Analytics-Con-202 Sample Questions Answers

Optimize Model is the feature designed to assess semantic models for AI readiness and recommend improvements. Salesforce describes Semantic Model AI Optimization as evaluating model quality, diagnosing problems, and supplying guided remediation that improves reliability for Tableau Agent. After the feature is enabled, an Optimize Model indicator appears in Semantic Model Builder and shows the model's AI-readiness strength.

Opening the Optimize Model panel provides a breakdown of AI-readiness indicators together with recommended corrective actions. The rating is recalculated as the semantic model changes, allowing consultants to iteratively improve metadata, structure, and agent compatibility.

Q & A Calibration serves a different purpose: it tests representative natural-language questions and helps calibrate how Tableau Agent responds. Business Preferences provide explicit business context and rules, such as how ambiguous terminology should be interpreted. Neither performs the automated holistic readiness evaluation described in the question.

References/Topics: Data Setup - > Semantic Model AI Optimization - > Optimize Model - > AI-Readiness Strength - > Recommended Actions.

===========================================================================

Tableau Next provides native interoperability with Tableau Cloud Published Data Sources (PDSs). Salesforce allows a consultant to create a Data 360 semantic model by connecting directly to an existing PDS after establishing the required trust between the Salesforce org and Tableau Cloud site.

The PDS remains the external data definition and query source. Tableau Next inherits fields such as their names and data types, and the semantic model can then be enriched with calculated fields, metrics, descriptions, and business context. Critically, Salesforce states that this approach allows organizations to use existing Tableau content without data or metadata migration.

That precisely addresses the customer's concern about preserving substantial prior investment in Tableau Cloud modeling and calculations.

Options A and B introduce mechanisms that are not the native Tableau Next interoperability pattern documented for PDS reuse. There is no requirement to translate the PDS into Snowflake SQL or convert it into a CRM Analytics recipe.

This architecture enables Tableau Next to extend rather than replace mature Tableau investments, while exposing existing governed analytical logic to Tableau Agent.

References/Topics: Embedding, Cross-Cloud, and Interoperability - > Tableau Cloud PDS - > Tableau Semantics - > Create Semantic Model from Published Data Source.

===========================================================================

Tableau Next asset sharing and Data 360 data access are separate layers. Granting users Viewer access to a workspace can provide inherited access to its dashboards, visualizations, and semantic models, but it does not automatically grant access to the underlying Data 360 data. Salesforce explicitly states that access to a semantic model requires access to the data space containing the model as well as access to the objects and fields used by that model.

Therefore, C is correct.

End users do not require Edit access to underlying Data 360 objects merely to view dashboard data, making A unnecessarily privileged. Similarly, B is incorrect because Consumer-level permission sets can view Tableau Next assets; Platform Analyst or Self-Service Analyst permissions are not universally required for dashboard consumption.

The architecture can be understood as three layers: license/permission access, Tableau Next asset sharing, and Data 360 data visibility/governance. Successful dashboard rollout requires all applicable layers to authorize the user. Sharing alone cannot override an administrator's data-space, DMO, DLO, field, or policy restrictions.

References/Topics: Managing Workspaces and Orgs - > Asset Sharing - > Data Spaces - > Data 360 Governance - > Semantic Model Access.

===========================================================================

Full Refresh is required when the accelerated Tableau Next cache must reflect records that were deleted in the external source. Salesforce documentation distinguishes the two acceleration modes explicitly. An Incremental Refresh adds or updates records that changed after the previous refresh, but it does not update deleted records.

By contrast, a Full Refresh removes all previously cached data during each refresh cycle and replaces it with a newly retrieved copy of the current source dataset. If a source record has been deleted, it no longer exists in the replacement dataset and consequently disappears from Tableau Next.

Incremental refresh can operate as frequently as every 15 minutes, making it attractive for high-frequency changes, but that frequency does not alter its deletion limitation. If deletion synchronization is a requirement, the architecture must account for periodic or continuous full-refresh behavior according to supported refresh intervals.

“Snapshot refresh” is not one of the documented Tableau Next Acceleration refresh methods for this use case.

This is the same architectural principle tested earlier: Incremental = added/changed data; Full = complete cache replacement, including removal of deleted source records.

References/Topics: Data Setup - > Acceleration for Data Connections - > Cache Refresh Method - > Full versus Incremental Refresh.

===========================================================================

The documented limitation of Tableau Next Incremental Refresh for accelerated external connections is that it does not process deleted records. Salesforce states that an incremental refresh updates cached data with records that have been added or changed since the previous refresh, but deleted source data is not updated in the cache.

