Fabric is a team sport
ONE PERSON CAN’T LEARN OR DO EVERYTHING
…why “Fabric experts” don’t exist and there’s a need for a clearer vision about how different teams can use Fabric
In the goblin world, there are no sports. Instead, goblins go exploring dungeons for loot and treasure. It’s basically the same thing. So, when goblins go to do this, it’s important that they put together a good team. As we all know, each dungeon is different, with unique challenges, threats and opportunities. If a goblin were to go into one of these dungeons alone, trying to be a hero… well… let’s say it doesn’t end well when that happens. To avoid this, different goblins with complimentary skills cover different parts of the dungeon from end-to-end. No goblin is identical in their unique sets of skills, yet each of them do fill a defined role. Together, each of them play a part in this process to ensure an optimal outcome; they do together what no goblin can do, alone.
While we are not goblins, we face similar challenges in our own data dungeons with tools like Power BI and Fabric. When we expect a single expert to do everything, we set ourselves up for failure. That should be an obvious statement, and yet, it’s not clear in a lot of content about Fabric. Furthermore, there’s a need for more information and guidance about Fabric that’s tailored to specific personas, teams, and scenarios.
In this article, I share some unfiltered thoughts and concerns about how we talk and produce content about Fabric and its capabilites.
The purpose of this article is threefold:
1. To call out that it's unrealistic to expect people to "learn Fabric" and it's impossible to be a "Fabric expert".
2. To reflect on the consequences of this language in the market for orgs, professionals, and job-seekers.
3. To suggest that content about Fabric would be more helpful if it addresses specific personas or scenarios, instead of lumping everything in one pot.
UNIFICATION IN FABRIC
In brief, Fabric is appealing for reasons like the below:
It unifies what were previously separate parts in the end-to-end analytics pipeline.
It provides an elegant all-in-one user interface, with a low-code user experience familiar to Power BI.
It introduces many novel innovations that have the potential to help you do more with your data.
While this unification can simplify many things from billing to day-to-day use, this doesn’t mean it unifies the roles and skills of the people who will use and manage these things. This is unchanged; with Fabric, you still need a team of people with different skills and specialties that cover your analytics pipeline. Unfortunately, however, I’ve noticed a problem where many people and organizations are simply not thinking this way.
‘DATA HEROES’ AND OTHER PROBLEMS
I always take issue with requests for a Power BI expert, a Power BI consultant, or a Power BI developer. Power BI is such a massive tool, and roles like this seem to perpetuate the myth that a single person can know it all, which is just silly. However, I also understand that it’s very difficult to stratify individuals and identify roles for Power BI, since there’s so much diversity in the composition of teams and how people use this tool. Furthermore, many organizations just don’t know any better; if they’re new to Power BI, they don’t understand what roles exist and what they’re looking for.
However, with Fabric, I’m concerned that we are seeing the next evolution of this trend, as expectations emerge for “Fabric experts” and “Fabric developers”. Unfortunately, I think this is not just a natural evolution of the ambiguous roles and terminology used; rather, it’s a direct response by the market to the way that we’ve been talking about Fabric as “one big thing”. While this makes for easier, cleaner storytelling and marketing, it does also misrepresent the complexity and scale of the platform, everything inside of it, and the different expertises required to use all of this effectively.
PROBLEM 1: UNREALISTIC EXPECTATIONS AND STANDARDS ARE BEING SET
I’m concerned that organizations and individuals are forming unrealistic expectations about what Fabric is and who uses it. I hypothesize that this is an indirect consequence of all the content that talks about Fabric as “one big thing” and the calls to “learn Fabric” / “upskill to Fabric”. This has already happened in the past with Power BI, as many organizations and individuals approach it as a tool that only produces visualizations—PowerPoint for Data as it was once called. I’m concerned that with Fabric this might be happening again, but in a different way.
In my professional environment, I have observed this take several different forms in recent months:
More frequently seeing or hearing requests for consultants or applicants that fit the role of a “Fabric expert”, either explicitly or implicitly (“applicant has deep knowledge and experience with all of the Fabric workloads…”).
Hearing of professionals (either FTEs or consultants) that are tasked to “learn Fabric” or to “upskill from Power BI to Fabric”. In many cases, this is a direct consequence of the first point.
Power BI professionals are being expected to know about areas in Fabric that they previously had no experience with or responsibility for. A good example is a quote from a friend who is a Power BI consultant: “Why are you asking me about the Fabric Data Warehouse? I make Power BI reports; I have no desire to build a data warehouse, or else I’d have chosen to be a data engineer.”
