Creating dashboards

Learn to construct powerful data visualizers for monitoring key metrics in your NLX workspace

Using templates

For your convenience, templates are available to use or easily build from. Simply start by selecting the + Add dashboard link:

  • Enter name of dashboard

  • If needed, modify panels by hovering & selecting from the icons that appear (edit, duplicate, remove)

  • When done, click Save


Analytics templates

Expand to read a description and panel setup of each template:

Overview template

Overview provides general traffic data happening with your application. If a system Analytics tag has been added to nodes in your intent flows, stats on these will also appear.

Panels included:

  • All conversations: Displays the total number of all conversations; a conversation is defined as a single interaction session between bot and user

  • All messages: Displays the total number of all messages sent by users

  • Traffic: Provides a graph that shows the number of conversations for each of the days defined in the From/To filter fields

  • Three analytics system tags labeled Automated, Helpful, and Not_helpful. Each gives a percentage for the metric and, below it, the aggregate ratio that led to the calculation of that percentage. The numerator in the ratio is the number of conversations or messages (depending on the data being calculated) that apply to that metric, while the denominator is the total number of conversations or messages. These analytics are based on tagging

    • Automated conversations: Displays the percentage of conversations when the user reaches a node in a flow tagged with the Automated system tag

    • Helpful conversations: Displays the percentage of conversations when the user reaches a node in a flow tagged with the Helpful system tag

    • Not helpful conversations: Displays the percentage of conversations when the user reaches a node in a flow tagged with the Not helpful system tag

Performance template

Performance monitors response times and operations of your API integrations; specifically, it provides a view into the performance of your Actions and Data requests.

Panels included:

  • Action invocations: The number of times an Action is invoked

  • Action success rate: The rate determined by the number of times an Action is invoked and successfully executed

  • Action response times: The average time it takes for an Action to be invoked and executed

  • Performance by action: The average response time (seconds) and the success percentage for each Action

  • Data request invocations: The number of times a Data request is invoked

  • Data request success rate: The rate determined by the number of times a Data request is sent and successfully retrieved

  • Data request response times: The average time it takes for a Data request to be responded to

  • Performance by data request: The average response time (seconds) and the success percentage for each Data request

User experience template

User experience focuses on how your users are experiencing a conversational AI application.

Panels included:

  • Flows recognized: The percentage determined by how many times the application is successfully able to recognize a user's intent and route to the correct flow

  • Avg NLP confidence score: The confidence level that the NLP has understood the users' utterances and has directed them to the correct flow

  • Dropoff rate: Percentage determined by how many times a user drops off after calling into the application before being routed to any flow

  • Escalation rate: Percentage determined by how many times the application escalates the user because it is not able to complete or route the user to a flow that matches their intent

  • Trending phrases: The utterances that are frequently said by users to the application, the number of times it has occurred, and whether it is trending up or down in the given time period

  • Flow performance: The average confidence score and number of conversations that have been routed to that flow


Custom dashboards

To create your own dashboard of tracked metrics and chosen charts, start by clicking the + Add dashboard link:

  • Enter name of dashboard

  • Click + Add panel button; repeat as needed

  • Hover on panel > Choose edit icon

  • Enter a name for the panel and description of what it tracks

  • Select visual Chart type > Choose from compatible metrics for your chosen chart type

  • When done, click < Back to panels link to edit other panels or select Save

🧠 When editing a panel, a checklist is conveniently provided on the right to indicate the setup requirements of your selected Chart type.


Edit a dashboard

To edit an existing dashboard in Analytics, tap the Edit dashboard link:


Chart type

Chart
Description
Example

Value

Displays a single numerical output of a chosen metric

Sum of all Data requests invoked

Gauge

Displays a numerical value of a chosen metric against its maximum range

Sum of successful SMS actions sent against the range of all SMS actions sent

Funnel

Displays stages of a metric as it filters downward

Sum of all Conversations filtered by language (Stage 1) then a specific intent (Stage 2) then an analytics tag (Stage 3)

Pie chart

Displays numerical proportions of a chosen metric within a selected group

Data requests invoked as the chosen Metric and Data request as the selected Grouping for showing all data requests created in the workspace with their proportions when triggered in conversations

Bar chart

Displays vertical columns of the values of one or more chosen metrics over time

Sum of conversations through voice channels versus conversations through chat channels (using filters), set to a time interval of each day

Line chart

Displays changes in the values of one or more chosen metrics over time

Sum of Data requests invoked that have timed out (using filters with analytics tags applied to timeout nodes), set to a time interval of every hour

Box plot

Displays the median value of a chosen metric along the range of its captured minimum and maximum values

Median value of Action response times (in seconds) displayed within its range of min and max values

Table

Displays one or more chosen metrics in rows and columns. Rows are auto-generated based on the Grouping type selected in Column 1 of your table

Sum of conversations occurring with applications and subsequent columns showing the sum of conversations occurring in different cities (using filters)


Theme

Neutral: Provides data visualization of the panel in a black scheme

🟢 Success: Provides data visualization of the panel in a green scheme

🔴 Error: Provides data visualization of the panel in a red scheme


Metric

📊 Make information visible from an external data source using the Table chart's Webhook.