Consequently, options B and C are incorrect. Inserts are specifically part of incremental processing, as are updates to existing records. The limitation applies only to deletions.

This matters architecturally when data freshness requirements include removal of records. If the external source deletes a transaction, customer, product, or other row, repeated incremental refreshes can leave the old cached representation present. A Full Refresh is required to reconcile the cache completely because full refresh deletes the existing cached dataset and replaces it with a fresh source dataset.

Consultants should therefore choose refresh mode based on more than refresh frequency. Incremental refresh is efficient and can run frequently, but Full Refresh is required when complete source-state reconciliation, including deletions, is a business requirement.

References/Topics: Data Setup - > Data Acceleration - > Incremental Refresh - > Full Refresh - > Deleted-Record Handling.

===========================================================================

Business Preferences are the semantic-model mechanism for supplying organization-specific instructions and contextual guidance to Tableau Agent. They can define how the analytics agent should interpret business concepts, terminology, and preferred response conventions. Salesforce recommends keeping each Business Preference short, focused, and high-value because these instructions are added to the agent's prompt.

Accordingly, preferences such as “Use GBP when presenting monetary values” or “Keep responses concise” belong in Business Preferences rather than individual field descriptions. A field description explains what a particular semantic field means; it is not the proper location for broadly applicable output behavior. The Einstein Trust Layer addresses AI trust, security, privacy, and governance controls rather than organization-specific response formatting.

References/Topics: Agentic Experiences - > Business Preferences - > Organizational Terminology - > Response Guidance - > Semantic Model Context.

===========================================================================

Tableau Agent conversational analytics is delivered through the Analytics and Visualization agent template. Salesforce's current documentation states that enabling Tableau Agent makes the Analytics and Visualization template and its prebuilt capabilities available in Agentforce Builder or Agentforce Studio. The consultant then creates an agent from that template and includes the Data Analysis subagent to enable conversational analytical responses and generated visualizations.

Therefore, B is correct.

Einstein Discovery Prediction is associated with predictive analytics and is not the preconfigured framework for Tableau Agent's conversational analytics. Tableau Next Marketplace contains reusable analytical assets and templates but is not the mechanism for constructing the core Analytics and Visualization agent.

References/Topics: Agentic Experiences - > Tableau Agent - > Analytics and Visualization Template - > Data Analysis Subagent - > Conversational Analytics.

===========================================================================

When Acceleration for Data Connections is enabled, Tableau Next temporarily caches external-source data in Data 360 to reduce runtime latency and improve query performance. Two refresh approaches are available: Full Refresh and Incremental Refresh. Salesforce explicitly states that an incremental refresh processes records that have been added or changed since the previous refresh but does not update deleted data. Therefore, records removed from the source can remain present in the accelerated cache even after several incremental refresh cycles.

This explains why the 15-minute frequency does not solve the issue. Frequency controls how often incremental processing occurs; it does not change what the incremental algorithm processes. Option A incorrectly attributes the behavior to ordinary cache latency. Option C is also too broad because acceleration itself supports deletion synchronization when a Full Refresh is used. A full refresh removes the existing cached dataset and replaces it with the current source dataset.

References/Topics: Data Setup - > Data Connections - > Acceleration - > Full versus Incremental Refresh - > Cache Refresh Method.

===========================================================================

The semantic model should form a single coherent connected cluster whenever the objects are intended to participate in the same analytical domain. Therefore, the consultant should establish valid relationships for the disconnected objects or remove them from the model when they do not belong.

Salesforce's current AI-readiness guidance explicitly warns about island tables and isolated objects. Disconnected objects can confuse both users and Tableau Agent because a query can imply relationships that the model has not defined. Salesforce recommends ensuring that all relevant objects are part of a single connected cluster and avoiding ambiguous or unjoined objects that could cause invalid agent responses.

Leaving objects isolated, as proposed by A, therefore preserves the exact modeling condition that decreases agent confidence. Option B is also architecturally incorrect. Business Preferences provide business terminology and interpretation guidance; they are not a substitute for structural relationships between data objects.

This principle is especially important for Tableau Agent because its semantic query generation relies on the relationship graph to determine valid navigation paths between dimensions and measures. A connected, unambiguous model substantially reduces unsupported joins and misleading answers.

References/Topics: Data Setup - > Semantic Model AI Readiness - > Connected Model Structure - > Relationships - > Remove Ambiguous Objects.

===========================================================================

The limiting factor in this scenario is the user's license level. Setting proactive data alerts through Tableau Agent is an authoring-oriented capability that requires Tableau Next Creator access. A Consumer user can view and interact with published analytical content, but the Consumer persona does not receive the full set of creation and configuration capabilities needed to define alert conditions.

