Custom Dashboard

Custom Dashboard allows you to design your own dashboards to meet business needs using a combination of built-in metrics and custom KPIs based on the user’s demographic or context information to provide a more tailored and relevant view of the data.

Using Custom Dashboard, you can select the metrics most relevant to your needs and display them in a way that is easy to understand and use. This can be particularly useful if you have a lot of user conversations and data that you need to track and monitor, as it allows you to focus on the most important information and ignore the rest.

Custom Dashboards are available along with the otherout-of-the-box dashboards in the Bot Builder. For more information on Default Dashboard, see here.

Follow these steps to build Custom Dashboards:

1. Identify the data points that you need to capture for deriving the metrics, based on your business requirements and needs.
2. Define suitable Custom Meta Tags to emit these data points in the Bot definition. To know how to add custom meta tags to your Virtual Assistant (VA), see here
3. Design widgets, preview, and update the dashboard with these widgets.

The following sections describe the steps to create custom dashboard widgets.

Go to Analyze → Custom Dashboards in the left navigation menu, to add multiple custom dashboards for your VA.

• Each of these dashboards can have one or more widgets.
• Each widget in the dashboard must be associated with a dataset, and you need to define a query to extract the required data to be displayed in a widget.
• Information related to messages, sessions, analytics, and meta tags can be used to generate the data.

Add Custom Dashboards

You can add one or more Dashboards by providing basic details of the Dashboard.

• To add a new dashboard, click on the kabab icon on the top right and select the ‘New Dashboard’ option. Enter the name of the dashboard.
• Please note that you can edit the name of the Dashboard anytime.
• By default, each custom dashboard comes with a Date Filter, allowing you to filter the records for all the widgets in the dashboard. You can choose between 24 hours, 7 Days, and a custom date range.
• You can choose a Color Theme for your dashboard.
• You can Add Widget to a given Dashboard (see below for details).
• You can reorder the widgets within a dashboard using the move cursor, visible on mouse hover over the widget, to drag and reorder the widget anywhere on the Dashboard.
• Using the Kebab menu (vertical ellipses) icon, you can do the following:
• Clone Dashboard: Copies the dashboard to a new dashboard with the same configuration
• Export Dashboard: Exports the configuration of the dashboard in a JSON file
• Delete Dashboard: Allows you to Delete the custom dashboard.

Add Widgets

You can add one or more widgets to a Dashboard using the Add Widget button. Widget configuration involves two steps:

1. Data Definition
2. Data Representation

Data Definition

Every widget must have a query defined, to retrieve and represent the required information on that Widget. Following are the configurations used to generate the query definition:

• Dataset defines the source of data. This can be:
  • Analytics: Analytics gives access to data about Success Intents, Failed Intents, Success Tasks, and Failed Tasks associated with your VA. You can view key fields like MetricType, Channel, UserId, etc.
  • Message: This dataset provides VA and User messages for your VA. You can view key fields UserId, Channel, etc.
  • Sessions: The dataset lists conversation sessions associated with your VA. You can choose to view key fields like UserId, Channel, etc.

  Note: You can view the top 20 records for all the dataset fields.

• Date Range is set by default to the past 7 days and can be customized to a range of 90 days. This date range is only used for preview purposes.
• Select fields to be depicted by the Widget.
  • The fields differ for each selected dataset. See the table for more details. For example to list the success & failed intents, metricType:
  • You can apply on these fields the aggregation functions like “min”, “max”, “sum”, “count”, or “avg”“. For example to count the total triggered intents, count(metricType), you can give a display name as an alias – count(metricType) as total.
  • Apart from these if you have defined any Custom/Meta Tags for your VA, you can use them under the appropriate heading with the following notation: userTag.tagname = value. If you declared a Message level custom tag, select the Message Dataset and enter messageTag.TagName. This would result in an empty dataset, and as a result the display would be empty.
• Filter By clause is used to extract only those records that fulfill a specified condition. You can apply the following operators on these fields: “=”, “>=”, “<=”, “>”, “<“, “in”, “not in”.
  For example, to obtain the count of a specific FAQ, the following details are entered in the query setup.

  taskName = 'What account privileges does an authorized user have?' and metricType = successtasks. While conjugating multiple conditions, they would be evaluated left to right, and this ordering cannot be changed using parentheses ()

Note: In the query setup, for Filter By, individual AND, OR operators, and multiple ANDs, ORs, can be applied, but a combination of AND/ORs, is not supported.

• Group By fields for applying the aggregating functions. For example you can display the count of all messages, grouping by the userId. For more examples of Group By, see Create Custom Dashboard
• Having clause is used to filter the results with aggregate functions as ‘Where’ keyword cannot be used here. For example, count(messageid)>10. The clause works only with the ‘Group By’ function.
  To understand how the Having clause works, see Filter the Messages using the Having Clause.
• Sort By fields (actual field names, not alias as give in the Select clause) to order the results in ascending or descending order – “asc”, or “desc”. For example to sort in the descending order of the metric type, metricType desc

Run the query to see the results in a tabular format.

Data Representation

The next step would be to render the data in a visually appealing way. The following options are available for data representation:

• Table will render the data in a simple row and column format. You can specify the Columns and their order from the Dimensions option.
• Pivot chart will help summarise the data. You can specify the Dimensions – the columns to be displayed; Metrics – the value against the column; and Overlay – the column to be considered in case a data series needs to be represented. For example, if the dimension is set to be ‘date’, metric as ‘number of chats’ and overlay to be ‘customer type’ then the number of chats per customer type where each distinct customer type as series would be displayed.
• Bar Chart can be used to depict the data across the X- and Y-axis. The results can be split into Data Series based on the Overlay field and get multiple lines plotted.
• Horizontal Bar Chart can be used to depict the data across the X- and Y-axis, a flipped version of Bar Chart. The results can be split into Data Series based on the Overlay field and get multiple lines plotted.
• Line Chart can be used to depict the data across the X- and Y-axis. The results can be split into Data Series based on the Overlay field and get multiple lines plotted.
• Pie Chart can be used for aggregation data to depict part-of-whole scenarios. Use Dimensions to set the fields to be depicted and the Metrics to set the aggregation function to be depicted.
• Donut Chart can be used similar to a Pie chart for better visualization.
• Label Chart can be used to highlight value or metric in a space of its own.

Use the Preview button to visualize the Widget. If it suits your purpose, add the widget to your to Dashboard to save the changes.
NOTE: You have to Run the query to be able to Preview it.

Widget Actions

Using the more icon (vertical ellipses) against each Widget, you can:

• Edit option will open the Widget definition page where you can make changes to an already defined Widget.
• Clone option can be used to duplicate the Widget definition and modify it.
• Export option can be used to export widget data
  • JSON format will include the final results that are displayed in the Widget UI
  • CSV format will export the results of the query that is associated with the widget (before converting the data as per the Widget UI)
• Delete option can be used to delete the widget from the Dashboard.

Limitations

• You can define a maximum of 100 custom dashboards.
• Each dashboard can include a maximum of 100 widgets.
• A maximum of 3 metrics can be added to the chart.
• Each chart can render 1 dimension.
• The custom date range can be set to a range of 90 days up to 365 days in the past.

Dataset and Fields

The dataset fields and values are listed in the following table.

Analytics

Messages

Sessions