The Tableau Sharing and Collaboration Platform: Deployment Models in Higher Education

Posted On August 18, 2020December 28, 2020

Tableau’s platform of products is intentionally flexible and scalable. Unsurprisingly, Tableau deployment models range from a narrowly focused, single use-case with a handful of users to an extremely complex ecosystem of thousands of people, functions, and processes across multiple campuses. Across the spectrum of models, the theme of sharing and collaboration around insights, data, and content is consistent. 

As education sector Solutions Engineers, we encounter Tableau deployment examples across this spectrum. We also see how institutions that increase their adoption of Tableau expand from narrow use-cases to broad applications, often in a measured, step-by-step manner. 

As they develop through a maturity cycle, a common question for colleges and universities is, “What are other institutions like us doing?”. Usually, the deeper root of the question is along the lines of, “How do our institutional peers structure their Tableau platforms to facilitate effective sharing and collaboration of resources?”

Fortunately, within the higher education Tableau community, institutions are not only willing but excited to share their Tableau stories and knowledge. Connecting customers together is an enjoyable part of our role as Solution Engineers. The one downside is that the scope of such knowledge sharing is limited to one-to-one interactions. 

Introducing the Four Tableau Platform Deployment Models in Higher Education

To address this, I thought it would be helpful to develop and introduce a schema based on the most common Tableau deployment models in the higher education sector. Four conceptual models of increasing complexity are presented. Each model highlights how sharing and collaboration (both in processes and features) are supported among an institution’s stakeholders. To make sure readers are taking full advantage of the sharing and collaboration features available to them in Tableau Server, I’ve included a list of the top features highlighted in the examples in the following section.

Quick Preface and Review of Terms

A few preliminary notes for context: first, the conceptual models presented certainly do not cover all possible or even successful Tableau environments. The models are intended only as a starting guide. Second, the underlying data sources and hardware architecture associated with the described deployments is outside of the scope of this post; these topics are deserving of their own in-depth examinations. Third, while only a subset of Tableau’s product offering is included in the four models, a basic knowledge of Tableau products and jargon is needed. Here are the key definitions to be aware of:

And finally, Tableau is on a quarterly release cadence. The products and sharing/collaboration features presented are current as of the 2020.3 release. Future version releases may look different or include changes. With that, let’s get into the models. 

Model 1: Legacy Starter for a Department

Across the spectrum of use-case and administrative complexity, we can broadly bucket Tableau deployments at higher education institutions into four models. The models are not strictly progressive, i.e., Model 2 leads to Model 3 and then Model 4. It is common to see two more models operating concurrently at an institution. Sometimes such deployments merge together to take advantage of economies of scale. Sometimes they do not for various reasons. Models 1 and 2 represent department scoped use-cases often with only a handful of licensed users.

Model 1 represents a legacy, department-starter approach that does not utilize the functionality of Tableau Server. This is an outdated model, as the stand-alone Tableau Desktop Professional license is no longer available for new customers. The Desktop Professional license was replaced by the Creator license back in 2018. A Creator license includes Tableau Desktop, Tableau Prep Builder, and a seat on Tableau Server (or Tableau Online) – three pieces of software in one. 

The hub of Model 1 is the Tableau Desktop product, or perhaps more accurately, the analyst who uses Tableau Desktop. For a team of two collaborating Tableau Desktop users, workbooks and data sources are saved in a shared drive or folder. Version control can quickly become a challenge without diligent organization, even for a team of two. 

For those without a Tableau Desktop license, the sharing options are limited. Content can be viewed via Tableau Public, but anything published there must be understood as accessible to the world; there is little to no governance over content published to Tableau Public. 

The second option is to export static files from Tableau Dashboards. Images, pdfs, and ppts are common examples. There are times where packaged workbooks, .twbx files, are shared with non-Tableau Desktop users. Packaged workbooks with a static snapshot of data included may be opened using Tableau Reader. This approach might seem sensible at first, but in practice often adds more headaches and frustration than the perceived dollar savings. Security concerns aside, regularly ensuring workbook-Tableau Reader version compatibility and guiding a casual user through the installation and workbook loading process is not without cost. This can be significant at scale. 