The fallback semantic model in option B is used when Tableau Agent lacks clear analytical context, such as when a user invokes the agent from a surface that does not already identify the relevant Tableau Next asset. Here, the user is explicitly accessing Tableau Agent from a page that contains Tableau Next content, so lack of fallback scoping is not the primary issue. Option A can prevent analysis of a particular metric, but the question establishes a broader inability to set the alert and emphasizes that the user has a Consumer license.

The appropriate remediation is therefore to use the Creator-level entitlement for users who must configure proactive alerts. This preserves the distinction between consumption permissions and analytical creation/configuration permissions.

References/Topics: Agentic Experiences - > Tableau Agent - > Proactive Data Alerts - > Creator Permissions.

===========================================================================

The Tableau Next Analytics Embedding SDK supports an AnalyticsAgent component with both single-context and multi-component operating modes. In Multi-Component mode, the application omits the contextConfig property. The agent then automatically tracks the embedded AnalyticsDashboard and AnalyticsMetric components present on that page.

Therefore, B directly describes the intended architecture for a portal containing multiple analytical components.

Option A describes single-context mode, where contextConfig explicitly binds the agent to one dashboard, metric, or semantic model. That mode exists but is not a platform limitation. Option C introduces a same-semantic-model restriction that the current Embedding SDK documentation does not impose. The SDK explicitly describes automatic context tracking across embedded dashboard and metric components.

This capability is particularly important in embedded analytics because it allows conversational analytics to understand the broader analytical surface rather than forcing the external application to continually rebind the agent as users move between embedded components.

References/Topics: Embedding, Cross-Cloud, and Interoperability - > Tableau Next Embedding SDK - > AnalyticsAgent - > Multi-Component Mode.

===========================================================================

Tableau Agent's conversational analytics depends on an AI-ready Tableau semantic model, making A correct. Salesforce positions Tableau Semantics as the governed business layer between connected Data 360 sources and analytical or AI consumption experiences. Semantic models define standardized measures, dimensions, relationships, calculations, metrics, terminology, and business context that Tableau Agent uses to interpret natural-language questions.

For Snowflake data, the consultant establishes an appropriate Data 360 connection—including supported zero-copy federation where appropriate—and exposes the required Snowflake data through Data 360. The consultant then builds and prepares a semantic model containing the marketing concepts that Tableau Agent must analyze. Salesforce's Tableau Agent implementation guidance explicitly requires determining the required data, connecting it through Data 360, and creating suitable semantic foundations before configuring conversational analytics.

CSV export is unnecessary and creates an avoidable duplicate dataset. Building dashboards directly in Snowflake also does not establish Tableau Agent's governed semantic context.

The architecture is therefore: Snowflake - > Data 360 connectivity - > Tableau Semantic Model - > Tableau Agent - > Conversational Analytics.

References/Topics: Data Setup - > Snowflake Integration - > Data 360 - > Tableau Semantics - > AI-Ready Semantic Models.

===========================================================================

The correct approach is to use the Tableau Semantics connector from Tableau Cloud. Salesforce explicitly supports connecting Tableau Desktop and Tableau Cloud directly to semantic models defined in Data 360/Tableau Semantics. This allows downstream Tableau authors to consume the same governed measures, dimensions, calculations, relationships, metrics, and business definitions that Tableau Next uses.

Option B would undermine the principal benefit of the semantic layer. Reconnecting directly to Data 360 objects and recreating joins would force analysts to reconstruct business logic and could lead to inconsistent calculations or relationships across analytical applications. Option C is also incorrect because Data Kits are Salesforce/Data 360 deployment mechanisms; they are not used to install semantic models into Tableau Cloud as independent Tableau data models.

This interoperability capability is strategically important: Tableau Semantics provides one governed analytical definition layer that can be consumed across Tableau Next, Tableau Cloud, Tableau Desktop, AI experiences, and other supported Salesforce applications. Analysts can therefore create custom Tableau Cloud visualizations without duplicating semantic logic.

References/Topics: Embedding, Cross-Cloud, and Interoperability - > Tableau Semantics Connector - > Tableau Cloud - > Reusable Semantic Models.

===========================================================================

For a live Tableau Next preview inside a Slack canvas, Salesforce instructs users to paste the Tableau Next asset URL into the canvas and select Card from the Paste As menu. The resulting card can display a metric, dashboard, or visualization and regularly refresh its underlying data.

