The Tableau Exodus Has Begun

What to do when your executives pull the plug on your BI tool.

Over the last few weeks I’ve had no fewer than five conversations with data leaders that all told me the same story; they were dropping Tableau. And while some of them had criticisms of how Salesforce has managed the platform following the acquisition, none of them mentioned that as the driving factor in the decision. Instead, they all said it was just too expensive.

In each case, the decision was made above the data team at the executive level, and in each case the data leader I spoke with was not thrilled. Migrating BI platforms is expensive even when there are significant licensing reductions, it’s a huge distraction from any strategic work the data team could deliver instead, and usually nobody is better off afterwards. You’re just re-arranging the deck chairs.

---

A special thank you to MotherDuck for sponsoring Super Data Summer. I love their new Dives feature; AI authored analytics apps delivered in hours, not weeks. Check out the financial reporting Dive I built here, and learn more at motherduck.com/product/dives

---

A few trends revealed themselves during the conversation that we should really sit with as analytics and business intelligence practitioners. First, executives often thought BI was ‘just dashboards’ and that AI could deliver the same results for less money. The hard work of BI was invisible to them. Next, the value the BI department delivers was not enough to overcome the expense of the tooling. Finally, cultural factors were ultimately the driver of low perceived value; reducing license costs and onboarding new software does nothing to fix that.

In each case the data leader was clearly frustrated and looking for answers - I have less than a year to get off Tableau, what should I do? What software should I consider instead?

If you’re in this situation, first of all I’m sorry. I know it sucks. And secondly, here is my advice on what to do. I don’t know your specific situation; if you want to deep dive on it DM me, I do this for a living.

Focus On Metrics Locked In BI

There’s this platonic form of a data environment where all the metrics are in a data warehouse with a well documented and designed semantic layer serving up verified results to users and developers. We sometimes catch glimpses of its shadow on the cave wall.

In the real world, critical metrics that run the business are often defined ONLY in the BI layer, nowhere else. And nobody knows that, until migration time comes and suddenly you have a problem on your hands. This is compounded by the sheer size and complexity of old BI environments, like many Tableau installs today. Tens of thousands of reports with little understanding of who does what with them or what they contain.

So your first task is to map all that out with a particular focus not on the dashboards, but the metrics they contain. You need to untangle the web of what operationally or strategically critical data points exist only in BI and plan your migration around replicating them into your warehouse if possible, or safely shepherding them into the new BI tool.

Data that exists in five places doesn’t need to be migrated at all, and taking one necessary calculation out of a Tableau workbook and into a database is much easier than replicating the entire dashboard that delivers it.

Get Creative Finding New Tools

Some of the leaders I spoke with were in a real tough situation. One of them had done a pilot of Omni and felt it was the best option. The Omni semantic layer would help organize their data and the AI tooling was ahead of what they expected. There was just one problem.

They couldn’t afford Omni either.

If there’s huge cost pressure to reduce BI spend, you are probably not shopping at BI Nordstrom, you are shopping at BI Nordstrom Rack. And that’s okay - there are lots of options out there that are very good at much lower prices. Look outside the ‘Leaders’ quadrant of the Gartner MQ for enterprise vendors at lower price points for traditional BI replacement.

One obvious option if you’re in this trap - and honestly I expect for large companies this is probably the most appealing - is to just port your Tableau workload to Power BI… or your Power BI workload to Tableau, or Amazon. The reality is you can probably lower BI spend significantly just through vendor consolidation. There are some really good specialty consultancies who can make this way easier for you. DM me if you need a recommendation.

But don’t restrict yourself to popular vendors. Search for smaller startups that might offer what you need. There are really interesting options like Golden Analytics, Ridge Data, Zenlytic, Count etc that can take a significant portion of your BI workload, often times for a reasonable price. Some of these are more like traditional BI, others are pretty different. But being traditional brought you to this point, maybe it’s time to shake things up. And getting in early has its advantages.

Start A Vibe Coding BI Pilot

AI is the elephant in the BI room. Multiple people I spoke with shared that executives seem to think that BI is easily replaced by AI, so why are we paying for this? If you read Super Data Blog you already know that this isn’t true. Yet.

But let’s be real with one another here - the day is fast approaching where vibe coded BI applications will compete with traditional BI workloads directly. They are already more flexible and easier for both self-service and professional users to build. Things like MotherDuck Dives or Zenlytic Artifacts create extremely good looking reports very, very quickly.

The challenge is all the other stuff BI does - security, auditability, content management, reusability, load balancing, caching, distribution, metrics stores - it’s a big list. AI can already make a good dashboard, it can’t automatically personalize a report and send it to 1,000 people every Monday morning. There’s nothing out of the box that combines vibe coding with classic BI infrastructure to my knowledge. Hex Data Apps might be closest - I haven’t dug in yet.

It also brings up an existential question for BI platforms - does the presentation layer even need its own tool anymore? I am working with MotherDuck on some really wild ‘BI right in your database, developed by Claude’ stuff that goes beyond dropping AI built HTML dashboards in a shared folder. If that sounds crazy, well it is. But crazy in a good way.

Do Some Soul Searching

You’re walking out of a meeting shell shocked. The powers that be just told you that they won’t be signing your upcoming BI renewal. Panic is setting in. Take a deep breath and follow the advice above.

But not before you do some soul searching. Here’s a painful fact - if your BI practice was delivering real value, you probably wouldn’t be in this situation. One of two things is true; either your work just isn’t that important, or you’ve done a terrible job marketing it.

Business intelligence is often pigeonholed as the dashboard factory. Nobody sees the hard work that goes on inside the factory, they just get a car from the dealership and drive away. And they don’t think about it again until the car breaks down. That’s where you are now.

It’s time to get creative and throw out your old ways of working. That’s the one silver lining of a migration - suddenly the operational drag of ‘I need this slight report variation by 5PM’ stops and you can step back and ask, what would make this function really valuable at this company? So look at new tools, yes, but also new approaches. Embrace AI to the greatest extent possible to make your work faster, better and more targeted to end user needs. It’s an incredible tool to help you think, even if you never use it to write a single line of code. My upcoming book teaches these techniques, but you don’t have to wait. Start now.