34 Tableau Server and Sharing Strategies

34.1 Why Tableau Server and Cloud Matter

A dashboard nobody else can reach is a dashboard nobody else uses.

Tableau Desktop is where the analyst builds. Tableau Server (on-premise) and Tableau Cloud (Tableau-hosted SaaS) are where the audience consumes. Without a sharing platform, every dashboard sits on the analyst’s laptop, accessed only by file copies and screenshots — which is exactly how dashboards die.

The standard practitioner references on the deployment side are Tableau Your Data! by Daniel G. Murray (2013), which devotes substantial coverage to Server administration, and Practical Tableau by Ryan Sleeper (2018), which covers the publishing-and-sharing patterns most analysts use day to day.

For a visualisation-focused book, this chapter is where the dashboards built in the previous chapters become shared institutional artefacts: governed, refreshed automatically, accessible from any device, and visible to the right audience under the right permissions.

34.2 The Tableau Deployment Landscape

flowchart TD
    T["Tableau<br>Deployment"]
    T --> D["Tableau Desktop<br>Authoring on<br>analyst workstations"]
    T --> S["Tableau Server<br>On-premise platform<br>for enterprise hosting"]
    T --> C["Tableau Cloud<br>SaaS hosted by Tableau<br>(formerly Tableau Online)"]
    T --> P["Tableau Public<br>Free public hosting<br>for shareable workbooks"]
    T --> M["Tableau Mobile<br>iOS and Android<br>consumption apps"]
    T --> Pu["Tableau Pulse<br>AI-driven metric<br>monitoring and alerts"]
    style T fill:#e3f2fd,stroke:#1976D2
    style D fill:#fce4ec,stroke:#AD1457
    style S fill:#fff3e0,stroke:#EF6C00
    style C fill:#fff8e1,stroke:#F9A825
    style P fill:#e8f5e9,stroke:#388E3C
    style M fill:#ede7f6,stroke:#4527A0
    style Pu fill:#f3e5f5,stroke:#6A1B9A

Server versus Cloud at a Glance

Property	Tableau Server	Tableau Cloud
Hosting	On-premise on your own infrastructure	Tableau-hosted SaaS
Maintenance	Customer manages OS, hardware, upgrades	Tableau manages everything
Customisation	Deeper — direct OS access, custom auth	More limited — managed within Tableau’s controls
Scaling	Customer adds nodes and capacity	Automatic, paid by user-tier
Compliance	Data never leaves the customer’s network	Data resides in Tableau’s chosen region
Best for	Large enterprises with strict data-residency or customisation needs	Most other organisations, especially those preferring not to manage infrastructure

For most new deployments, Tableau Cloud is the pragmatic choice; Server is the fallback for organisations with compliance or customisation requirements that Cloud cannot satisfy.

34.3 Architecture: Sites, Projects, Workbooks, Data Sources

A Tableau Server or Cloud deployment is organised in a four-level hierarchy. Understanding this hierarchy is essential because permissions, sharing, and governance all hang off it.

The Four-Level Content Hierarchy

Level	Purpose	Example
Site	Tenant boundary; users in one site cannot see another’s content	A separate site per business unit, or a single corporate site
Project	Folder within a site, used to group related content	Marketing, Finance, Operations
Workbook	Published Tableau workbook with sheets and dashboards	Yuvijen Sales Q1 Review
Data Source	Published data source reused across workbooks	Customer 360 (Certified)

Within a project, additional content types include Flows (Tableau Prep flows), Custom Views (saved filter states), Subscriptions, and Alerts. Permissions can be set at any level and inherit downward unless overridden.

34.4 The Publishing Workflow

Publishing a workbook from Desktop to Server / Cloud follows a consistent six-step path:

In Desktop: open the workbook → Server → Publish Workbook.
Sign in to the Server / Cloud instance with your credentials.
Select site and project.
Configure: name, description, tags, project, permissions, data source handling (embed vs published), data refresh schedule (if extract).
Publish.
Open in browser: the published workbook appears immediately at a URL the audience can be given.

For the data source, the analyst chooses between embedding (the data travels with the workbook) or pointing to a separately-published certified data source (preferred for governance). The certified-data-source pattern was introduced in Chapter 30 — most mature organisations point every workbook at certified sources.

