The Disconnect behind Tableau Pulse

One of the most misunderstood roles in marketing is that of the data analyst. The popular perception is that BI is mostly about creating attractive graphs in one of the leading data visualisation tools. While data presentation is indeed a core skill for any data analyst, it only forms a small part of the role. The day-to-day language of a data analyst is based around extracting, transforming and loading data rather than discussing the merits of pie charts versus bar charts. After all, few things annoy a CMO more than seeing inconsistent or inaccurate reporting.

The bulk of a data analyst's time is spent in back-end data manipulation, building the unified data models needed to power the charts that everyone sees. These data models take the disparate datasets that exist across the organisation and transform them into a set of metrics and KPIs that can be filtered and drilled into using whatever attributes make sense to the business. Many of the leading BI tools even have separate applications for building the back-end data model as compared to the front-end dashboard. This allows one data model to be shared across multiple dashboards, ensuring that key reporting metrics are tracked consistently across the entire enterprise.

As AI enters mainstream business usage, the distinction between front-end data presentation and back-end data model is poised to become even more important than ever. Both Salesforce and Microsoft are embedding automatic data visualisation tools into their applications. A few years ago, Microsoft introduced a 'visualise this view' feature to Dynamics 365, allowing any user to automatically create a report summarising the contents of a CRM lead queue or opportunity view. The graphs this feature creates are fairly basic, but they are useful for creating operational reports showing the status of open leads.

Metrics Layer

The rise of Generative AI has allowed Salesforce to take the concept of self-service visualisation one stage further. Last week saw the general availability of Tableau Pulse, which introduces the concept of headless BI to the Salesforce-owned BI platform. Tableau Pulse introduces a new metrics layer to Tableau that allows data analysts to define key KPIs the business wishes to report on, as well as build the data flows needed to enable that reporting. The headless concept refers to the fact that this metrics layer doesn't include any capability for creating graphs or charts, merely the raw data with the most important data points tagged as a named KPI.

Headless BI is a relatively new concept in the BI world that formalises the separation between front-end data presentation and back-end data model. The theory is that business users shouldn't be limited to the visualisations created by a data analyst but instead can create and customise their own reports and visualisations using the data stored in the metrics layer. This enhances data discovery while ensuring data accuracy is maintained. Dashboards can be customised in real-time to the specific questions that need to be answered without calling upon a data analyst to create a whole new set of reports.

Insights Layer

Naturally, Generative AI plays a vital role in enabling anyone to find the most relevant reports and insights for their current challenges. Tableau Pulse also introduces a separate Insights platform that allows users to track key KPIs and provides status updates on the state of those metrics. These insights are presented using natural language alongside the relevant graphs. Business users can even interrogate the numbers using a Q&A style conversation interface, making it easy to drill into the details behind the reports being presented.

The concept of an Insights layer itself marks an important shift in the role of BI within the enterprise. Making dashboards more interactive has been a trend for many years. However, Tableau has taken this one stage further. In Tableau Pulse, users no longer monitor reports or dashboards. They monitor metrics directly. The BI tool then generates the relevant reports necessary to drill into that metric at whatever level is required. This changes how executives interact with reports, allowing them to focus on the most important information. Then, when needed, the traditional Tableau dashboards will still be there to present a more holistic view.

To the uninitiated, the whole concept behind Tableau Pulse sounds like a threat to the traditional role of a data analyst. However, that's not actually the case. Instead, it allows data analysts to focus on the skills that deliver real value to the business. The ability to present data in an easily digestible visual is not a skill unique to data analysts, many marketers can do it too. It's the ability to organise different data sources into a usable format that sets a good data analyst apart from the rest.