When Microsoft launched Power BI in its modern “2.0” form 10 years ago, it was all about speed and agility. Led by James Phillips who later became the owner of Power Platform, the modern BI designed for the SaaS era was aimed to disrupt the traditional enterprise BI with how fast the end-user could get value from the tools.

In many ways, it did change the game. When Power BI was later grouped under the Microsoft Power Platform product family umbrella (at least in marketing comms), this package introduced an incredibly powerful toolkit for the business users. The modern era of citizen development was given SaaS tools, instead of Excel and Access clients, to build solutions that weren’t trapped on their individual PCs.

Unlike low-code application platforms, business intelligence was a much more mature market. It had established budgets and owners in corporations. As a result, the uphill battle that tools like Power Apps and Power Automate faced with finding executive sponsors to fund the low-code initiatives was not a barrier with Power BI’s adoption. The market was ripe for disruption. Whereas Power Apps are still struggling to bring the “fusion development” story into reality, both the professionals and the citizens in the analytics space found a common ground in Power BI.

Unfortunately, in the software business, there rarely is a “happily ever after” stage where things just settle down. Even when you succeed in solving the problems of your user base, there’s always gotta be more. The incentives both inside the software vendor organizations as well as in the stock markets are aligned to keep the product teams “reimagining” things. That’s also what is happening to BI at Microsoft. Let’s look at what precisely is changing and why there’s a growing sentiment that MS is leaving the SMB audience of Power BI behind.

Introducing Microsoft Fabric

The duality of Power BI as a member of both the business applications platform as well as data platform was always a source of friction. A bit like with the CRM and ERP products under Dynamics 365 umbrella, everyone had to pick a side when stepping down from the product marketing gospel down to everyday reality of using and developing for the tools. You were either an apps person that also knew a bit about BI, or an analytics professional that combined apps, flows and Dataverse with the end products published via Power BI.

When Microsoft launched Fabric, it became increasingly clear that Power BI belonged to this product family - rather than in the Power Platform family of low-code tools. There would not ever be a “Power Platform SKU” to bundle both the BI and app products together. No matter how much customers would have wished for a simpler way to acquire the necessary licenses.

Power BI licensing had offered an enterprise friendly option for buying Premium capacity instead of per-seat licenses for users. This model has often been proposed as a way out of the complexity and barriers that have slowed down Power Apps & Automate adoption at scale. Fabric licensing is all about capacity - and it starts at a minimum of $262.80 per month for the smallest F2 SKU.

Microsoft assures that Fabric is just like Power BI Premium. At least that’s what they said when announcing the retirement Power BI Premium per capacity pricing options in March 2024. For any Fabric capacity smaller than F64 at ~$8k per month, you’ll still need to equip end users with a Power BI Pro or Premium Per User license, though. Otherwise, it would simply be too cheap, considering everything that Fabric offers. Which is basically “a solution to any data need you could ever have”:

How good is Fabric at meeting those bold claims then? If you listen to analysts like Gartner, they say it’s a leader in their MQ for Data Integration Tools. (No direct link included here, because Gartner, Forrester et al are a PITA when it comes to free promotion for their publications. 🙄) That’s hardly a surprise, though, given how the business for these big analyst corporations works.

What we really care about is the honest opinion of customers and partners that aren’t afraid to speak up. Looking at the more independent, smaller analyst firms, such as Directions on Microsoft, we get to hear things from the perspective of current Microsoft customers. Their recent presentation “Microsoft’s AI Strategy: Clever, Visionary… and Risky?” by Barry Briggs included plenty of insights on the role of the data layer in AI, as well as the competition that Fabric faces from the established titans of Snowflake and Databricks:

Recently, I’ve encountered opinions expressed by data platform professionals on LinkedIn that are questioning the readiness of Fabric for real-life production scenarios. This article written by Nadim Abou-Khalil reflects on the year 2024 in Microsoft Fabric and draws attention to issues that are causing friction:

The instability of the platform and missing features are something you could naturally expect from a new product offering. As things become more mature, issues on these fronts are likely to decrease.

What’s more worrying, though, is how the product fit in the broad market where Power BI has been successful is also decreasing. The growing complexity of what Fabric consists of is making it hard or impossible for PBI professionals and customers to follow along the path that Microsoft has chosen.

In particular, this post from Sam McKay (of Enterprise DNA) tells the story without trying to find excuses. Instead of saying “I’m sure it will get better over time”, the message is that there’s no change of direction to be expected.

“A massive departure from what Power BI used to be.” “So far down the road of complexity, there’s no turning back.” Those aren’t merely concerns about Fabric not yet being mature enough. It’s about the platform taking a direction that may leave a significant user audience behind - for good.

Subscribe to Plus to read the rest.