Certifications covering a wide scope of the platform (DP600 “Implementing analytics solutions using Microsoft Fabric”). (I added this point as an edit on April 17, 2024; thanks to Marc for mentioning it)
PROBLEM 2: NON-EXPERT AUDIENCES ARE VERY OVERWHELMED
In my personal environment, I’ve observed a fatigue among people who are learning Power BI or Fabric. It goes beyond the daunting difficulty of “Where do I start?”. Outside of the expert communities, I notice that layman audiences are confused or overwhelmed by the volume of information, the features available, and how (or what) they should use for their specific scenarios (and perhaps more importantly, what they should not use and why). My concern is that this will lead to more people “tuning out” the new updates or giving up on learning, altogether.
I think that this is due to several things:
An over-focus on marketing features and technology (the “whats” and “hype”) rather than more grounded discussions about the problems that these features and technology address (the “hows” and “whys”).
Lack of a clear vision about how a team can use Fabric in different real-world scenarios. Instead, most of the content seems to imply (implicitly) that a single individual is using all these tools and features (the Power BI developer who evolved into the “Fabric analytics engineer”).
Lack of content that presents a balanced perspective, which highlights both when to use something (the benefits) and when not to use it (the caveats). An example of content that does present a good balanced perspective are the articles from Marc Lelijveld (like his recent post about refreshing semantic models by using Fabric Pipelines). Marc presents the scenario, clearly outlines the use-cases, then reflects upon the shortcomings. Perfect, clear, practical.
At the conclusion of the FabCon 2024 keynote, the speaker joked, “Who can remember everything that we talked about in the keynote?”, implying that it was too much for the audience to remember. This is precisely the problem. It is too much. If this is true and we are self-aware of it, then why do we keep spraying people with information, announcements, and content like pressurised water on a forest fire? Why do we not find more efficient ways to deliver this information, and deliver the right information to the right audience?
Other examples of this include:
The monthly announcements of the entire Fabric platform; an overwhelming wall of text (March 2024 is 76 pages if you use print to PDF) without visual or video overviews. Thankfully, updates to specific products like Power BI come with their own blog.
Content channels like YouTube, and blogs that used to produce more focused content about Power BI now produce content about Fabric. This can be confusing to audiences (it sure was confusing for me, last year), and sends the implicit message that they, too, should pivot to “learning Fabric”.
Many Power BI user groups were now re-named to Fabric user groups. This can be similarly confusing, particularly since many of these channels and user groups provide deep-dive sessions that are highly specialized, so the content focus will vary wildly from one session to the next.
To make this situation more grounded in our everyday reality instead of “the world of tech”, let’s compare to the field of medicine. In medicine, you have general practicioners (GPs or house doctors), who are generalists with a broad knowledge across many fields. However, many problems require more specialized knowledge and skills. In these situations, a GP refers you to a specialist, like a neurologist (for neurological issues), a cardiologist (for cardiovascular issues), or a dermatologist (for skin-related issues).
“LEARN MICROSOFT FABRIC”
When you’re telling someone to “learn Fabric” or “become a Fabric expert”, to me, this is the same as telling someone to “learn medicine” or become a “medicine expert”. It sounds ridiculous; it’s just not possible. You’re telling that person to learn an extremely broad spectrum of theory, tools, and technologies. What’s worse, you’re implying that people do this in addition to their day jobs. It would take significant time and resources for one person to truly “learn Fabric”. It’s a non-statement because it’s impossible to achieve the implicit outcome; you can’t “learn Fabric”; there’s simply too much there.
“UPSKILL TO FABRIC FROM POWER BI”
If you’re telling a Power BI specialist to “upskill to Fabric”, to me, this is like telling a neurologist to “upskill to cardiology and dermatology”. You’re telling a specialist in one area to expand their specialization to encompass many more areas. Even as-is, the concept of a “Power BI Expert” is already dubious; Power BI itself is such a massive ecosystem that involves a tremendous number of areas from data visualization to tabular modeling to an increasingly complex administration.
THE FABRIC DATA HOSPITAL?
Ideally, this dialogue shifts away from these broad calls to learn the platform and steers more toward learning that’s grounded in real-world scenarios, challenges, and roles.
I’d expect that people should choose between becoming generalists (your Fabric GPs) who have a very broad but shallow knowledge of Fabric and its components, or specialists (your Fabric neurologists, cardiologists, and dermatologists) with deep knowledge in one part of the platform, although a limited knowledge of other areas. Ideally, a team who works with Fabric comprises sufficient generalists and specialists to cover the data needs and use-cases of an organization. Continuing with this overdrawn medical analogy, a team who works with Fabric isn’t a thinktank of “Fabric experts”, but a data hospital where everyone plays their role to achieve an optimal outcome.
(Fun fact, I wrote this from a hospital, thus the analogy).