34.5 Permissions and Site Roles

Tableau’s permission model has two axes: Site Roles (what a user can do across the platform) and Permission Rules (what they can do for a specific item).

34.5.1 Site Roles

Site Roles

Site Role	Capability
Server / Site Administrator	Full administrative access
Creator	Connect to data, build new workbooks, publish, edit, manage
Explorer	Interact with published content, create web-edit workbooks
Explorer (can publish)	Explorer plus the ability to publish workbooks
Viewer	Read-only access to published content
Unlicensed	No access until a role is assigned

The Creator licence is the most expensive; Viewer the cheapest. Most large deployments have a small number of Creators, a moderate number of Explorers, and a large number of Viewers.

34.5.2 Permission Rules

Within a project, workbook, or data source, permissions are set as a list of capabilities granted or denied to a specific user or group. Capabilities include:

View — see the item.
Filter — apply filters and customise within the published view.
Download Image / Data / Workbook — export to PNG, CSV, or .twbx.
Web Edit — modify in browser without Desktop.
Save — overwrite the published version.
Add Comments, Set Permissions, Delete, etc.

A common pattern: at the project level, grant View, Filter, Download Image to a Business Users group; grant the full set to a Content Owners group. Inheritance pushes the rules to all workbooks and data sources within the project, with overrides for exceptions.

34.5.3 Row-Level Security

Row-Level Security (RLS) restricts which rows of data each user sees, not which workbooks. The pattern: a user-to-data mapping table joins to the underlying data; a calculated field uses USERNAME() to filter rows where the user is allowed.

Example calculation:

[Region] = LOWER(SPLIT(USERNAME(), '@', 1))

In a workbook published with this filter as a Data Source Filter, each user sees only their region’s data automatically.

34.7 Refresh Schedules and Tableau Bridge

Published extracts and data sources need to refresh on a schedule. On Server / Cloud:

Schedules are managed by administrators; refresh windows can be daily, hourly, or sub-hourly.
Refresh tasks are queued; failures notify the data-source owner.
Incremental refresh appends new rows since the last refresh, useful for very large extracts.

For Cloud customers whose data sits on-premise behind a firewall, Tableau Bridge is the connector: a small Bridge client runs inside the customer’s network, refreshes the on-premise data on a schedule, and pushes the result to Cloud. Bridge is the standard mechanism for hybrid deployments.

34.8 Tableau Pulse

Tableau Pulse is a 2024-introduced metric-monitoring layer above Server / Cloud. The user subscribes to metrics (daily revenue, order count, NPS) rather than to dashboards, and Pulse delivers:

A natural-language daily summary of the metric’s behaviour.
Automated drivers analysis (revenue is up because the West region grew 8 %).
Threshold alerts when the metric breaks a predefined band.
A unified mobile and email feed of the user’s tracked metrics.

Pulse turns subscriptions into proactive, AI-augmented monitoring — useful for executives who want the headline without opening dashboards.

34.9 Governance and Content Management

A mature Tableau deployment has explicit governance:

Certified Data Sources: A Tableau-Server feature that marks a data source as reviewed and trusted; certified sources show a small badge in the catalogue.
Content Catalogue: Searchable inventory of all workbooks, data sources, flows, and metrics; supports tagging and ownership.
Lineage View: Tableau Catalog shows the upstream data flow from source to dashboard, so consumers and stewards can trace any number to its origin.
Stale Content Cleanup: Server / Cloud admin can find unused workbooks (last viewed > 90 days ago) and archive them.
Project Hierarchies: Projects can be nested up to 10 levels for large enterprises.

Without these practices, the deployment becomes a content swamp within a year — analogous to the data-lake-becomes-data-swamp problem from Chapter 25.

34.10 Performance and Monitoring

Server / Cloud provides administrative views for monitoring:

Background Tasks: Status of extract refreshes, subscription delivery, and data-quality runs.
Workbook Performance: Slowest views, most-viewed views, average load times.
User Activity: Active users, sessions, content consumed.
Postgres Repository: For Server, direct access to the underlying metadata for custom dashboards.

