CCNA Visualize and analyze the data Questions

The correct option is B: the DonorType field must be placed in the Legend well so the scatter chart splits into one point per donor type (Individual and Corporate). In Power BI, the Legend well is what creates separate series/points by category, while the Values well holds the numeric measures (Total Donation Amount and Number of Donations). With DonorType in Values, the chart aggregates everything into a single point, which matches the symptom described.

Option A is wrong because an inactive relationship would affect filtering/aggregation, not the number of scatter points, and the model already relates Donations to Donors via DonorID. Option C is wrong because SUM and COUNT are appropriate for total amount and donation count, and incorrect aggregation would not by itself collapse categories. Option D is wrong because scatter charts do support multiple categories via the Legend well; small multiples are not required.

The correct answer is B: calculated columns are only refreshed when the dataset is refreshed. In Power BI, a calculated column is computed during dataset processing and its values are stored in the model, so it will not reflect underlying data changes until the dataset is refreshed (or the column is recalculated during that refresh). Option A does not fit because a measure is not stored row-by-row and would not appear as a column that simply fails to refresh.

Option C does not fit because incorrect DAX syntax would typically produce an error or blank values, not a stale column. Option D does not fit because calculated columns do not dynamically update regardless of the functions used; the refresh behavior is inherent to how calculated columns are processed.

The correct option is B: One-to-many from dimension to fact. In a star schema, the dimension table holds unique key values (the "one" side) and the fact table has many rows per key (the "many" side), so a one-to-many relationship from dimension to fact lets filter context propagate from the dimension down to the fact table correctly. Option A describes the same relationship from the opposite direction, which is not how Power BI models it, and it would also be a many-to-one from fact to dimension rather than the required propagation direction.

Option C (one-to-one) is wrong because a fact table typically has multiple rows per dimension key, and option D (many-to-many) is unnecessary and would introduce ambiguous filter propagation unless a bridge table is involved.

Option A is correct because the CALENDAR function is a DAX table-returning function that generates a contiguous date table, and calculated tables are created with DAX table expressions. Option B is correct because DAX table functions such as SUMMARIZE and ADDCOLUMNS return tables and are commonly used in the DAX formula bar to build calculated tables. Option C is incorrect because the 'New Table' button under the Modeling tab expects a DAX table expression, not a Power Query expression.

Option D is incorrect because M language in Power Query Editor creates queries/imported tables, not calculated tables. Option E is incorrect because there is no such right-click 'New calculated table' command in the Fields pane.

DATESINPERIOD is the correct choice because it returns a table of dates spanning a specified number of intervals (e.g., -12 MONTH) ending on a given date, which is exactly what a rolling 12-month running total requires when combined with CALCULATE and SUM. In this scenario, you would write something like CALCULATE(SUM(Sales[Amount]), DATESINPERIOD('Calendar'[Date], MAX('Calendar'[Date]), -12, MONTH)), where MAX returns the last date in the current filter context. DATEADD shifts a date range by an interval but does not by itself produce a continuous 12-month window ending on the last date.

PREVIOUSMONTH only returns the single prior month, not a 12-month span. DATESBETWEEN requires explicit start and end dates and does not automatically anchor to the last date in context with a rolling interval.

The CLV measure is defined as total revenue divided by distinct customer count. When this measure is used in a visual with Product category, the context filters both the revenue and the customer count to that category. The DISTINCTCOUNT(CustomerID) returns the number of customers who purchased at least one product in that category.

If a customer buys multiple categories, they are counted in each category's distinct count. This makes the denominator per category smaller than the total distinct customer base, leading to a higher CLV value than expected. Option A is incorrect because returns would reduce revenue, not cause higher CLV.

Option C is incorrect because using COUNTROWS would count transaction rows, not distinct customers, making the denominator larger and CLV smaller. Option D is incorrect because DIVIDE handles division by zero, and the scenario does not involve zero customers.

