Where does Tableau’s 4th wave take us?

At the recent Conference, Tableau presented a future vision for the “4th wave on analytics”. In the keynote, it felt like the announcements fell a bit flat and rewatching the keynote confirmed this impression. I think the main issue on the day was that many of these announcements are aimed at a different audience to what was in the room. Let me pick apart those four main areas that Tableau is investing in and put it into the context of the Salesforce mothership and how Salesforce is finally embracing Tableau and data in general!

Stating the obvious problems

Ryan Aytay, Southard Jones, and Padmashree Koneti opened the segment with four major challenges that everybody working with data faces:

1. The Data landscape is large and fragmented
2. Users don’t trust the data or insights
3. Insights are overlooked or ignored
4. You can’t reuse what you build

This does not come as a surprise to anybody. If you have worked with data for any length of time, you will have run into all these issues at least once. The irony is that every tool ever promised to address these issues over and over again, in fact Tableau has features to address all of them already (connecting to different data sources, data lineage, subscriptions and pulse metrics, data sources and pulse metrics – to name a few).

The solution: Salesforce

What follows, will not be a dunk on Salesforce. There were definitely changes (good and bad) at Tableau after the acquisition, but it happened and it’s cumbersome to imagine “what could have been.” What’s far more interesting to me is how Salesforce is actually following through on the acquisition. You do not spend $15.7bn to just add an AI component to a product!

The first point that will stand out to anybody who has worked with Salesforce before is that all new screens are based on the Salesforce UI. While it is a “new Tableau interface”, it is also the “old Salesforce interface”! A very obvious hint on where the journey will be going but clearly something that most people would probably not have picked up upon.

The second detail I noticed was that most announcements that followed exist today in one way or another in Salesforce already. That is not to say that there wasn’t any innovation or significant development involved, but from a Salesforce perspective, it can be reframed as supercharging their capabilities with a robust data platform (and I expect this to happen at Dreamforce in September) opposed to innovating Tableau.

Challenge 1: Large and fragmented data landscape

There has been more and more of a push for semantic layers for your data. Traditionally, that has been one of Tableau’s weaker spots, where you were able to create data models in the tool, however it had multiple limitations. These will be addressed with the new shared dimensions feature which enables users to replicate complex data models within Tableau. This allows developers to define loose relationships between tables that are then fully determined when a user actually interrogates the data. What they announced in the vision goes a step further, where different parts of the model can be certified and reused in other contexts in an easy to understand visual interface. It also introduced real-time data processing, which will allow completely new use cases that currently require hacked workarounds.

Their demo was on a marketing use case and they did mention “lake houses” as part of it, which heavily hints on a more extended and tighter integration with Salesforce Data Cloud. While the Data Cloud product and its positioning changed significantly over the last 18 months, this is one of the core promises: to gather customer data across platforms and unify it in one place to make it easily accessible. While already built on top of the core Salesforce platform and utilising Tableau’s Hyper technology, this announcement shows the Tableau interface reimagined in the Salesforce UI.

Challenge 2: Lack of trust in data

Tableau’s solution here is also something that is – in parts – already there. A way to easily ingest and cleanse data to combine with governed enterprise data models. Tableau Prep can do most of these steps already and it actually does suggest the next steps to you for a long time. The big change here is that it moves away from a technical “flow” interface to a less “threatening” interface for business users. The system suggests changes to cleanse the data and a user just needs to accept those suggestions.

What is new is the fact that it also has access to all other models in the semantic layer and suggests how your new data can link into those models.

While this in itself sounds very interesting, it will be a no-brainer for Salesforce customers. Salesforce itself is already an enterprise data model, so this now allows a Tableau user to just hook into the relationships that are already defined with little to no additional work. These days I have to define all the relationships myself, potentially over and over again for different data sources. In the future, Tableau will read it out of Salesforce and allow me to go on with my day.

Challenge 3: Insights are overlooked or ignored

It is no secret that Salesforce currently has two competing products (of course Salesforce would not put it this way): Tableau and CRMA.

Whenever the discussion came to which tool to use, there were two big arguments for CRMA:

1. It respects the native Salesforce permission structure
2. It allows you to take an action (eg. add a note to an opportunity) from the dashboard

Tableau/Salesforce clearly addresses the second argument with this announcement, where you can take an action (ie. kick off a flow) from a metric or a dashboard. Again, capabilities that are already native in Salesforce but will be available to Tableau as well.

And while nobody explicitly mentioned the permissions at the event, my best guess would be that it will come soon, considering how close Tableau is moving to the core Salesforce platform.

Challenge 4: You can’t reuse what you build

Accepting the risk to oversimplify, this is something that Salesforce already had great success with. Their AppExchange has thousands of extensions for Salesforce and entire companies were founded on an app idea for a feature that Salesforce themselves wouldn’t or couldn’t support. When it comes to data, it is already something that organisations like Snowflake and data.world are very successful in. Opening this up to the customer base of Tableau will allow companies to monetise their data, models and analysis more easily (something that many companies are interested in but usually lack the infrastructure to do) and for more products around the Tableau and Salesforce ecosystem to be developed.