This behavior differs materially from sharing a static Tableau Next preview in an ordinary Slack channel or direct message. In a Canvas, Salesforce retrieves the data from each viewer's perspective. Consequently, each person must have appropriate Tableau Next access to the underlying asset and data. Unauthorized users receive an access message instead of seeing protected data. This preserves individual data governance while still providing a rich collaborative experience.

Option A does not describe the documented Slack Canvas workflow. Option C produces a static image and therefore fails the requirement for regularly refreshed information.

The exam distinction is important: standard Slack message preview = snapshot governed by sharing behavior; Slack Canvas Card = live, refreshable preview evaluated against each viewer's access.

References/Topics: Embedding, Cross-Cloud, and Interoperability - > Tableau Next in Slack - > Slack Canvas - > Paste As Card - > Viewer-Based Data Access.

===========================================================================

For a standard Tableau Next Slack message preview, Salesforce explicitly warns that anyone who can see the Slack message can see its preview, including filters applied when it was shared. Consequently, users should verify that the preview does not expose inappropriate sensitive information before posting it into a channel or conversation.

Tableau Next's Slack integration generates previews according to the sender's permission access and Tableau Next governance context. Once that snapshot-style preview is shared, visibility of the message governs who can see that static representation.

Therefore, C correctly describes the behavior. There is no indication that Data 360 security has malfunctioned.

Option A is wrong because recipients do not automatically obtain unrestricted interactive access to the underlying Tableau Next asset. Option B incorrectly assumes that each static Slack preview is dynamically authorization-checked as though the recipient were opening Tableau Next.

A critical distinction is that Slack Canvas cards behave differently: Canvas previews can refresh using each viewer's Tableau Next access, and unauthorized users cannot see the live underlying content.

References/Topics: Embedding, Cross-Cloud, and Interoperability - > Tableau Next in Slack - > Shared Previews - > Governance.

===========================================================================

The administrator must enable Collaborate with Slack and Tableau Next, making A correct. Salesforce's documented setup requires connecting the Salesforce org to Slack and then navigating to Tableau Next Administration settings to confirm that Collaborate with Slack and Tableau Next is enabled.

For Inspector Proactive Data Alerts specifically, Salesforce states that Slack must first be enabled for Tableau Next. Users can then receive Inspector alerts through Slack after the Agentforce app and associated connection are configured.

Option B is not the documented prerequisite. Tableau Next relies on appropriate Tableau Next permissions plus Salesforce/Slack identity and connection configuration; assigning a generic “Slack User” permission set is not the required answer.

Option C is also inaccurate as written. A properly configured Analytics and Visualization agent must contain the relevant Data Analysis and Data Alert Management capabilities, but administrators do not necessarily need to create an entirely new agent solely because Slack is being enabled; an existing qualifying agent can be configured and activated.

References/Topics: Embedding, Cross-Cloud, and Interoperability - > Tableau Next with Slack - > Inspector Alerts - > Collaboration Settings - > Agentforce App.

===========================================================================

Tableau Next applies a deterministic precedence model for conditional formatting. When multiple conditions overlap, Tableau Next attempts to combine their formatting. When the formatting definitions actually conflict—for example, two matching rules specify incompatible formatting for the same property—the most recently created rule takes precedence. Therefore, B precisely matches documented Tableau Next behavior.

The conditions are not discarded simply because they overlap, so A is incorrect. Likewise, Tableau Next does not give permanent precedence to the earliest rule, making C incorrect.

This behavior is important when designing KPI text components or executive dashboards containing several threshold-based rules. Rule ordering can materially affect what users see when a measure simultaneously satisfies multiple business conditions. Consultants should therefore inspect overlapping ranges and understand which formatting attributes can coexist versus which produce a direct conflict.

Salesforce specifically documents two applicable principles: overlapping conditions have their formats combined, and conflicting rules resolve in favor of the most recently created rule.

References/Topics: Visualizations and Dashboards - > Analyze and Share Data - > Create Effective Dashboards - > Conditional Formatting.

===========================================================================

User Interaction Logging is the capability used to begin collecting interaction data for governance and adoption analysis. In this scenario, Cloud Kicks needs visibility into how users interact with Tableau Next dashboard assets so administrators can evaluate usage patterns, adoption, and engagement. Enabling interaction logging creates the foundation for capturing those user events and making them available for downstream monitoring and analysis.

Tableau Next Auditing is associated with broader administrative audit and governance use cases, while Session Tracing is oriented toward tracing agent sessions and AI behavior. Neither is the specific capability identified here for initiating dashboard interaction collection. The requirement is centered on user activity against Tableau Next assets rather than agent reasoning or session diagnostics.