The correct answer is A: the many-to-many relationship is causing ambiguity in measure evaluation. In a bridge-table (ProductSales) many-to-many model, a single fact row can be reached through multiple product paths, so measures like SUM over Sales can be double-counted or misallocated when users slice by multiple related fields, producing incorrect totals. This is a classic DAX/modeling ambiguity that requires resolving the relationship (e.g., proper bridge filtering or distinct-count logic) rather than a simple setting change.

Option B is unlikely because key data type mismatches would typically block relationship creation or cause blanks, not selective wrong totals. Option C is not the root cause, since single vs. both cross-filter direction affects filter propagation but does not by itself create the many-to-many double-counting ambiguity. Option D is unrelated, as RLS would consistently hide rows for restricted users, not produce incorrect aggregate totals for executives with full access.

The correct answer is D: Bing Maps geocoding inaccuracies. Power BI's map and filled map visuals rely on Bing Maps to geocode location data (such as region names) into geographic shapes, and ambiguous or non-standard region names can be matched to the wrong boundaries, producing incorrect shapes. This is the most likely cause of misdrawn region boundaries, since the visual itself is functioning but the geocoding lookup resolves to the wrong place.

Option A is incorrect because visual choice affects suitability, not boundary accuracy. Option B is incorrect because a stale data source would show outdated values, not wrong geographic boundaries. Option C is incorrect because overlapping labels are a formatting/rendering issue, not a cause of incorrect region shapes.

Option B is correct because the Performance Analyzer in Power BI Desktop records the DAX query, visual display, and other timing metrics for each visual, letting you pinpoint exactly which visuals are slowest and target optimization efforts. Option D is correct because each field added to a visual increases the size of the DAX query result and the rendering work, so trimming visuals to only the necessary fields reduces query and render time. Option E is correct because cross-filtering and cross-highlighting force dependent visuals to re-query and re-render whenever a selection is made, so disabling interactions between unrelated visuals cuts unnecessary query and rendering overhead.

Option A is not correct because refresh frequency affects data latency and load on the source, not the rendering performance of a report page. Option C is not correct because adding more calculated measures generally increases model complexity and query cost rather than precomputing aggregations; pre-aggregation is achieved through import mode, aggregation tables, or summary tables, not by adding calculated measures.

The correct answer is B: the 'Date' table is not marked as a date table or is missing dates. Time-intelligence functions such as SAMEPERIODLASTYEAR, DATEADD, and TOTALYTD rely on a contiguous, complete date table marked as a date table; if dates are missing or the table is not marked, the measure returns blank for those months even when underlying sales exist. Option A is wrong because SUM vs.

SUMX affects row-context aggregation, not time-intelligence blank results. Option C is wrong because DIVIDE handles division by zero by returning BLANK or an alternate result, which is not the cause of missing months. Option D is wrong because single cross-filter direction is the default and correct setting for a one-to-many Date-to-Sales relationship and does not cause blanks in time intelligence.

The correct answer is C: the matrix rows do not have a hierarchy defined, with Category and SubCategory as separate fields. In Power BI, drill-down in a matrix requires a single field well containing a hierarchy (or nested fields in the Rows bucket) so the expand/collapse icon can navigate from Category to SubCategory; placing Category and SubCategory as independent fields prevents the expected drill behavior. Option A is wrong because a blank Total Sales measure would show empty values, not block the hierarchy expansion.

Option B is wrong because a Store Region slicer filters data but does not disable matrix drill-down. Option D is wrong because single-direction relationships affect filter propagation and drillthrough pages, not in-matrix hierarchy expansion.

The correct option is A: the Date table does not contain all dates from the previous year. A typical year-over-year DAX measure uses SAMEPERIODLASTYEAR or DATEADD over the marked Date table, and if the Date table is missing dates from the prior year, the time-intelligence function returns an empty set, so the measure evaluates to blank even though current-period sales exist. Time intelligence in DAX requires a contiguous, complete Date table marked as a date table; gaps in prior-year dates break the shifted filter context.

Option B is too generic and does not explain blanks tied specifically to prior-year periods. Option C is unlikely because a variable referencing current sales would still evaluate in the current context. Option D is incorrect because DIVIDE returns blank only when the denominator is zero or blank, which would not selectively blank prior-year comparisons when current sales exist.

