Details of Tableau’s integration with Salesforce

Francois Ajenstat, Chief Product Officer, Tableau

On 1 August 2019, Salesforce announced the completion of its $15.7bn acquisition of Tableau, bringing cloud-CRM “no-software” and the analytics giant together. With the recent release of Tableau 2021.1 in March, the two companies are exploring this newfound synergy and paving new roads towards creating concrete value for users. To find out more about the integration process and what lies ahead, we spoke to Francois Ajenstat, Tableau’s Chief Product Officer.

What has been the overall approach towards the integration of Salesforce and Tableau?

Although the deal closed in August 2019, we were still under regulatory review in the UK, and so we have not been able to integrate until the beginning of 2020. Firstly, our focus on innovation has not slowed down. We continue to deliver new capabilities to our customers, and continue the core promise of helping people see and understand data. At the same time, we are looking at Salesforce to find the opportunities to do more together. What are Salesforce’s capabilities, such as Einstein Discovery? What are Tableau’s capabilities? How do we deliver more to our customers together?

We are also looking at how to leverage innovations across Salesforce. For example, we have added connectivity to Datorama, and we’ve added new dashboard templates for all the clouds. This comes from thinking through how we can accelerate our mission of helping people see and understand data, and leveraging Salesforce innovations to help us deliver more value quickly to Salesforce and non-Salesforce customers.

Einstein Discovery is one of the first products to be integrated into Tableau. What made it the ideal AI plug-in for Tableau?

This is an area that we have been very passionate about. It is something that we were investing in, but being part of Salesforce has accelerated our ability to deliver by 3-5 years. We have a team that is dedicated to bringing powerful data science capability to non-technical people, and that aligns with Tableau’s mindset and philosophy of democratising data.

There are many AI engines out there such as Azure Machine Learning, Sagemaker, Tensorflow, and Tableau integrates with them by essentially taking data science and operationalising it with Tableau. However, you still need the business user to write their own models, and that is a gap that is unaddressed. Einstein Discovery has been doing this work for Salesforce users. It is a proven capability, and it has been running and delivering millions of models and predictions everyday. Why not take it over? It was the right team, the right technology, and the right mindset, and it has expanded our opportunities.

What was the process of integration like?

We started by getting the teams together and sharing each other’s innovations. What was unique? What was fun? Then, we started doing some hackathons to let the developers just have fun. Now that you have two toolsets, go and figure out what you can do, and the amount of innovation, ideas, and creativity that came out was very inspiring. From there, we started to have a lot of customer conversations, because we are focused on customer success and innovation at the end of the day. That was where this first innovation to bring Einstein Discovery and Tableau together started to blossom.

Customers told us three things. First, they said Tableau is a fantastic solution, but it looks at data that happened in the past. We would love to have more predictive or prescriptive capabilities, and that would be a game-changer. Second, they told us their organisations were doing a lot of data work, but there were limited skills, people, and talent. If we could accelerate the speed of decision-making and bring it to more people, it would help drive value to these organisations. Last, we heard that Einstein Discovery was extremely valuable inside the Salesforce workflow to help sales people or service agents be more productive and have insights within the point of consumption.

Our thought was to leverage this value and open it up, so that we can deliver value not only in the Salesforce workflow, but for any workflow. It was a win-win-win scenario, and I think it brings tons of value to Tableau and Salesforce customers, and opens up a new category for customers.

Could you elaborate on the use cases for data scientists?

There are a few patterns that we see with data scientists. In the first case, they run models in different tools, get a result set, then visualise that result set in Tableau. They use Tableau as the interface, then they go and refine their models, so there is a back-and-forth approach. The second approach is to use Tableau to see all the data and figure out the subset that they need to go and launch their models.

The third approach is where it gets really interesting. Instead of a data scientist being a model factory, they can start operationalising Tableau so that the user can interact with the data, and as they point and click, it recomputes the models dynamically. From the interaction paradigm, we can integrate with any kind of data science pattern in the background. It works great if it is fast and interactive, and if it requires a lot of compute and time, we can integrate it into the data prep pipeline so that they can enrich data models with classification, risk scoring, and all the other different factors.

Can you describe what are some of the new things a typical Tableau user will see when they open it up? What new insights can they gain from the new capabilities?

We have integrated Discovery into Tableau in three ways. First is in the dashboard extension, which now has predictive and prescriptive capabilities. The second is that we have included this in our calculation language as well, so you can create sophisticated calculations that react to predictions from the auto-ML model. Thirdly, we have integrated into our data prep engine so that you can do scoring or classification directly in prep. These are capabilities that you could not do natively in Tableau before.

When you are consuming a model, you can choose the fields that you want to map it to, and how you want to display the data. This is fully interactive, so you can click on the different categories or geographies and it recomputes the percentages. It also shows you the factors that affect the percentage, and suggest possible actions to increase the probability. So, it is not only giving you a prediction, but it is also giving you the context of the why, and the ways you can improve. And because it is interactive, you are getting it directly in the flow of the analysis, and you can ask an unlimited number of questions. This is something that has never been possible in Tableau without writing a lot of code or being a data scientist, but now I am able to do this as a typical business user.

Will this entire Tableau product be available natively within Salesforce?

The answer is “yes-ish”. These predictions are available in Salesforce, so users can still get the same value from Discovery in their workflow. I can also bring the Tableau experience into Salesforce and have it contextualised and actionable, but what we see is people taking the predictions and putting it into the workflow of the user, whereas the traditional Tableau is more exploratory, where you look across transactions and find patterns.

For branding, the Tableau brand is now the analytics brand for Salesforce. Before this there was a product called Einstein Analytics in Salesforce, but we have rebranded that to Tableau CRM, as it is essentially our analytics capability inside of the CRM experience.

CRM solutions always face the issue of “garbage in, garbage out” when it comes to data and insights, whilst Tableau’s expertise is in working with precise data. What are your thoughts on how you can overcome this problem?

We can only make decisions with the data that we have, so we are going to use the best data available and make as best of a decision as we can. What we have seen is that if you can add value to users and they see that value, they realise they can get even more value by putting in better data in the front-end. It becomes a self-reinforcing prophecy and they can become more motivated to make sure they do not get bad predictions.

The other aspect is speed — where they are able to see gaps quickly. Speed and agility is extremely important in delivering value to organisations so that their staff can react faster, the organisation can pivot faster, and they can adapt quickly to the changing landscape that we are dealing with.

Could you share the longer roadmap for integrating Salesforce and Tableau in the next few years?

Firstly, we are bringing Einstein Analytics, now known as Tableau CRM, and Tableau closer together, so that the AI capabilities, data capabilities, and experiences will become more seamless, and that will help our customers bring powerful, easy-to-use analytics inside Salesforce and across their enterprises. Second, we want to create a smarter Customer 360, so that what Salesforce provides is not just a system of record or engagement, but a system of insight and ultimately a system of truth about their customers, so that they can truly understand what powers their business and how they can react faster.

The last point is — independent of Salesforce — Tableau continues to focus on democratising analytics, and adding capabilities to make analysis simpler, more engaging, and able to work at scale so that customers can create data cultures and transform their companies with data and analytics. That is our ultimate vision, and we believe that in order to do so, we need to deliver analytics in the flow of business, in the flow where users work, in the way that people think, and that is really where all of the innovations come together to create an incredible analytics experience for customers.

In that regard, Tableau 2021.1 is a big release where we are bringing data science to every customer. We are democratising data science in the same way that Tableau changed the game for Business Intelligence. By bringing self-service analytics to the world, we are giving more people this powerful capability in an easier way, and our entire customer base will be able to use this predictive power to drive better decisions faster.