Ticket form used on the ticket.
Ticket organization	The name of the organization associated with the ticket.
Ticket priority	The ticket’s priority.
Ticket problem ID	The ID of the associated problem ticket.
Ticket requester	The name of the user that requested the ticket.
Ticket satisfaction rating	The satisfaction rating of the ticket, Good or Bad.
Ticket subject	The subject of ticket.
Ticket tags	The tags associated with a ticket. For important information about filtering reports using tags, see Reporting on ticket tags using filters.
Ticket type	The ticket type.
Sharing agreement inbound	Affiliated instances of Zendesk Support and companies who share tickets with the current instance of Zendesk Support.
Sharing agreement outbound	Affiliated Zendesk accounts and companies tickets are shared with.
User name	The name of the user who was assisted by Answer Bot.
User role	The role of the user who was assisted by Answer Bot.
User ID	The ID of the user who was assisted by Answer Bot.
User email	The email address of the user who was assisted by Answer Bot.
User locale	The locale of the user who was assisted by Answer Bot.
User status	The Zendesk status of the user who was assisted by Answer Bot. Values are Active, Suspended or Deleted.
User tags	A list of tags associated with the user who was assisted by Answer Bot. For important information about filtering reports using tags, see Reporting on ticket tags using filters.
User time zone	The time zone of the user who was assisted by Answer Bot.
User organization name	The organization name of the user.
User organization ID	The organization ID of the user.
User organization domains	The domain name of the organization of the user, for example, zendesk.com.
User organization status	The system status of the organization of the user, either active or deleted.
User organization tags	The tags associated with the organization of the user. For important information about filtering reports using tags, see Reporting on ticket tags using filters.
Time – Answer Bot answer created	Includes several Answer created attributes that refer to the time when an answer was generated by Answer Bot containing relevant knowledge base articles.
Time – Answer Bot answer last event	Includes several Answer event attributes that refer to the last answer notification from Answer Bot.
Time – Answer Bot suggested article last event	Includes several Suggested article event attributes that refer to the last time Answer Bot triggered a suggestion.
Time – Answer Bot suggested article clicked	Includes several Suggested article clicked attributes that refer the time a suggested article was clicked from an answer generated by Answer Bot.
Time – Answer Bot suggested article resolved	Includes several Suggested article resolved attributes that refer to the time a suggested article was clicked from an answer generated by Answer Bot, which led to ticket being resolved.
Time – Ticket created	Includes several attributes that return the time and date when tickets were created.
Time – Ticket solved	Includes several attributes that return the time and date when tickets were solved.
Time – Ticket last updated	Includes several attributes that return the time and date when tickets were last updated.
Time – Article created	Includes several attributes that return the time when an article was created.
Time – Article updated	Includes several attributes that return the time when an article was last updated.

Flow Builder dataset

This section contains the following topics:

• Flow Builder metrics
• Flow Builder attributes

Flow Builder metrics

This section lists and defines all metrics available in the Flow Builder dataset.