Option A is correct because Import mode loads a compressed copy of the data into the Power BI in-memory engine (VertiPaq), so queries against aggregated data are served from memory and are typically much faster than DirectQuery, which pushes queries to the source. Option C is correct because Import mode datasets are constrained by capacity memory limits — for example, a 1 GB dataset size limit per dataset in shared/Pro capacity — which is a genuine factor when deciding between the two modes. Option E is correct because DirectQuery sends queries directly to the underlying source at query time, so it reflects near real-time data changes without the scheduled refresh latency required by Import mode.

Option B is wrong because DirectQuery does not support all DAX functions; some functions are restricted or behave differently, and certain calculated columns/measures are unsupported. Option D is wrong because DirectQuery can query large data sources — it is often chosen precisely to avoid importing huge volumes, since processing happens at the source.

Option C is correct because in DirectQuery mode every visual issues its own queries against the source, so reducing the number of visuals per page directly cuts the number of round trips and the volume of data the source must process. Option E is correct because pre-aggregating data into summary tables in the source means DirectQuery retrieves smaller, already-computed result sets instead of scanning large detail tables, which lowers query cost and latency. Option A is wrong because calculated columns in a DirectQuery model are computed at query time (or force the column to be materialized), adding processing overhead rather than reducing source load.

Option B is wrong because disabling 'Reduce queries sent to the source' removes the optimization that consolidates and limits queries, increasing the number of queries hitting the source. Option D is wrong because complex, deeply nested DAX measures translate into more elaborate SQL and heavier source-side computation, degrading rather than improving performance.

The correct option is B: create a calculated column that combines city and country to provide unambiguous location. When a map visual can't place a city because multiple places share the same name, supplying a more specific, hierarchical location string (for example, concatenating City and Country) gives the geocoding engine enough context to resolve each point uniquely. Option A is unnecessary and less precise here because adding raw latitude/longitude is a heavier modeling change than disambiguating the existing location field, and it doesn't address the root cause of ambiguous city names.

Option C does not fix the data ambiguity—it merely abandons the map visualization. Option D is irrelevant, since bubble size affects rendering of plotted points, not the geocoding of missing locations.

With a live connection to an Azure Analysis Services tabular model, Power BI Desktop is only a thin client and cannot create or edit model objects such as measures, calculated columns, or calculated tables. Measures must be authored in the source tabular model itself, so option C is correct: you add the measure to the AAS model using SSMS (via an MDX/DAX query window or Tabular Model Scripting Language) or Visual Studio with the Analysis Services projects extension, then it becomes available to the live-connected report. Option A is wrong because calculated columns cannot be created in Power BI Desktop against a live-connected model.

Option B is wrong because Power BI Desktop cannot modify the AAS model directly in a live connection. Option D is wrong because calculated tables also cannot be added in Power BI Desktop when using a live connection.

Option B is correct because SAMEPERIODLASTYEAR returns the equivalent date range shifted back exactly one year, and wrapping it in CALCULATE lets you evaluate the prior-year sales in the current row context; DIVIDE then safely computes the growth ratio (Current − PriorYear) / PriorYear without divide-by-zero errors. This is the standard DAX pattern for year-over-year percentage growth in Power BI and Analysis Services Tabular models. Option A is less appropriate because DATEADD requires an explicit number and interval and is more error-prone for a simple YoY shift, while SAMEPERIODLASTYEAR is purpose-built for this.

Option C's TOTALYTD computes a year-to-date aggregate, not a prior-year comparison, so it cannot produce YoY growth. Option D's PREVIOUSMONTH shifts by one month, not one year, so it answers a month-over-month question instead.

Option B is correct because a slicer on category lets users interactively choose the category, while a Top N filter on the product field of the visual dynamically limits the display to the top 5 products by sales within the selected category. This combination directly satisfies the requirement for user-driven category selection and automatic top-5 ranking by the visual's sales measure. Option A does not fit because a waterfall chart with category on the breakdown axis shows contribution changes, not a top-5 product ranking.