A MESSAGE FOR THOSE WHO WANT TO LEARN ABOUT FABRIC
If you or your organization are interested in Fabric, want to learn about it, or want to use it, then the following messages are for you.
1. THERE IS NO SUCH THING AS A “FABRIC EXPERT”
Given the size and scale of the Fabric platform, it’s simply not possible for one person to be an expert in Fabric. The bredth and depth of knowledge goes beyond what any single individual has or can obtain; it encompasses numerous technical areas, disciplines, and processes.
If someone is claiming to be a “Fabric expert” (or an expert in any data platform), then one of two things are true:
They are intentionally lying or misleading you for their own benefit, such as to grow a following or to sell you their product (i.e. a course, a tool, or Fabric itself) or their service (i.e. consulting). This is unethical, for obvious reasons.
They are ignorant of how big Fabric is and the different skills needed to manage it. This is dangerous, as such “experts” can be false prophets that lead you into deep waters without a boat when it’s not clear that either of you will swim.
This is not a caveat of Fabric. If anything, this is somewhat a bragging right; Fabric is so big and powerful, with so much potential, it’s simply not possible for one person to master it all. You need a team.
2. ONE DOES NOT SIMPLY “UPSKILL TO FABRIC FROM POWER BI”
“The majority of your Power BI users are bottom-up business analysts without a technical background, yet you want them to use very technical parts of Fabric like lakehouses and notebooks. You want people to ‘upskill’ to this massive new tool, yet you don’t give them the time or resources to learn. Most of your issues are related to people or processes and not technology. You say your main reason to use Fabric is for Copilot and generative AI, yet your master data has more blanks than a prop firearm, you still can’t agree to a definition of “margin” or which adjustments are in “gross sales”, and your data is mostly hundreds of Excel exports on a SharePoint site.
Think, Bonk the Business Goblin, think!”
This phrase implies that you can and should expand your knowledge to encompass the many items, features, and capabilities of Fabric. It implies that the natural learning path is to go from Power BI to Fabric. This isn’t realistic, and it’s also not true. It is simply too much.
Instead, you should see the capabilities of Fabric as an opportunity to compliment your existing processes and address existing or new problems in better ways. For instance, if you develop Power BI semantic models, you want to focus on features that relate to semantic models, like Direct Lake or semantic link in notebooks. Then, you might understand how you might benefit from them, and what caveats or limitations they might have for your needs. As such, it’s not about upskilling, but rather improving the efficiency of what you do today.
Furthermore, if you’re just interested in something like notebooks and semantic link, you can also see this as an opportunity to benefit from those tools. If you’re interested in learning them, you can and should—provided that your organization gives you the time and resources to do that. You’re under no obligation to spend your free time on this stuff.
A MESSAGE FOR THOSE WHO WANT TO TEACH FABRIC
If we want to teach or produce content about Fabric, it could be helpful if we frame our content with respect to the real-world personas, scenarios, and teams that use these tools, today. This could be more beneficial rather than providing a large, generic, all-in-one perspective. Instead, we should think more about who will use these tools, and what they will use them for.
This is true for articles, videos, presentations, and even social media posts. Doing this will not only help people to learn, but prevent the market from misunderstanding Fabric and who uses it, or creating unrealistic expectations. Consider the following:
Who is this content for? Which roles, or which tasks will benefit from this? Who is it not for?
What existing problems does this solve, or what new business value does it create?
Is this something for an enterprise developer? Citizen developer? Self-service user? Someone else?
Is there nuance that you can use instead of “X all the things” or “Y is amazing”? What are the shortcomings or contra-indications?
TO CONCLUDE
Fabric unifies many tools and technologies, which simplifies how you manage your end-to-end analytics pipeline. However, it doesn’t simplify it to the extent that a single person can possibly know (or learn) all of Fabric; Fabric experts do not exist. The platform is simply too big, and covers too many technical domains. Unfortunately, however, there’s emerging indicators that suggest that some organizations and decision-makers do not see it this way.
I think that this is in part due to the nature of content about Fabric, which so far tends to focus on features or technology, or the platform as a whole. Content creators and learners might find it more helpful to instead approach Fabric from real-world scenarios, grounding this content in how and why different parts of the platform are used. Furthermore, I think that we have to acknowledge that it’s simply unrealistic to ask someone to “learn Fabric” or “upskill from Power BI to Fabric”, and not produce content that delivers this message, explicitly or implicitly.
In the future, as Fabric is more widely adopted, usage scenarios, architectural patterns, and common challenges will emerge. When this happens, I hope that there will be more clarity about what a team who uses Fabric can look like, and how these different personas leverage these features to make a whole that’s greater than the sum of its parts.