Table 3. Flow Builder metrics

Metric	Definition	Explore formula
Total users	The number of unique users who received a message from Answer Bot. A user does not have to engage with the bot (sending a message or clicking a quick reply) to be included in this count.	[Sunshine Conversations app user ID]
Engaged with bot	The number of unique users who have sent a message to Answer Bot or responded to a prompt.	IF ([Activity event type] = "user_input_received") THEN [Sunshine Conversations app user ID] ELSE NULL ENDIF
Transferred to agent	The number of unique users who have successfully completed the transfer step from Answer Bot to an agent. A Support ticket is created for each successful transfer.	IF ([Activity event type] = "agent_transfer_completed") THEN [Sunshine Conversations app user ID] ELSE NULL ENDIF
% total users who engaged with bot	The percentage of total users who have sent a message to Answer Bot or responded to a prompt.	D_COUNT(Engaged with bot) / D_COUNT(Total users)
% engaged users transferred to agent	The percentage of engaged users who have successfully completed the transfer step from Answer Bot to an agent.	D_COUNT(Transferred to agent) / D_COUNT(Engaged with bot)
User input count	The total number of user inputs (messages of any type) received by the bot.	[Content text]
Total users in previous 30 days	The total number of users in the period from 62 days to 32 days before today.
Total users in last 30 days	The total number of users in the period from 31 days ago to yesterday.
Engaged users in previous 30 days	Users who engaged with the bot in the period from 62 days to 32 days before today.
Engaged users in last 30 days	Users who engaged with the bot in the period from 31 days ago to yesterday.
Transferred users in previous 30 days	Users who were transferred from the bot to an agent in the period from 62 days to 32 days before today.
Transferred users in last 30 days	Users who were transferred from the bot to an agent in the period from 31 days ago to yesterday.
Containment rate	The proportion of engaged users whose conversation was not transferred from the bot to an agent. The inverse of % transferred to agent.
Resolution feedback prompts	The number of times the bot asked for feedback via an “Ask if question resolved” step. Note: This metric and the following metrics are counted based on each instance of the event and not aggregated per user. So if a user went through the same flow and replied to a feedback prompt more than once, this metric counts each instance of a prompt being sent or a feedback reply interaction.
Resolution feedback replies	The number of feedback prompt replies received by the bot.
% resolution feedback replies	The proportion of times the bot received a reply after sending a feedback prompt. This is the count of replies of any type divided by the count of prompts sent.
Resolved feedback replies	The number of resolved feedback prompt replies received by the bot (for example, “Yes, problem solved”).
% resolved	The proportion of total replies that were resolved replies. This is the count of resolved replies divided by the count of total replies.
Unresolved feedback replies	The number of unresolved feedback prompt replies received by the bot (for example, “No, I still need help”).
% unresolved	The proportion of total replies that were unresolved replies. This is the count of unresolved replies divided by the count of total replies.

Flow Builder attributes

This section lists and defines all attributes available in the Flow Builder dataset.

Table 4. Flow Builder attributes

Attribute	Definition
Brand ID	The ID of the brand associated with the bot the user engaged with.
Brand name	The name of the brand associated with the Flow Builder flow the user engaged with.
Sunshine Conversations app user ID	The unique identifier of the user participating in a messaging conversation.
Activity event type	The event type recorded as a result of certain bot conversation interactions. Values include: Conversation started (journey_started), Engaged with bot (user_input_received), Transferred to agent (agent_transfer_completed), (resolution_prompt_sent), (subflow_switched), and (user_feedback_received). Events are recorded only for published answers. If answers are removed, unpublished, or renamed, historical data isn’t affected.
Channel	The Zendesk channel that the Answer Bot interaction took place on. In Sunshine Conversations terminology, a “channel” is referred to as “source.type”.
Time - Activity occurred	Includes several attributes that return the time and date when an activity (Conversation started, Engaged with bot, or Transferred to agent) happened.
Language	The standard name for a language (for example, Simplified Chinese).
Language code	The shorthand code for a language (for example, zh-CN).
User input type	The type of user input received by the bot. Possible values include Option selected, Message sent, Form sent, and Unknown.
Content type	This is effectively the same as user input type. User input type is the human friendly name, while content type is the raw value used by the system.
Content text	The text of the message sent to the bot. This includes only messages sent as a result of selecting a quick reply option (the Option selected value in the User input type attribute). Free text entered by end users is returned as blank values.
Answer ID	The ID of the answer.
Answer name	The name of the answer. This is referred to as “intent” when creating an answer in Flow Builder.
Answer type	The type of answer. Possible answer types are: answer initial (For the greeting standard response, if enabled) fallback (For the fallback standard response) unknown (For legacy and unexpected values)
Option is resolution feedback	Indicates whether the type of option selected was a resolution feedback reply, as opposed to a standard option. Possible values are true or false. An attribute of the user_input_received event.
Resolution feedback reply	Indicates how the user replied to the resolution feedback prompt (when the bot asks for feedback via an “Ask if question resolved” step). Possible values are Resolved or Unresolved. An attribute of the user_feedback_received event.