Option C is less appropriate because a ranking measure alone still requires a visual-level filter to restrict to top 5 and does not by itself provide the category slicer interaction. Option D is not suitable because a calculated column computes rank at data refresh time and cannot respond dynamically to slicer selections or the visual's filter context.

Option D is correct because time-intelligence functions like DATESINPERIOD require a dedicated Date table that is marked as a date table, ensuring a continuous, unbroken date range for accurate period calculations. Option C is correct because the Date table must be related to the Churn Date column in the fact table, which is what allows the filter context from the date table to propagate to the churn records. Option A is correct because the measure itself is built with CALCULATE to modify filter context, COUNTROWS to count the churned customers, and DATESINPERIOD to return the set of dates covering the last 30 days.

Option B is not needed because a calculated column flag is static and evaluated at refresh, and it cannot respond dynamically to slicers or report filter context the way a measure does. Option E is not needed because a disconnected table with a date range would not filter the fact table through a relationship, so it cannot drive the time-based calculation correctly.

Drill-down (option C) is correct because it lets users navigate a single visual's hierarchy from a yearly summary to quarterly and then monthly levels while preserving the ability to go back up the hierarchy. This matches the requirement of moving through levels of the same data rather than jumping to a separate page or filtering via saved states. Bookmarks (A) capture and restore view states but do not provide hierarchical level navigation.

Tooltips (B) show supplementary detail on hover and do not change the visual's level. Drillthrough (D) navigates to a separate detail page filtered by a selected value, not down a hierarchy within the same visual.

Option B is correct because each visual on a report page issues its own DAX queries against the dataset, so reducing the number of visuals per page lowers the number of concurrent queries and the rendering workload, improving report responsiveness. Option D is correct because pre-summarizing data in the source (for example, using SQL GROUP BY, views, or Power Query aggregation) reduces the volume of data loaded into the model and lets the engine scan smaller tables, which speeds up refresh and query execution. Option E is correct because cross-highlighting and cross-filtering force Power BI to re-query and re-render every affected visual whenever a selection is made; disabling them where interactivity isn't needed avoids that cascading query overhead.

Option A is not appropriate because calculated columns are computed at refresh time, stored in the model, and consume memory, whereas measures are evaluated at query time and are the recommended approach for complex aggregations. Option C is not appropriate because importing all columns bloats the model, increases memory usage, and slows refresh and queries; best practice is to import only the columns actually needed.

Option A is correct because slicers for date, product, and region let users interactively filter the report across all three required dimensions, providing direct control over the data shown. Option B is correct because bookmarks capture the current state of a report page—including filters, slicers, and visibility—allowing users to switch between predefined views without rebuilding selections. Option C is correct because drill-through pages let users right-click a data point and navigate to a detail page filtered to that specific context, enabling deeper analysis of sales by the selected dimension values.

Option D is not appropriate because disabling cross-filtering removes the interactive highlighting and filtering between visuals that improves exploratory analysis. Option E is not appropriate because relying only on default tooltips limits the contextual detail users can see on hover, reducing the overall user experience.

The combo chart (clustered column and line) with salesperson bars and average line is correct because it plots each salesperson's SalesAmount as a column while overlaying the team average as a line on the same axis, enabling direct individual-versus-average comparison. This dual-axis visual is the standard Power BI approach for comparing categorical values against a reference measure. A clustered bar chart with an average measure would show the average as another bar per salesperson rather than a single team reference, which is misleading.

A waterfall chart shows cumulative contributions or changes, not comparisons to an average. A table with conditional formatting can display numbers and color scales but does not visually juxtapose each value against the average as clearly as a combo chart.

Option A is correct because executive dashboards must center on KPIs tied directly to strategic business objectives, so leadership can immediately gauge organizational performance against goals rather than sifting through operational noise. Option B is correct because consistent colors and fonts reduce cognitive load and reinforce a credible, professional presentation, which matters when executives make rapid decisions from visual cues. Option C is correct because executives need high-level summaries first, with drill-down capability available on demand, balancing at-a-glance insight with the ability to investigate anomalies.