A common Server/Cloud admin task is identifying slow workbooks and either refactoring them, switching them to Extract, or moving them to a faster data source. Performance Recording (covered in Chapter 33) runs at the workbook level; the admin views run at the deployment level.

34.11 Best Practices

One certified data source per business entity: Customer, product, region, employee — each has one trusted source that all workbooks consume.
Project hierarchy by audience: Projects organised by who uses the content (Executive, Marketing, Operations), not by who built it.
Standard naming convention: Workbooks named {Domain} - {Topic} - {Owner} for findability.
Group-based permissions: Permissions assigned to groups, not individuals; group membership maintained via SSO / Active Directory.
Refresh schedules off-peak: Heavy extract refreshes overnight; lightweight ones during the day.
Tag content liberally: Tags drive findability in the catalogue.
Train end users: Explorers and Viewers benefit from a one-hour orientation more than from a thicker permission scheme.
Audit quarterly: Review permissions, content usage, refresh failures.
Use Pulse for executives: A daily Pulse digest is more effective than a daily-emailed dashboard PDF.

34.12 Common Pitfalls

Embedded Data Source Sprawl: Every workbook with its own embedded extract; no single trusted source. The mature pattern is one published certified source per entity.
Permission Per User: Permissions assigned to individual users instead of groups; impossible to maintain at scale.
Stale Content Hoarding: Years of orphaned workbooks; nobody trusts the catalogue.
No Refresh Failure Alerts: Refresh fails overnight; the dashboard silently shows yesterday’s data; no one notices.
Open Project as Default: Default permissions allow everyone to publish anywhere; the catalogue becomes a write-anywhere mess.
Site as Project: Using sites where projects would suffice; sites are tenant boundaries and should be used sparingly.
No Version Control: Workbooks overwritten in place with no history; a bad change cannot be rolled back.
Mixing Certified and Non-Certified: A dashboard joining a certified source with an analyst’s local Excel file produces the worst of both worlds — the dashboard is non-reproducible.
No Pulse / Subscription Strategy: Executives still pulled daily-emailed PDF dashboards in 2026 when a Pulse digest would do the job better.

34.13 Illustrative Cases

A Bank’s Project Hierarchy by Function

A retail bank organises its Tableau Server with one site, projects for Retail Banking, Risk, Treasury, and Operations, and sub-projects within each for Certified Data Sources and Workbooks. Permissions are inherited from the project level and assigned to AD groups. Onboarding a new analyst takes a single AD group change.

A Retailer’s Pulse-Led Executive Briefing

A retailer’s CEO no longer opens dashboards on most days. Tableau Pulse delivers a daily metric digest at 7 a.m. covering five tracked KPIs with natural-language summaries. The CEO clicks through to the dashboard only when a Pulse alert flags an anomaly.

A Manufacturing Group’s Bridge Deployment

A manufacturing group’s operational data lives on-premise but the firm has chosen Tableau Cloud for its analytics platform. Tableau Bridge runs in the corporate network, refreshes plant-floor extracts every 30 minutes, and pushes the data to Cloud. Plant managers consume the dashboards on their phones via Tableau Mobile. The hybrid pattern works without re-platforming the operational systems.

34.14 Hands-On Exercise: Publishing and Sharing a Tableau Workbook

Aim: Publish a Tableau workbook to Tableau Cloud (or Server), configure permissions, set up a refresh schedule, create a subscription, and configure a data-driven alert.

Scenario: Take the interactive sales dashboard built in Chapter 33 (or any working .twb file) and deploy it for the audience at Yuvijen Stores.

Deliverable: A published workbook, a configured subscription, a data-driven alert, and a one-page sharing reference.

34.14.1 Step 1 — Publish the Workbook

In Tableau Desktop, open the workbook.
Server → Publish Workbook.
Sign in to Tableau Cloud (or Server). The Publish dialog opens.
Set Project (e.g., Sales Analytics).
Set Name and Description. Add Tags (monthly-review, executive).
Permissions: choose Locked to Project for governance simplicity.
Data Sources: prefer Use Published Data Source if a certified source exists; otherwise embed.
Click Publish. Tableau opens the workbook in a browser at its new URL.

