Comprehensive and Detailed Explanation From Exact Extract:

When Tableau Data Management is enabled, Tableau Catalog provides Lineage capabilities that map connections between:

External databases

Tables

Data sources

Workbooks

Fields

Tableau documentation states that the Lineage tab for any external asset (such as a database or table):

Shows all connected workbooks

Shows dependencies and impact analysis

Allows admins to instantly assess how many analytics assets will be affected by a data warehouse migration

Option A directly uses Tableau Catalog to perform exactly this task.

Option B is unnecessary because the Catalog lineage tool already provides this information without development effort.

Option C is completely inappropriate because it offers no analysis or planning.

Option D is too time-consuming and unnecessary, especially when Tableau Catalog provides an automated dependency map.

Therefore, the correct method is to use the Lineage tab in External Assets .

Tableau Catalog lineage documentation showing how to view impacted workbooks.

External Assets and data source dependency features in Tableau Data Management.

Impact analysis best practices for data warehouse migration using Tableau Catalog.

Comprehensive and Detailed Explanation From Exact Extract:

Tableau Cloud does not natively contain a Sankey diagram in Show Me. Such advanced charts often require:

A custom extension

Or specialized templates built into Tableau Exchange

A sandboxed Viz Extension allows users to embed specialized visualization components (like Sankey diagrams) securely in Tableau Cloud. These extensions are designed for advanced chart types that are not available natively.

Accelerators provide prebuilt dashboards but are not intended for custom visual types such as Sankey.

Connectors relate to connecting to data sources, not visualization.

Show Me does not include a Sankey option.

Therefore, downloading a sandboxed Viz Extension is the correct approach.

Viz Extensions documentation explaining support for custom charts, including Sankey.

Tableau Exchange listing providing sandboxed visualization extensions for non-native chart types.

Show Me panel documentation showing Sankey is not an included chart type.

Comprehensive and Detailed Explanation From Exact Extract:

The problem:

Custom SQL joins two multi-row tables , causing → many-to-many duplication .

This artificially multiplies claim costs.

The extract becomes heavy and slow due to Custom SQL.

Tableau’s recommended solution:

✔ Use Relationships in the Logical Layer

✔ Instead of physical joins

✔ Tableau resolves many-to-many issues automatically

✔ Query is generated at the appropriate granularity to avoid duplication

This is exactly Option A.

Relationships allow the Claims facts to remain at the claim grain and Members to remain at the member grain. Tableau resolves aggregations correctly, preventing inflated values.

Why the others are incorrect:

B — Physical Join

Would continue the same duplication problem because multi-row joins multiply rows.

C — LODs

Would require complex calculations and are error-prone.

They do NOT fix the duplication in the underlying extract.

D — Table Calculations

Happen after Tableau aggregates the duplicated data — too late to fix the inflated baseline numbers.

Thus, the only correct and modern solution is relationships .

Relationships documentation explaining resolution of many-to-many granularity issues.

Guidance recommending avoiding Custom SQL for performance reasons.

Logical Layer behavior preventing row-duplication errors.

Comprehensive and Detailed Explanation From Exact Extract:

A LOOKUP table calculation requires access to rows outside the filtered date range so that the calculation can reference prior data. When a Relative Date Filter removes older data before the table calculation is evaluated, the LOOKUP loses the needed rows, causing the prior year’s data to disappear.

Tableau’s order of operations states:

Relative Date Filters act early (at the dimension filter stage).

Table calculations act very late.

To preserve table calculation context, filters must not remove necessary rows.

Two Tableau-documented approaches address this:

Explanation for B

Setting the Relative Date Filter as a Context Filter allows table calculations to operate on the full dataset needed for LOOKUP. Context filters create a separate temporary table, and subsequent filters like table calculations evaluate after the context is established.

This ensures older rows are still available to the LOOKUP function.

Explanation for C

Creating a field such as:

LOOKUP(MIN([Order Date]), 0)

and filtering on this field instead of Order Date converts the filter into a table calculation filter , which occurs after the LOOKUP computation. Tableau documentation explains that table calculation filters preserve the full dataset required for the LOOKUP window.

This ensures that the LOOKUP still has access to last year ' s values even when filtering for the current 12 months.