Option D does not belong because excessive interactive slicers add complexity and clutter, slowing comprehension rather than aiding executive decision-making. Option E does not belong because detailed tables of all underlying data overwhelm a strategic audience and belong in operational or analyst-level reports, not executive dashboards.

The correct answer is A: there is no separate date table marked as a date table. Time intelligence functions in DAX (such as TOTALYTD, SAMEPERIODLASTYEAR, and DATEADD) require a dedicated date table that is marked as a date table and has a continuous, unbroken range of dates at the day level, which a transactional fact table's Date column cannot reliably provide. Without this marked date table, the time intelligence measures will either fail or return incorrect results.

Option B is irrelevant because the schema version number has no bearing on time intelligence functionality. Option C is incorrect because the Amount column's data type does not affect time intelligence calculations. Option D is incorrect because changing the Date column's data type from dateTime to date does not satisfy the requirement for a separate, marked date table.

The correct answer is D: the Calendar table does not have a continuous date range, causing SAMEPERIODLASTYEAR to return blank for some months. Time-intelligence functions like SAMEPERIODLASTYEAR require a contiguous, gap-free date column in a marked Date table; if any dates are missing, the shifted period lookup fails and the measure returns BLANK for the affected rows. Option A is wrong because SAMEPERIODLASTYEAR works at any granularity, including month, as long as the date table is continuous.

Option B is wrong because a divide-by-zero would typically surface as an error or infinity, not blank, and DAX's DIVIDE handles it gracefully. Option C is wrong because an inactive Sales–Calendar relationship would break the measure for all rows, not just some months.

Option B is correct because aggregations let you pre-summarize large fact tables at higher grain levels (for example, daily or monthly totals) and cache those smaller tables in memory, so most report queries hit the compact aggregation table instead of scanning billions of detail rows, dramatically reducing query time. Option D is correct because every visual on a page issues its own DAX query against the dataset, so reducing the number of visuals lowers the total number of queries and the volume of data retrieved per page render, which directly improves report responsiveness. Option A is not appropriate because adding more columns to slicers increases the cardinality of filter combinations and generates larger, more complex filter context, which typically slows queries rather than speeding them up.

Option C is wrong because deeply nested calculated measures are evaluated at query time and consume CPU and memory, degrading rather than improving performance. Option E is incorrect because DirectQuery pushes queries to the source database on every interaction and generally performs worse than Import mode for large datasets, not better.

The correct answer is C: the relationship between Shipments and City is based on OriginCity, but the table shows DestinationCity, so the filter does not apply. In Power BI, cross-filtering propagates along active model relationships, so clicking a city bubble filters the Shipments table only through the column used by the map's active relationship (OriginCity); if the table visual displays DestinationCity, the filter on OriginCity does not restrict those rows. The fix is to use the same city column in both visuals or add a role-playing dimension/relationship for DestinationCity.

Option A is wrong because a legend is not required for a map bubble to act as a filter source. Option B is wrong because a table filter would not silently block cross-filtering unless it explicitly conflicts, and the scenario says cross-filtering is enabled. Option D is wrong because a slicer on Region does not prevent a map selection from filtering the table.

The correct option is D: enable multi-select in the slicer formatting options. In Power BI, the Selection Controls section of the slicer's formatting pane includes a Multi-select toggle (on by default for most slicers), and turning it on lets users Ctrl-click or check multiple product categories so the visual aggregates combined data. Changing the slicer style to dropdown (A) only alters the display, not the selection cardinality, and sync slicers (B) merely propagate an existing slicer's state across pages.

Adding a second slicer (C) creates two independent filters that intersect rather than combine categories, so it does not achieve multi-selection.

PREVIOUSMONTH (option C) is correct because it is a time-intelligence function that returns a single-column table of dates for the entire month immediately preceding the dates in the current filter context, so wrapping it in CALCULATE with SUM of sales yields the previous month's total sales. It works directly against a marked date table and shifts the context back exactly one month, which matches the requirement. SAMEPERIODLASTYEAR shifts the context back one year, not one month, so it would return the same period last year.