Model 2: Modern Starter Deployment for a Department

Model 2, also for the department, shows how modern Tableau licensing is deployed. The fundamental difference between Models 1 and 2 is that Tableau Server now serves as the hub of the Tableau Deployment, rather than Tableau Desktop. This change opens up new sharing and collaboration features and reduces the risk of user bottlenecks. It supports scaling to Explorer licensed users: a far more secure way of sharing Tableau content than relying on Reader.

An Explorer’s Web Authoring functionality supporting self-service within the governed Tableau Server environment is one significant example. In Model 1, ad hoc requests for dashboard changes require the time and attention of a Tableau Desktop user, even for something very small. A properly permissioned Explorer self-serves such changes and might even develop new dashboards all within the browser interface of Tableau Server: no software downloaded needed.

A further advantage of Model 2 sharing is the ability to use Tableau Mobile to access and interact with visualizations.

Model 3: Tableau Deployment for a Division or College

Model 3 expands from the scope of a single department to a division with multiple departments. Scaling to more users is apparent in both the additional Creators in departments as well as more Explorers. We also introduce the Viewer license type, which effectively supports roles where strict consumption of dashboard content, rather than authoring or editing, is the primary need.  

Model 4: Tableau Deployment for the Institution or Enterprise

Model 4 depicts an enterprise or institution-wide Tableau deployment. Along with the multiplication of divisions, we also see the partitioning of use-cases into internal and external-facing content. This partitioning is a far more scalable solution than relying on Tableau Public for external content, but comes with increased administrative complexity and support costs. 

The Four Models in Context: Higher Education Use-case Examples

The product-centric model diagrams are helpful for readers with more familiarity with Tableau products: perhaps less so to anyone newer to the platform. To provide better intuition, I’ll now apply use-case examples from the higher education context to illustrate the interactions of model components. Some of the common challenges experienced in each model are also highlighted as a way to assist institutions considering a new model better prepare. I’ll start with Model 1. 

Model 1: Institution Research Factbook

The Model 1 scenario is very common in higher education institutional research departments. Once a year, the institution’s interactive factbook must be updated with the prior year’s data. The updated dashboard is embedded in the IR’s web page, so that anyone with an internet connection can view the statistics for the institution.

Here are the steps depicted above:

(1) Our IR analyst assigned to this task must first search through the shared drive of last year’s data and saved workbooks. Even with careful documentation and organization, it still takes time to confirm the right files are identified. 

(2) Next, the analyst updates the Tableau workbook, replacing the data source and modifying some of the visuals based on feedback. (3) After some back and forth reviewing and approving, the new factbook is republished to the institution’s Tableau Public site, which can be used for embedding in a webpage. A copy of the workbook is also saved to the shared drive for future reference.

(4) Last, a series of static images of some of the workbook’s dashboards is exported. These snapshots are pasted into a Word document with additional commentary in preparation for the upcoming board of trustees meeting.

While this example is for a single department and very narrow use-case it has limitations. First, the time spent tracking down the most up-to-date version of a workbook and data can be significant. Second, as people view the workbook on Tableau Public, responding to ad hoc requests and changes can become unmanageable for the analyst. Third, the data included in the images is static. If data changes the Word document’s contents becomes inaccurate and there is no simple way to deliver up-to-date content. The example for Model 2 highlights how some of these limitations are overcome.

Model 2: Registrar’s Office Classroom Assignment

In this use-case example, the Registrar’s Office is responding to reduced classroom capacity implications due to COVID-19 social distancing requirements. Classroom assignments for courses must be re-evaluated and the information efficiently distributed to the various faculty department contacts.

