CFM Copilot helps you explore and analyze survey response data using natural-language questions. You can use it to summarize data, review metrics, compare segments, identify trends, analyze customer feedback, and investigate changes in customer experience metrics.

Understanding which query types are supported can help you phrase questions effectively and get more relevant results

Supported Query Types

CFM Copilot supports reporting queries that use the survey data available in your current context. This includes questions about dashboards as well as questions that require calculations or analysis of response data.

Descriptive Queries

Use descriptive queries when you want to understand the current or aggregated value of a metric.

For example:

• “What is the average CSAT this month?”

• “What percentage of respondents selected ‘Very satisfied’?”

• “How many responses do we have for Chat support?”

Descriptive queries are useful when you need a direct answer about values, counts, percentages, averages, or distributions.

Comparative Queries

Use comparative queries to compare a metric across different groups, entities, or periods.

For example:

• “Compare NPS across regions.”

• “Which account tier has the highest satisfaction rating?”

• “Compare Completion Rate across regions.”

CFM Copilot can return ranked or comparative breakdowns and relevant response volumes based on the available data.

Trend Queries

Use trend queries to understand how a metric has changed over time.

For example:

• “How has NPS changed since last month?”

• “How has NPS changed quarterly across Arizona locations?”

CFM Copilot can identify changes across reporting periods and highlight relevant variations or outliers.

Diagnostic Queries

Use diagnostic queries when you want to investigate where a change in a supported metric appears to be concentrated.

For example:

• “Why is NPS decreasing?”

• “Why did Europe CSAT drop in Q3?”

CFM Copilot can examine patterns such as changes in response volume, question-level performance, sentiment, segments, and other related factors.

Qualitative Queries

Use qualitative queries to understand themes, sentiment, and patterns in customer feedback.

For example:

• “What do customers most complain about?”

• “What are the top complaints from low CSAT responses in the last 60 days?”

CFM Copilot can identify thematic clusters and sentiment patterns and surface representative customer feedback where supported.

Text Fields support structured facets such as sentiment and labels. Keyword-search queries against Text Fields are not supported.

Segmentation and Drill-Down Queries

Use segmentation queries when you want to break down results for a particular population or dimension.

For example:

• “Show CSAT by region.”

• “Show CSAT by store tier.”

• “Show sentiment for detractors.”

CFM Copilot supports segmentation and hierarchical drill-down scenarios using data available within the current survey or dashboard context.

Relationship and Correlation Queries

Use relationship queries to investigate associations between eligible variables or metrics.

For example:

• “How does sentiment correlate with NPS?”

• “Is there a relationship between satisfaction and another supported variable?”

CFM Copilot can select an appropriate statistical method based on the query and eligible variable types. You do not need to specify the statistical test, although you can explicitly request one and CFM Copilot will validate whether the variables are eligible.

Supported Question and Field Types

CFM Copilot supports analysis for several survey question types, including:

• Multiple Choice

• Matrix

• Slider

• CSAT Rating

• Text Field for structured facets such as sentiment and labels

• NPS Rating

• Rating Scale

• Scale

• Rank Order

For custom fields, Numeric and Picklist types are supported. Response, Profile, Transaction, and Hierarchy field types are also supported within the applicable reporting scope.

Multi-select questions are supported by evaluating individual options separately. Where appropriate, p-values can be aggregated to provide a combined signal.

Unsupported Queries

Some requests fall outside CFM Copilot's analytical scope.

Action-Based Queries

CFM Copilot does not perform platform actions or modify product configuration.

For example:

• “Create a new dashboard for Support CSAT.”

• “Add a widget that shows CSAT trend by week.”

• “Plot a bar chart for NPS by region on Tab 2.”

Requests to modify rules, perform case actions, or create or modify dashboards and widgets are not supported.

Metadata-Only Queries

Questions that can be answered only from survey metadata are not supported.

This includes questions about:

• Survey title or description

• Exact question text

• Question type

• Question options

• Custom field configuration

For example:

• “What is the survey title and description?”

• “What is the exact wording of the NPS question?”

• “List all question options for the ‘Reason for score’ question.”

• “Which questions have custom fields attached?”

Unsupported Question and Field Types

The following survey question types are not supported for the documented analysis scenarios:

• Text/Media

• File Upload

• Heat Map

• Audio/Video

• Conjoint

• Image Upload

Case and Distribution Case fields are also not supported.

Data Scope and Analysis Considerations

CFM Copilot performs analysis within your current context. The dataset is determined by the selected survey or dashboard, RowScope, active filters, and applicable time window. CFM Copilot should not silently broaden that scope.

Keep the following considerations in mind:

• Permissions: Results respect your data access, field visibility, dashboard restrictions, and masking or redaction rules.

• Missing values: For applicable statistical analyses, responses with missing values for the variables being analyzed are excluded rather than automatically imputed.

• Small samples: When the available data is too limited or sparse, CFM Copilot should indicate the limitation instead of presenting an overconfident conclusion.

• Unsupported variable combinations: If the variables are not suitable for the requested analysis, CFM Copilot can provide a fallback or recommend a more appropriate analysis.

• Follow-up questions: CFM Copilot can retain the current statistical context for follow-up questions, such as narrowing an analysis to a particular channel or segment.

FAQs