34.14.2 Step 2 — Configure a Refresh Schedule

In the browser, navigate to the workbook’s data source.
Click Schedule a Refresh (or the calendar icon).
Choose a schedule (Daily, 6:00 AM IST, for example).
Save.

If using Tableau Bridge for an on-premise data source, ensure Bridge is running on a server inside the corporate network with sufficient capacity for the refresh window.

34.14.3 Step 3 — Set Permissions

From the workbook page → Permissions.
Add the Sales Team group with capabilities: View, Filter, Download Image.
Add the Sales Leadership group with capabilities: View, Filter, Download Image, Download Workbook.
Remove default All Users permission.
Save.

The workbook now appears only to members of the two groups, with appropriate capabilities.

34.14.4 Step 4 — Create a Subscription

Open the workbook in the browser.
Click Subscribe (the bell icon).
Choose Subject: Yuvijen Sales — Daily Snapshot.
Frequency: Weekday 8:00 AM IST.
View: Select the Executive Dashboard page only.
Format: PNG image (or PDF for a wider audience).
Save.

Subscribers receive an email with the dashboard image at the chosen time. Multiple subscribers can be added by repeating the subscription for each.

34.14.5 Step 5 — Configure a Data-Driven Alert

Open the dashboard. Select the visual containing the NPS KPI.
Click the Alert icon (bell with arrow).
Configure: NPS goes below 40 with frequency Daily.
Add recipients (yourself initially).
Save.

When tomorrow’s refresh produces an NPS below 40, the alert email fires automatically.

34.14.7 Step 7 — Configure Tableau Pulse (Optional)

From the Tableau Cloud home, navigate to Pulse.
+ New Metric: choose the data source, select the measure (Sales) and time dimension (Order Date).
Configure the comparison period (week-over-week, month-over-month).
Add follower groups.
Pulse begins delivering daily summaries to followers.

Pulse is the modern subscription replacement; for executives, it is often more effective than a static emailed dashboard.

34.14.8 Step 8 — Document the Sharing Plan

The Sharing Reference Sheet

Audience	Access Method	Cadence
Sales Team (50 users)	Live Cloud URL	On-demand
Sales Leadership (5 users)	Live URL + Daily Subscription	Daily 8:00 AM IST
CEO	Pulse digest	Daily 7:00 AM IST
Marketing partners (external)	Embedded view in partner portal	On-demand
Audit and compliance	Read-only with audit logs enabled	Quarterly review

The reference sheet documents who sees what, how, and how often — the operational form of the sharing strategy.

34.14.9 Step 9 — Connect to the Visualisation Layer

The hands-on closes the loop on the Tableau journey across Chapters 30–34:

The workbook built in Desktop (Ch 30) becomes the published artefact in Server / Cloud.
The calculations and parameters (Ch 31) and advanced charts (Ch 32) and interactive dashboard (Ch 33) all travel with the publish step.
Permissions and subscriptions decide who sees what and when.
Pulse and alerts make the dashboard push to the audience rather than waiting to be pulled.

A Tableau workbook that was never published reaches one user. The same workbook published, scheduled, and subscribed to reaches an entire organisation — and is the operational difference between a chart and a system.

Files and Screen Recordings

Tableau workbook (yuvijen-published-dashboard.twbx), screenshots of the Cloud / Server publish flow, permission setup, refresh schedule, subscription, alert, and Pulse configuration will be embedded here.