DATEADD shifts dates by a specified number of intervals but requires an explicit interval argument and is more general, not purpose-built for the immediately preceding month. PARALLELPERIOD shifts by a specified number of intervals and returns a full period, but it also requires an interval argument and is not the specific previous-month function.

The correct option is A: add the store name and sales fields to the tooltips bucket of the map visual. In Power BI, the Tooltips field well on a visual lets you bind additional fields (here, store name and total sales) so they appear automatically when a user hovers over a location, which is exactly the required behavior. Option B is wrong because bookmarks capture and switch report state rather than provide hover tooltips.

Option C is wrong because drill-through requires the user to right-click and navigate to a separate page, not hover. Option D is wrong because the page tooltip property configures a report-page tooltip (often for a whole page), not the field-level hover details on the map visual.

A stacked area chart is correct because it plots each product category as a cumulative band over a continuous time axis, so the total height shows overall sales while each band's thickness shows that category's contribution to the total at every point in time. This directly satisfies the requirement to show both contribution by category and the trend over time. A clustered bar chart compares discrete category values side by side but does not naturally show their combined contribution to a running total over time.

A scatter plot shows relationships between two numeric variables, not compositional contribution over time. A pie chart shows parts of a whole for a single point in time and cannot represent change over time.

The correct option is B: add a relative date slicer to the report page. A relative date slicer is a built-in Power BI slicer type that lets users pick dynamic ranges such as 'last 30 days', 'last 3 months', or 'this year', and it automatically recalculates based on the current date without any manual selection. Option A (bookmarks) only switches between pre-defined static views and cannot dynamically follow the current date.

Option C (a calculated column flagging the last 30 days) hard-codes a fixed window at refresh time and doesn't give users an interactive range control. Option D (a filter on the date field in the Filters pane) requires users to manually specify explicit start and end dates, which is exactly the manual selection the scenario wants to avoid.

A line chart with a continuous date axis (option C) is correct because a continuous axis plots dates proportionally along the X-axis, so gaps for weekends and holidays are preserved as empty intervals rather than being collapsed, giving an accurate time-based trend. Line charts are also the appropriate visual for showing trends over time, and the continuous axis correctly reflects the true temporal spacing between data points. Option A is wrong because a categorical date axis treats each date as an evenly spaced label, which distorts the time scale and hides gaps.

Options B and D are wrong because bar charts are suited to comparing discrete categories rather than depicting continuous trends over time, regardless of axis type.

The correct option is C, the Matrix visual, because it natively supports hierarchical row/column groupings (such as Year > Quarter > Month) with expand/collapse controls, which is exactly what the report requires. A matrix also supports drill-down and drill-up through the hierarchy, letting users navigate from year to quarter to month levels. Option A is wrong because a Table visual does not provide expand/collapse hierarchy interaction.

Option B is wrong because a stacked bar chart with drill-down changes the chart's level rather than offering expandable/collapsible hierarchy rows. Option D is wrong because a Card visual displays a single aggregated value and has no hierarchy support.

The correct answer is C: the variable PreviousDate is evaluated within the current filter context, which may equal LastDate for single rows. In DAX, a VAR is computed once in the filter context where it is defined; if the row context is not transitioned or the filter context already restricts OrderDate to the current row, then CALCULATE(..., 'Sales'[OrderDate] = PreviousDate) filters on the same date as the current row, returning the current amount instead of the prior purchase. To fix this, the previous date must be computed in a context that ignores the current OrderDate filter, e.g., using CALCULATE with ALL or REMOVEFILTERS on 'Sales'[OrderDate] before assigning the variable.

Option A is wrong because a text OrderDate would typically cause type or sort errors, not a consistent return of the current row's amount. Option B is wrong because ALLEXCEPT removing filters except CustomerID would actually help find prior dates, not prevent it. Option D is wrong because CALCULATE does support boolean filter syntax like 'Sales'[OrderDate] = PreviousDate.