Why A is incorrect

Replacing LOOKUP with an LOD changes the logic entirely.

LOD expressions cannot replicate moving-window or lag-type behavior.

Why D is incorrect

DATEDIFF logic can replicate a rolling window, but hiding False values is essentially a manual filter and does not preserve the integrity of the LOOKUP’s required partitioning. It also contradicts Tableau’s recommended approach for maintaining table calculation context.

Tableau Order of Operations explaining why table calculation filters preserve data for LOOKUP.

Tableau documentation on context filters and how they allow more data to remain available for downstream table calculations.

Tableau guidance on how Relative Date Filters interact with table calculations.

Best practices for preserving table calculation window rows when filtering.

Comprehensive and Detailed Explanation From Exact Extract:

The dataset contains timestamps for each individual user interaction. Growth in user adoption over 6 months means the number of rows has expanded significantly. Tableau’s performance documentation states that large row-level datasets can cause performance degradation even when using extracts, especially when:

The visualization is aggregated to a higher level (such as daily), and

The underlying extract still contains much more granular data than needed.

Tableau recommends pre-aggregating data before it reaches Tableau Desktop, which reduces extract size, memory use, and query time. This improves performance without changing what the visualization displays.

Option D uses Tableau Prep, which is included with Tableau Creator licensing and therefore incurs no additional cost. Tableau Prep can aggregate raw timestamp data into daily totals per product feature, which matches the visualization’s actual granularity. This results in:

A dramatically smaller extract

Faster queries

No change to how the dashboard looks or functions

Option A would remove product features from the visualization, altering the dashboard content and reducing insight, which does not meet the requirement of minimal impact.

Option B requires purchasing an external ETL tool, which violates the requirement of no additional cost.

Option C reduces the number of extract refreshes but does not improve dashboard performance; the data would remain equally granular and equally slow.

Therefore, Tableau Prep aggregation is the correct solution that improves performance while maintaining the same visualization and incurring no additional cost.

Tableau performance guidelines recommending pre-aggregation of highly granular datasets.

Tableau Prep documentation stating it can be used to aggregate data before creation of extracts.

Tableau’s extract optimization guidance describing how reducing row counts improves query and visualization performance.

Comprehensive and Detailed Explanation From Exact Extract:

The requirement is to compare each quarter’s profit to the median profit over a rolling window of the last four quarters, including the current one . This is a classic use case for WINDOW_ table calculations in Tableau.

Tableau documentation explains:

WINDOW_MEDIAN( expression, start, end ) computes the median of the expression over a window of rows defined by start and end, which are offsets relative to the current row.

To create a rolling calculation that includes the current row and the three preceding rows, the window frame must span four rows ending at the current row .

Conceptually, the correct pattern is:

Current quarter’s profit: SUM([Profit])

Rolling four-quarter median: WINDOW_MEDIAN(SUM([Profit]), previous_3, current)

In actual Tableau syntax, that pattern is written with a frame that begins three rows before the current row and ends at the current row.

Among the options provided:

Options A and B use INDEX() or FIRST() as the start of the window, which creates frames anchored to either the first row or varying positions in the partition, not a consistent four-quarter trailing window.

Option D anchors the frame relative to LAST(), which makes the window depend on the final row in the partition, not a trailing four-quarter window for each bar.

Option C uses a fixed frame of four rows expressed as (3, 0) in the argument list. While, in exact Tableau syntax, a trailing 4-row frame is typically written with a negative start offset and zero as the end offset, this option is clearly intended to represent the frame “three rows back through the current row” and is therefore the only answer that matches the required rolling four-quarter window conceptually.

So, using a WINDOW_MEDIAN over a four-row frame ending at the current row, as shown in option C, is the intended solution for coloring each bar based on whether:

SUM([Profit]) > rolling_median_over_last_4_quarters

Tableau table calculation reference describing WINDOW_ functions and their start/end frame parameters.

Examples in Tableau help that use WINDOW_SUM or WINDOW_AVG with a frame spanning a fixed number of previous rows to compute rolling-window metrics.

Best practices for using WINDOW_MEDIAN to compute rolling medians over sliding time windows.