(1) The registrar staff member assigned to perform the updates logs into Tableau Server and finds the most current workbook and data sources published in the Data Server. She opens them in Tableau Desktop.

(2) After analyzing and making the needed updates in Tableau Desktop, the staff member publishes the new workbook back to Tableau Server for governed, internal access by others. 

(3) For external users, she publishes one of the dashboards that has a redacted view of course format and assignment information to the Registrar’s Tableau Public profile. Like in Model 1, this can be used for embedding in a web page.

(4) To notify the various faculty department contacts of the new information, the staff member shares the link to her dashboard. Because each of the department contacts is an Explorer user in Tableau Server they can make modifications to the updated workbook. This allows them to focus only on their discipline area. The save their own modified workbook to Tableau Server for future quick access. 

(5) Key enrollment KPIs from some dashboards are also saved as Metrics – a new content object in Tableau Server. The faculty department contacts keep an eye on the Metrics from their mobile device.

The new self-service capabilities of Tableau Server enhance the sharing and collaboration potential of Model 2. However, the addition of Tableau Server requires some administrative responsibilities, such as maintaining the environment and permissions configuration for data and content. For some institutions, Tableau Online provides an effective way to reduce such administrative burdens. This can be particularly helpful for less-technically inclined departments. 

While we now have a platform to support more user sharing and collaboration, the Explorer functionality may still be a bit too robust for the majority of faculty and administrators who strictly want to consume dashboards, not create them. 

Model 3: Student Affairs Division

The first two models focused primarily on a single department’s operations. Model 3 considers the new complexities and sharing potential that emerges when a division of multiple departments uses the Tableau Platform. In this use-case example, the student affairs division monitors and is responsive to the well-being of its students. Similar to Model 2, Tableau Server is the central hub for nearly all of the interactions among users.

(1) We start in the Student Health Services Office. Recently, they conducted a well-being survey for all students. Because the results are incredibly valuable for departments across the division, the Health Office publishes the data source to Tableau Server. They certify it to indicate its official status. The data is now securely available on-demand for other departments to perform analysis and build dashboards. 

(2) One of the departments that sees great value in the well-being data is the Residence Life Office. The director is concerned about providing effective programming tailored to the needs of different segments of campus. The analyst for the department connects to the certified data source and builds a set of well-being summary and program priority dashboards. These are published back to the Server. 

(3) Transparency is a critical value for the institution. Communicating the well-being sentiment of students is something the institution’s leadership is supportive of sharing. The Residence Life analyst publishes some of the aggregate results and programming dashboards to the institution’s Tableau Public site for embedding in content as needed. 

(4) Within Residence Life, the campus housing directors are also able to access and customize the unredacted programming and summary information for students in their areas. There are certainly differences of sentiment between both undergraduate and graduate students, so a one-size-fits-all approach is not going to be effective. After applying custom filtering, the housing directors subscribe their staff teams to the views on a weekly basis. The teams know where their supervisor’s attention is focused at their weekly staff meetings. 

(5) The Vice President of Student Affairs is stretched even further than usual this fall. While she recognizes the incredible value of the work that her team has put together in the data and dashboards, she has no time to explore the data or customize dashboards herself. She receives an emailed Alert only when certain well-being thresholds are exceeded to make appropriate data-informed decisions. Her needs are matched to the Viewer role, which is designed for the secure consumption of accurate, up-to-date information on-demand.

(6) Lastly, just in case the Vice President needs to brief the president’s cabinet on the real-time value of the metrics, with a single click she Favorites the student well-being summary dashboard. This will make it easy to pull up on her mobile phone in a meeting via a shortcut list. 

The Server administration challenges of Model 2 are amplified with the increasing scale of users in Model 3. Specialized knowledge of Tableau Server Administration, performance tuning, and SSO or user management functionality becomes key in larger division deployments, where there may be upwards of hundreds of users. The usage profiles and needs of the Tableau Server users also begin to segment at this scale. Consequently, ensuring analytics proficiency and fostering a community that supports other Tableau users becomes increasingly valuable. A range of education, from basic getting started to advanced functionality ensures the Creators, Explorers, and Viewers have access to the resources they need.

