What Q&A is and why it matters
The Q&A feature in Power BI allows users to type a question in plain English and receive an automatically generated visual answer. A user who wants to know the top ten booking agents by revenue does not need to navigate to the agent analysis page, find the right visual, and apply the right filter. They type "top 10 agents by YTD business" and Power BI generates a bar chart instantly.
For business users — sales managers, operations leads, finance teams — who understand their business deeply but have no technical background in data tools, Q&A is often the most natural interface to a Power BI report. It meets them where they are, using the language they already know, rather than requiring them to learn the structure of the data model.
How Q&A works under the hood
When a user types a question, Power BI\'s natural language processing engine parses the text and maps it to fields, measures, and values in the data model. It identifies the intent (what kind of visual to show), the subject (which fields to include), the filters (which values to restrict to), and the aggregation (sum, count, average, top N).
The engine relies on field names, table names, and any synonyms you have configured to make the mapping. If a user asks for "revenue" but your measure is called "YTD Business," the query fails unless you have taught Q&A that "revenue" means "YTD Business." This is why configuration — specifically field synonyms — is the most important step in making Q&A genuinely useful.
The engine also learns from usage. When a user asks a question and Q&A misunderstands it, the Teach Q&A feature lets you correct the interpretation. That correction is stored and applied to all future queries that use the same phrasing.
Setting up Q&A on a dashboard
Q&A is available in two places in Power BI. In Power BI Service (the online platform), a Q&A question box appears at the top of dashboards by default. In Power BI Desktop, you add a Q&A visual to a report page from the Visualizations pane — it appears as an empty text box with the prompt "Ask a question about your data."
Both use the same underlying engine, but the Service version has additional features: it logs all questions users ask, enabling the review workflow that surfaces common questions and misunderstandings. The Desktop version is better for testing and training before publishing.
For Q&A to work correctly, the dataset must have relationships configured between tables, field names must be readable and meaningful (not database column abbreviations), and the dataset must be published to Power BI Service to access the full synonym and training configuration tools.
Field synonyms — the most important configuration step
Field synonyms tell Q&A that when a user says one thing, they mean a specific field or measure in the model. Without synonyms, Q&A only recognises the exact names of fields and tables as they appear in the model. With synonyms, it understands the natural language your business uses to describe its data.
In a travel platform context, the synonyms configuration might include: "bookings" as a synonym for the BookingsData table; "revenue," "business," and "sales" as synonyms for the YTD Business measure; "agents" and "partners" as synonyms for the Agents table; "destinations" as a synonym for the Countries dimension; and "last year" as a synonym for the LYTD modifier.
To add synonyms, go to Power BI Service → your dataset → Settings → Q&A setup → Field synonyms. For each table and field, you can add multiple synonym terms. Power BI also suggests synonyms based on common language patterns — review and accept the suggestions that match your business terminology.
Teaching Q&A your business terminology
The Teach Q&A feature handles cases where a user asks a question and Q&A either cannot understand it or misinterprets it. Instead of leaving the user with a blank or wrong answer, an administrator can define the correct interpretation and save it for future queries.
The workflow is: a user asks "what is the booking ratio by branch?" and Q&A returns a blank or an irrelevant visual. An admin opens Teach Q&A, finds the failed query, and defines what "booking ratio" means — mapping it to the Booking Agents Ratio measure in the model. From that point forward, any query containing "booking ratio" is correctly interpreted.
Teach Q&A is also used for defining what specific terms mean in your business context. "Active agents" might have a specific definition — agents who have made at least one booking in the current year. Teaching Q&A this definition ensures that "show active agents by branch" returns the correctly filtered result rather than all registered agents.
Reviewing questions users have asked
One of the most valuable features for maintaining a well-functioning Q&A is the Review Questions panel in Power BI Service. This panel logs every question asked against a published dataset, shows the result rating (whether Q&A understood the question), the frequency (how many times it was asked), and whether a fix has been requested.
Reviewing this log regularly reveals patterns: which questions are asked most frequently (candidates for dedicated report visuals), which questions Q&A consistently misunderstands (candidates for Teach Q&A training), and which questions return correct answers most reliably (validation that synonyms are working).
In a live dashboard used by a team of twenty people, the review log after one month will typically surface five to ten commonly asked question patterns. Addressing those ten patterns through synonym configuration and Teach Q&A training dramatically improves the experience for the entire user base.
Example queries and what makes them work
"Top agent countries by LLYTD bookings" — this works when Agents table is properly related to BookingsData, "agent countries" is a synonym for AgentCountry, and LLYTD bookings is a configured measure. The query returns a ranked bar chart automatically.
"Booking data of zone 1" — this works when the Branch-Market dimension includes a Zone column, "booking data" is a synonym for the BookingsData table, and zone values are stored consistently. The result is a filtered table showing all bookings from Zone 1 branches.
"Sort date tables by month no with day of week" — this works because the DateTable has a MonthNo column and a DayOfWeek column, and "sort" is a recognised intent by the Q&A engine. The result is the DateTable sorted as requested.
"What is the YTD supplier business by bookings data city" — this query returns a ranked bar chart of supplier business by city because "YTD supplier business" maps to a configured measure, "bookings data city" maps to the AgentCity field in the BookingsData table via the relationship, and the query intent is clearly a breakdown-by comparison.
Q&A in Power BI Desktop vs Power BI Service
In Power BI Desktop, the Q&A visual is available as a report page element. It is useful for testing during development — you can type queries and see whether the model is correctly structured to return expected results. However, Desktop Q&A does not log queries or support the full Teach Q&A and synonym configuration workflow.
In Power BI Service, Q&A appears as the question box at the top of dashboards and as a pinnable visual. Service Q&A logs all user queries, supports the full configuration toolset, and allows answers to be pinned as tiles to the dashboard. For production use, Service is where Q&A delivers its full value.
The practical workflow: build and test the data model in Desktop, publish to Service, configure synonyms and review questions in Service, iterate on the model in Desktop when field name changes are needed, and republish.
Limitations and when not to use Q&A
Q&A works best for exploratory queries where the user has a specific question and wants a quick visual answer. It is not a substitute for a well-designed report page with properly labelled visuals, consistent formatting, and narrative context. For regular, recurring reporting — daily booking summaries, monthly performance reviews — dedicated report pages are always more reliable and more communicative than Q&A.
Q&A struggles with complex multi-step analytical questions: "show me agents who booked more than average last year but are below average this year" requires DAX logic that cannot be expressed in a single natural language query. For those questions, a properly designed report visual with the right measure is the correct tool.
Q&A also requires ongoing maintenance. As the data model evolves — new fields are added, old ones are renamed, new measures are created — the synonym configuration and Teach Q&A training need to be updated. A Q&A setup that is not maintained gradually becomes less useful as the gap between configured terminology and actual model structure grows.
Interactive Q&A simulator
The simulator below demonstrates how Q&A maps natural language to data model elements — type or select a sample query to see how the engine breaks it down into intent, fields, filters, and aggregation.
In the next post in this series, we shift from tutorials to applied practice — building a KPI dashboard with gauge indicators, live scrollers, and card visuals that give leadership a single-screen view of business performance.
`