Summary

Concept	Description
Foundations
Why Server and Cloud Matter	A dashboard nobody else can reach is a dashboard nobody else uses
Tableau Deployment Family
Tableau Desktop	Authoring environment on analyst workstations
Tableau Server	On-premise platform for enterprise hosting
Tableau Cloud	SaaS hosted by Tableau, formerly Tableau Online
Tableau Public	Free public hosting for shareable workbooks
Tableau Mobile	iOS and Android consumption apps
Tableau Pulse	AI-driven metric monitoring with natural-language summaries and alerts
Content Hierarchy
Site	Tenant boundary; users in one site cannot see another's content
Project	Folder within a site used to group related content
Workbook	Published Tableau workbook with sheets and dashboards
Data Source	Published data source reused across workbooks
Flow	Published Tableau Prep flow
Custom View	Saved filter state on top of a published view
Publishing Workflow
Publishing Workflow	Six-step path from Desktop publish to a workbook URL
Embedded Data Source	Data source travels with the workbook; less governance
Published Data Source	Data source published separately and reused; preferred for governance
Site Roles
Server Site Administrator	Full administrative access across the platform
Creator	Connect to data, build new workbooks, publish, edit, manage
Explorer	Interact with published content, create web-edit workbooks
Explorer Can Publish	Explorer plus the ability to publish workbooks
Viewer	Read-only access to published content
Unlicensed	No access until a role is assigned
Permission Capabilities
Permission Capability View	Permission to see the item
Filter	Permission to apply filters and customise within the published view
Download	Permission to export to PNG, CSV, or twbx
Web Edit	Permission to modify in browser without Desktop
Save	Permission to overwrite the published version
Row-Level Security
Row-Level Security	Restricts which rows of data each user sees, using USERNAME() in a filter
Sharing Mechanisms
URL Sharing	Stable URL for any published view; the simplest share
Embedded View	Iframe embed of a published view in SharePoint, Confluence, or any HTML page
Subscription	User subscribes; Tableau emails a snapshot on a schedule
Data-Driven Alert	User sets a threshold; Tableau emails when the threshold is crossed
Slack Integration	Tableau Cloud posts notifications and snapshots to Slack channels
Public Sharing	Tableau Public hosts workbooks at no cost on the public internet
Pulse Metric Monitoring	Daily AI-driven natural-language metric digests with anomaly alerts
Refresh and Bridge
Refresh Schedule	Schedule for refreshing extracts and data sources
Incremental Refresh	Appends new rows since the last refresh; useful for very large extracts
Tableau Bridge	Connector running on-premise that refreshes data and pushes to Cloud
Governance and Catalogue
Certified Data Source	Server-side feature marking a data source as reviewed and trusted
Content Catalogue	Searchable inventory of workbooks, data sources, flows, and metrics
Lineage View	Tableau Catalog showing upstream data flow from source to dashboard
Stale Content Cleanup	Find unused workbooks last viewed beyond a threshold and archive them
Project Hierarchies	Projects can be nested up to ten levels for large enterprises
Performance and Monitoring
Background Tasks View	Status of extract refreshes, subscriptions, and data-quality runs
Workbook Performance View	Slowest views, most-viewed views, average load times
User Activity View	Active users, sessions, content consumed
Postgres Repository	Direct access to underlying metadata for custom admin dashboards
Best Practices
One Certified Source per Entity	Customer, product, region, employee each have one trusted source consumed by all workbooks
Project by Audience	Projects organised by who uses the content rather than by who built it
Standard Naming Convention	Workbooks named domain dash topic dash owner for findability
Group-Based Permissions	Permissions assigned to groups maintained via SSO and Active Directory
Refresh Off-Peak	Heavy extract refreshes overnight, lightweight ones during the day
Tag Content	Tag content liberally to drive findability in the catalogue
Train End Users	Explorers and Viewers benefit from one-hour orientation more than thicker permission schemes
Audit Quarterly	Review permissions, content usage, refresh failures every quarter
Use Pulse for Executives	Daily Pulse digest is more effective than a daily-emailed PDF for executives
Common Pitfalls
Embedded Source Sprawl	Pitfall of every workbook with its own embedded extract and no single trusted source
Per-User Permissions	Pitfall of permissions assigned to individual users instead of groups
Stale Content Hoarding	Pitfall of years of orphaned workbooks that erode catalogue trust
No Refresh Failure Alerts	Pitfall of refresh failing silently overnight while the dashboard shows stale data
Open Project as Default	Pitfall of default permissions allowing everyone to publish anywhere
Site as Project	Pitfall of using sites as projects when sites are tenant boundaries
No Version Control	Pitfall of overwriting workbooks with no history; bad changes cannot be rolled back
Mixing Certified and Non-Certified	Pitfall of joining a certified source with an analyst's local file producing non-reproducible results
No Pulse Subscription Strategy	Pitfall of executives pulled to PDFs when Pulse digests would do the job better