A consultant should distinguish telemetry collection from downstream analysis. First, interaction events must be captured through User Interaction Logging; those events can then be analyzed to understand which assets are used, how users engage with dashboards, and where adoption or governance attention is required. This supports operational oversight without changing the analytical content itself.

References/Topics: Basic Setup and Admin - > Tableau Next User Interaction Logging - > Governance and Adoption Monitoring.

===========================================================================

Agent Scoping determines which approved semantic models Tableau Agent can search when a user's question lacks sufficient analytical context. Salesforce describes Agent Scoping as defining a fallback list of semantic models for a particular analytics agent. This fallback is used when Tableau Agent cannot infer the appropriate model from the current page or analytical asset.

Examples include asking Tableau Agent a question from Slack or embedding the agent on a page that contains no Tableau Next metric or dashboard. In those circumstances, the agent uses its approved fallback models to determine the appropriate semantic context.

When a user asks a question from a dashboard or metric that already supplies clear analytical context, Salesforce states that Tableau Agent uses the semantic models associated with that asset instead of the fallback list. User permissions remain enforced at runtime; placing a model on the fallback list never grants unauthorized access.

Option B is incorrect because Agent Scoping does not enforce row-level security; Data 360 governance and user permissions perform that function. Option C is unrelated to agent routing.

References/Topics: Agentic Experiences - > Tableau Agent - > Agent Scoping - > Fallback Semantic Models - > Context Resolution.

===========================================================================

The semantic model should be enriched so the agent can associate the organization's business language “Earnings” with the existing Profit Margin concept. Therefore, B is the correct answer. Semantic models provide the business-context layer that allows Tableau Agent to translate natural-language terminology into governed analytical definitions. Salesforce recommends clear, contextual names and descriptions and explicitly highlights synonyms as an important semantic consideration for agent readiness.

Salesforce's Tableau Semantics training also explains that synonyms broaden the agent's understanding of terminology used by business users, allowing alternate terms to map back to authoritative metrics and definitions.

Creating a duplicate metric merely to support another phrase introduces redundant semantic definitions and increases ambiguity. Salesforce specifically recommends eliminating redundant or overlapping calculated definitions rather than multiplying them. Changing aggregation from Average to Sum would alter mathematical behavior and does nothing to resolve a terminology-discovery problem.

References/Topics: Data Setup - > Create Semantic Models - > Design Semantic Models for AI Readiness - > field descriptions, synonyms, ambiguity resolution.

===========================================================================

Tableau Next provides generative AI capabilities that translate a user's natural-language description of a calculation into a calculated-field formula. This directly addresses scenarios where an analyst understands the required business logic but lacks familiarity with Tableau Next formula syntax.

With Draft with Einstein, the user describes the calculation they need in natural language. Einstein can propose the field name, description, data type, aggregation type, and formula. The agentic calculated-field experience also supports conversational refinement, allowing the user to explain requirements and respond to clarification questions before reviewing the proposed field.

A batch transform, as described in A, is a data-preparation mechanism and is unrelated to helping an analyst overcome formula-authoring syntax problems. C is less accurate because the defining generative-AI capability being tested is the transformation of natural-language analytical intent into a proposed formula.

The generated output remains subject to user review before creation, preserving human control over the final semantic definition while substantially reducing formula-authoring friction.

References/Topics: Agentic Experiences - > Generative AI - > Draft with Einstein/Data Pro - > Natural-Language Calculated Fields.

===========================================================================

Batch Data Transforms are designed to perform repeatable data-shaping operations before downstream analytical consumption. Salesforce describes batch transforms as canvas-based processing flows capable of adding, manipulating, joining, aggregating, filtering, and applying formulas or other transformations to data before writing the resulting dataset to a target object.

This makes A the appropriate solution when the requirement involves multiple structural transformations—such as pivots, aggregations, and calculations—before the resulting dataset becomes part of a semantic model.

Identity Resolution solves a different problem: reconciling and matching records across data sources to build unified profiles. It does not provide the generalized transformation pipeline required here.

Calculated fields can create derived values, but they are insufficient when the preparation requirement involves broad reshaping and aggregation of the underlying dataset. The transformations should occur at the Data 360 data-preparation layer, after which the output object can be used by downstream analytics.

Batch transforms can write to supported DLO or DMO output nodes depending on the source and design, providing a persistent transformed object that semantic modeling and other Data 360 processes can consume.

References/Topics: Data Setup - > Data 360 - > Batch Data Transforms - > Aggregate, Join, Transform, and Output Nodes.

===========================================================================