Model 4: Institution-wide Ecosystem

Model 4’s use-case combines the scenarios from the previous two models. It extends the external-facing functionality, supporting a greater variety of use-cases. The scale and complexity is more representative of an ecosystem of interdependent actions and shared resources, rather than a series of linear causes and effects.

(1) Like Model 3, all internal-facing use-cases are built around the Tableau Server hub. 

(2) As described in Model 3, the Student Affairs division relies on Tableau Server to generate their Student Well-Being data and dashboards, which are saved to the internal-facing Server.

(3) Because the Well-being Survey will be run on a weekly basis for the semester, the manual publishing to Tableau Public process is replaced by a second, stand-alone Tableau Server environment that provides robust and automated functionality. 

(4) The second external server environment provides anonymous or guest user functionality for those outside of the institution. Similar to Tableau Public, external visualizations can now be embedded in the institution’s webpages, but with dedicated resources and an institutionally managed infrastructure.

(5) The Registrar’s Office, which falls under the Business and Administrative Operations Division at our institution, also completes their course and classroom assignment use-case in this model. They also benefit from the external-facing Tableau Server environment, rather than relying on Tableau Public.

(6) Similarly, the Academic Affairs Division has its own set of priorities and dashboard needs. The division is able to utilize the secure internal-facing Server environment. 

(7) The governed access to Web authoring content by Explorers continues to function as before, just on a much larger and segmented scale using Project and Nested Project Folders.

(8) The story is similar for the visualization consumption focused Viewer users. Department heads and other institutional leaders have the read-only access and alerts that they need.

(9) Finally, all users in the deployment have secure access to the content they are permissioned to see and interact with via the Tableau Mobile app, providing flexibility in how and when data is accessed.

The same challenges of Model 3 hold true in Model 4, but with a second, stand-alone Tableau Server environment — the external-facing server. Due to the nature of an institution-wide or enterprise deployment, it becomes important to pay attention to performance metrics and understand the options to tune the architecture of an environment. Because you may be supporting many thousands of users, understanding how to address server load and when scaling to a multi-node environment requires more more advanced administrator knowledge. All of the infrastructure complexity correlates with cost, so budget planning considerations, both initial and long-term, are critical.

Review of Highlighted Sharing and Collaboration Features of the Tableau Platform

Throughout the scenarios, I’ve highlighted some of the top sharing and collaboration features in functional context. They are numerous, but valuable features that you likely already have access to if your institution’s Tableau environment is similar to Models 2 through 4. Below is a figure with a brief summary of each. You can find more details on each of them by visiting the Tableau product guide web pages.

Concluding Thoughts and Recommended Next Steps

In closing, I am hopeful this information serves as a useful reference guide to Tableau Platform discussions at higher education institutions. For those ready to dig deeper into their data and analytics strategy, I encourage you to spend time reading through the Tableau Blueprint resources. Specifically, the section entitled “Use Cases and Data Sources” allows you to take your institution’s needs and processes and consider how you might structure your analytics environment effectively. 

As I stated early in the post, I could not possibly cover all of the variations in Tableau deployments because of the flexibility and scalability that the platform offers. If your institution has found a unique structure that should be considered a Model 5, please leave a comment below or reach – I’d be happy to learn more from you. 

Dan Bradley

Dan Bradley is a Principal Solution Engineer for Tableau’s Higher Education Field Education Team. Based in Chicago, he works with higher education institutions in the Central and mid-Atlantic regions of the U.S. In addition to technology, Dan has a background in education administration, including an M.S. in Higher Education Administration and Policy. Dan's mission is to help the people of higher education become data-reflective practitioners who can see, understand, and act on their data. *Opinions are my own and not the views of my employer*