It will also allow for internal sharing of assets, which has the potential to have a huge impact on governance. Internal subject matter experts can develop charts with complex business logic behind them which can then be embedded in different dashboards rather than having multiple people across an organisation create their own version of the truth. (Something that was actually my main point in this year’s TC presentation)

Tableau themselves made several attempts in this direction, with Web Data Connectors, the Tableau Exchange and its accelerators and now Viz Extensions. But this time it seems like a more thought through and comprehensive approach.

What does that all mean?

TL;DR: Salesforce finally capitalises on their acquisition and makes Tableau an integral part of the platform!

In a bit more detail, they do what I expected them to do a few years ago already: to bring the platforms closer together and leverage each other’s capabilities.

For Salesforce customers, I believe this will be a huge advantage. Historically, Salesforce always pushed the Customer 360 promise but most organisations I worked with struggled to actually lean into it and unlock the value out of the data they already had.

For Tableau customers without Salesforce, I am not sure about the impact and I think the reaction at the keynote is the best indicator for this. I can see the value of all of the features and I do believe that they are useful to a ton of organisations; for data people however, it is a bit more complicated. Most of us are acutely aware that all of these problems are not tool problems per se, they are skill and culture problems more than anything. Rolling out a tool like Tableau across the organisation does not magically solve all your problems if the new users are not trained and the culture doesn’t allow them to question decisions based on data.

Looking a bit more into the future, these announcements have some interesting implications:

1. Tableau and Salesforce become harder to separate. While writing this post I already struggled to decide in certain sentences whether to use “Tableau” or “Salesforce” as the platform name. I’m sure the brand will stick around but we might see a convergence of Tableau Cloud and Salesforce, technologically as well as in the UI. Taking a stab in the dark, I wouldn’t be surprised to see a toggle between the “traditional” and “new” interface in Tableau Cloud.
2. Tableau cannibalises CRMA more and more. Everybody I talk to at Salesforce vehemently states that they are committed to CRMA. About 3 years ago I wrote a blog post for a previous employer predicting that Tableau will replace CRMA (Tableau CRM at the time) in the next ~24 months. It took a bit longer but I am fully convinced that Salesforce bet on Tableau and while they will not retire CRMA due to customer investment, future innovations will be delivered on Tableau.
3. Ryan already announced that this “Vision” will be delivered at Dreamforce in September, which makes it more of a teaser than a vision but that’s semantics. Comparing the Salesforce and Tableau ecosystem, Salesforce (ie. the business side) is where the money is. If you compare attendance and spend at regional Salesforce and Tableau events and the average spend on Salesforce and Tableau implementations, you will know what I mean. Going forward, I think we will see more features that speak directly to the business and operations side of organisations opposed to the technology side.
4. This tighter coupling to the operational side also makes sense from a sales point of view. For an organisation it is much easier to justify additional investment in a platform that can automate business processes and maybe even monetise my assets than in a platform that lets me “just” see and understand data.

On a higher level

There are two main parts of this vision, and Salesforce’s vision in general, that I am still mulling over.

GenAI

I am fully aware of all the developments in the last 18 months and do realise the implications these developments could have for everything. I also understand that GenAI is what investors expect because it is “the next big thing!”. And while I generally agree that it will be important going forward, I think there are a ton of barriers for its adoption, especially in enterprise use cases.

Image and text generators have inherent licensing issues that are and will be addressed in court, both inherit biases from the source material which make automatic publishing any GenAI content quite risky and generating code or workflows quickly in a way that works but nobody understands goes against any development paradigm I am aware of. And we do see very large organisations that have very strict policies around using GenAI even in a supervised scenario (Please rewrite my email).

If, when, or how these issues will be solved is anybody’s guess. For niche and fairly controllable use cases (e.g. We train a bot on our knowledge base to improve customer support) I can see it work well, yet for large scale automation and development, I think we might see the adoption raise a bit more before the Trough of Disillusionment will make us rethink some of our methods. (I am sure these paragraphs will age like fine wine!)

Business Process Automation

Salesforce as a CRM is much more than “managing customers and opportunities”. That is obvious to everybody within the ecosystem but I think it is not for people who don’t work with Salesforce. It allows organisations to replicate any kind of business process and manage it.

Initially it has been mostly in platform but especially with the acquisitions of Mulesoft and Slack, it allows for automation across platforms. In general that sounds great and many of us will have seen the demos of a data driven alert being pushed to Slack, where somebody tags the responsible person and they take an action right in Slack to resolve the problem.

The reality though is – again – much more complex. Integration between systems is a huge hurdle for organisations due to the lack of strategy, skill, funds, and people. Anecdotally, few organisations seem to make use of the capabilities that are already there, my best guess for the reason is that they cannot justify the effort for integration just to shave off a few minutes here and there. This seems like a feature that could save millions of dollars for a few organisations that have the right use cases but is not worth the effort for the majority of the customer base for the foreseeable future.

What’s next?

We will see in September what exactly they are releasing and how it fits in with the vision. In the meantime, I am interested to understand what others got from this vision. Do you work with Salesforce and agree with my summary? Do you not know Salesforce and think my summary doesn’t make sense? Or did it help you to put some of the pieces into place?