SAMEPERIODLASTYEAR (option B) is correct because it returns a table of dates shifted exactly one year back from the dates in the current filter context, which is precisely what you need to compare this year's sales against the same period last year for a year-over-year growth measure. It is a time-intelligence function designed for this purpose and works on a marked date table, letting you write something like CALCULATE([Sales], SAMEPERIODLASTYEAR(Dates[Date])). DATEADD is more flexible (it shifts by any interval, e.g., -1 YEAR) but is not the dedicated same-period comparison function, and it can behave differently at partial periods.

PARALLELPERIOD shifts to a full parallel period (e.g., the entire previous year or quarter) rather than the same portion of the period, so it would not align day-for-day. PREVIOUSYEAR returns the whole previous calendar year regardless of the current period's partial span, which also breaks the same-period alignment needed for accurate YoY growth.

The correct option is A, the Quick Measure for 'Running total', because Power BI provides this built-in Quick Measure specifically to compute a cumulative sum of a value (e.g., sales) across a sorted axis such as a date, generating the required DAX automatically. It handles the filter context and ordering needed for a running total without manual coding. Option B is not the intended feature since writing a custom SUM/FILTER measure is a manual approach rather than the dedicated Power BI feature for this task.

Option C is unsuitable because a calculated table is static and does not dynamically respond to report filters or slicers like a measure. Option D is incorrect because a calculated column using EARLIER computes row-level values at data refresh and cannot produce a dynamic running total in the report visual.

Option A is correct because Power BI Desktop provides a dedicated Mobile Layout view (View > Mobile Layout) that lets you rearrange and resize visuals specifically for phone portrait orientation without affecting the desktop canvas. Option B is correct because small phone screens and varied lighting conditions make large font sizes and high-contrast colors essential for legibility and accessibility on mobile devices. Option D is correct because limiting each page to roughly 3-5 visuals keeps the mobile layout scannable and avoids excessive vertical scrolling and slow rendering on phones.

Option C is not a mobile best practice per se: custom visuals are chosen for analytical capability, not for mobile optimization, and some custom visuals render poorly or are not optimized for small screens. Option E is not recommended because stacking multiple slicers consumes scarce screen space and complicates touch interaction; mobile designs typically favor a single, well-chosen slicer or filter pane instead.

Setting tab order in Power BI allows keyboard-only users to navigate through report visuals in a logical sequence, which is a core requirement of WCAG 2.1 success criterion 2.4.3 (Focus Order). Option C is correct because Power BI provides built-in high contrast themes that enhance readability for users with visual impairments. Option E is correct because adding alt text to visuals ensures screen readers can convey the content to visually impaired users.

Option D is not considered a valid method for this question; while using a color-blind friendly palette is a good design practice, it is not specifically a method to enhance accessibility in the context of the exam. Option A is incorrect as animations can distract and are not an accessibility feature.

Option B is correct because drillthrough pages are purpose-built for this scenario: a user right-clicks a data point on a summary visual (such as a KPI card or chart) and navigates to a separate detail page filtered to that context, preserving the summary-to-detail flow executives need. This keeps the executive page clean with high-level KPIs while providing a dedicated, context-filtered detail page on demand. Option A is wrong because cramming all data into one table with drill-down does not deliver a true high-level KPI view and becomes unwieldy.

Option C is wrong because bookmarks merely toggle saved view states and do not pass the selected data context to a detail view. Option D is wrong because report page tooltips only show supplementary information on hover and are not a navigation mechanism for drilling into detail.

Option A is correct because with DirectQuery, every slicer and filter interaction is translated into a SQL query sent to the source, so performance depends heavily on the underlying SQL Server query efficiency and index availability; optimizing the SQL and adding appropriate indexes reduces query execution time and speeds up slicer/filter responsiveness. Option B does not fit because a live connection is used with Analysis Services (tabular/multidimensional) models, not a SQL Server data warehouse, and it would not improve the underlying query performance. Option C does not fit because increasing Power BI service capacity memory does not address slow source-side SQL execution in DirectQuery.

Option D does not fit because switching to Import mode changes the architecture and introduces refresh latency and dataset size limits, rather than directly fixing the DirectQuery query performance issue described.