Metric
Description

Actions invoked

Total number of Actions triggered during user conversations

Percentage of successful actions

Percentage of Actions triggered that were resolved during conversations

Actions response time

Length of time Actions took to resolve when triggered during conversations

Conversations

Total number of conversation sessions with users

Time spent per conversation

Length of time (displayed in seconds) of a conversation session with a user

Flows detected

Total number of flows users were directed to

Flows detected without input

Total number of flows users were directed to without use of a User input node

Flows invoked

Total number of times the conversational AI identified customer intent and invoked the appropriate flow

Time spent per flow

Length of time (displayed in seconds) users experienced a flow

Voice+ scripts

Total number of Voice+ experiences triggered in conversations by Voice+ nodes

Number of steps per Voice+ script

Number of Voice+ steps triggered by users in Voice+ experiences (may choose aggregation: Min, Max, Avg, Sum)

Time spent per Voice+ script

Length of time (displayed in seconds) users were engaged with a Voice+ experience (may choose aggregation: Min, Max, Avg, Sum)

Lifecycle hooks invoked

Total number of Lifecycle hooks triggered during user conversations

Knowledge base confidence score

Median confidence level of LLM when matching user queries to Knowledge base articles

Knowledge base invoked

Total number of times a knowledge base was invoked during user conversations

Knowledge base response time

Length of time for a response to be matched and relayed to a user

Messages per conversation

Average number of user messages per conversation session

Messages

Total number of user messages

Confidence score

Median confidence level of NLP for understanding users' utterances when matching them to an intent

Time spent per node

Length of time (displayed in seconds) users spent on user response nodes in flows

Phrase occurrences

List of captured user phrases and their total occurrences

Phrase trends

Percentage increase or decrease of phrase occurrences for a selected time interval

Response time

Length of time (in seconds) all API resources took to resolve when triggered during conversations

Users

Number of unique users to initiate a conversation

Data requests invoked

Number of Data requests triggered during user conversations

Percentage of successful data requests

Percentage of Data requests triggered that were resolved during conversations

Data requests response time

Length of time Data requests took to resolve when triggered during conversations


External analytics

To display data from an external data source, begin by creating or editing an analytics dashboard:

  • Add or edit a panel from your dashboard

  • Set Chart type to Table

  • Select Webhook as the Data source > Choose External for Type

  • Enter the URL of the endpoint (your endpoint must respond with a status code and string body of the data)

  • Click Save

Sample payload:

{          
  statusCode: 200,
  body: JSON.stringify("one,two\n1,2\n3,4\n5,6")
};

Optional

  • Use placeholders startTimestampExtended and endTimestampExtended in the URL field to pass the start and end timestamps from the dashboard to the endpoint request

  • Add one or more headers > Include name/value pair

  • Toggle Sensitive setting on to redact information from Event logs


Aggregation

  • Min: Displays the minimum value of the chosen metric

  • Avg: Displays the average, or mean, of the chosen metric

  • Max: Displays the maximum value of the chosen metric

  • Sum: Displays the total value of the chosen metric


Filters

Filters allow you to include or exclude specific resources or criteria from your data:

  • Navigate to your Analytics dashboard

  • Select Edit dashboard > Choose any panel in your dashboard

  • After selecting Chart type and Metric, expand the Filters section

  • Specify whether to include/exclude resources from data displayed

  • When done, click Save

When adding more than one item to include or exclude from your dataset, the following toggle setting is made available:

  • Any: Returns data if at least one item from the list is present in a conversation

  • All: Returns data if all items from the list are present in a conversation


Private/Public visibility

Each analytics dashboard created in a workspace may be set to one of two visibility settings:

  • Public: Visible to all who access the workspace

  • Private: Visible only to the creator of the dashboard


Setting up alerts

Alerts allow you to specify individuals that should receive emailed notifications when a particular metric's value is triggered:

  • Navigate to an Analytics dashboard > Hover on panel > Choose bell icon

  • Enter trigger criteria & the period of time over which the alert will continually be checked in 5-minute intervals

  • Select subscriptions to receive email notifications. Need to set up a subscription? See Subscriptions

Optional:

  • Region: Choose between Global or European Union, if applicable to your bot's deployment

  • Hours of operation: By default, alerts may be sent any time of day, Sunday through Saturday. You may adjust these settings to correspond with business hours, or to specify different subscribers for different intervals of the day

  • Click Create alert

Last updated