Field parameters are the correct choice because they let report authors define a dynamic list of fields or measures that users can switch between via a slicer, directly changing what a card visual displays. In this scenario, a field parameter containing the candidate measures bound to the card's value field enables single-measure selection at runtime. Bookmarks (A) capture and replay saved filter/visibility states rather than swapping the measure bound to a visual.

Tooltips (B) only show supplementary detail on hover and cannot replace the card's displayed measure. Drill-through (D) navigates to a filtered detail page and does not provide measure selection.

The correct option is C: remove the LOOKUPVALUE calculated column and instead create a relationship between the tables. LOOKUPVALUE is a row-by-row DAX function that forces the formula engine to scan the target table for every row during processing and can also cause expensive query-time evaluations, whereas a proper relationship lets the VertiPaq engine resolve the lookup through the model's relationship graph, which is far more efficient. Since the customer dimension and the other table share a key, a relationship is the native, performant way to propagate the region attribute.

Option A is unrelated to DAX performance and concerns Power Query folding behavior, not calculated columns. Option B (incremental refresh) only reduces refresh time and data volume loaded, not the cost of a LOOKUPVALUE column during report queries. Option D (more RAM) may mask memory pressure but does not address the inefficient row-by-row lookup logic.

The correct answer is A: relationships between tables must be defined correctly. In a Power BI composite model, a measure can aggregate data from both a DirectQuery source and an imported local table only when the tables are related through properly defined relationships, which allows the DAX engine to propagate filters across storage-mode boundaries. Without a valid relationship (correct cardinality and cross-filter direction), the measure cannot combine the two sources.

Option B is false because composite models explicitly support measures spanning Import and DirectQuery tables. Option C is wrong because composite models are designed to mix storage modes, not force them to match. Option D is unnecessary and would defeat the purpose of using an imported local table.

The correct answer is C, a bar chart, because bar charts encode values by length along a common baseline, which makes even small differences between categories easy to compare accurately, unlike the angle-based judgments required by pie slices. In Power BI, switching the pie chart to a bar chart (or clustered bar chart) directly addresses the users' difficulty in comparing sizes. A treemap (A) uses area to represent values, which is also harder to compare precisely than length.

A donut chart (B) has the same angle-comparison limitation as a pie chart, just with a hole in the middle. A scatter plot (D) shows relationships between two numeric measures rather than comparing parts of a whole, so it does not fit this scenario.

Option D is correct because a relative date filter on the date field set to 'Last 12 months' dynamically restricts the visual to the trailing 12 complete months, automatically updating as the data refreshes. This is the intended Power BI feature for time-relative filtering and works directly on the date column without extra modeling. Option A's calculated column flag would be static and require manual maintenance, and it would not automatically roll forward.

Option B's slicer is a manual selection control, not a relative date filter, so it won't reliably show the last 12 months. Option C's measure could compute a 12-month total, but it would not by itself filter the visual to show the monthly trend across those 12 months.

Suggested Questions (B) is correct because Power BI automatically generates natural-language question prompts for a dataset, letting report authors add them so users can click a question and get a Q&A visual result without typing. The Q&A visual (C) is correct because it is the core natural-language feature: users type questions in plain English (or another supported language) into the Q&A box and Power BI interprets them against the semantic model to produce charts and tables. Synonyms for tables and columns (D) is correct because defining synonyms in the model (via Model view or Q&A linguistic schema) improves Q&A's understanding of user terms, mapping words like 'revenue' to a column such as 'SalesAmount'.

Key Influencers (A) is not a natural-language query feature; it is an AI visual that analyzes which factors influence a metric. Quick Measures (E) is not natural-language either; it is a dialog-driven way to generate DAX calculations from templates.

The correct option is A: use bookmarks with buttons to toggle between charts. In Power BI, bookmarks capture the current state of a report page—including visual visibility and chart type—so pairing two bookmarks (one showing the bar chart, one showing the line chart) with buttons lets users switch views dynamically without editing the report page. Slicers (B) filter data and cannot change a visual's chart type, and no standard slicer offers chart-type switching.

A custom visual (C) is not required and would not provide the native toggle behavior described. Drillthrough pages (D) navigate to a filtered detail page rather than toggling the chart type in place.