Unravelling the Modern BI Puzzle: When Progress Looks Like Failure
Shifting the Lens on Success and Chaos in Data Implementations
As the wheels turn inside businesses, and people move on, I’ve worked with many data leaders who have recently inherited a self-serve BI platform. To be honest their assessments are sometimes bleak. As one professional said to me last week: “we have hundreds of unused dashboards, people in the business using tools to do simple tasks we can automate. Things have been allowed to grow out of control and it’s my job now to bring some order to the chaos.”
The indictment is strong; their predecessor didn’t have a clue how to manage a data platform. There is a clear need to establish a data pipeline, row back using from no code / low code tools, establish some data modelling, and start to build and support dashboards based on value. It’s clear that there is no data governance, with users unsure of where to obtain metrics and KPIs.
Ouch.
Over the last few years I’ve seen varying degrees of this story play out. Clearly the example above is extreme, but it’s not uncommon.
I’m going to dare to try and paint these “chaotic” implementations as successful, bear with me.
Chaos: The Uncomfortable Teacher
The entomology of the word chaos, is interesting. It derives from the late 15th century, denoting a gaping void or chasm, later formless primordial matter, via French and Latin, and ultimately from the Greek khaos “vast chasm, void”.
Chaos therefore isn’t the opposite of order, it’s simply the absence or removal of order. A void of a familiar structure. In that sense we can argue that chaos is simply a shake of the snow globe, a period where the natural order is shaken up and there is a void of order, while things settle and re-establish themselves.
The great thing about shaking things up is that things don’t settle in the way they have done before. As there world evolves, so our thoughts set themselves to the new reality.
Chaos has a way of being a helpful teacher, if an uncomfortable one. Moreover, allowing some deliberate, and controlled, chaos can be a great way of launching a data platform.
The Dark Ages
To understand what I mean by this, we have to step back before our data implementation, back to the “dark ages” if you will.
I recently had the pleasure of speaking with Dmitry Gudkov while he was in the UK earlier this year, and he explained it better than I ever could. He described these “dark ages” as having high levels of IT control, with databases and systems stripping away users access to data, and leaving business users with a lack of agency to work with data in the ways they needed. Spreadsheets gave a little of that agency back, but not enough.
As such there was very little decision making done around data, and very low data literacy. Those who could use VBA and coding skills gained the most agency, and the business turned to those individuals to drive value. Reporting looked backwards, with report packs helping educate leadership on the state of the business.
(Dimitri’s full article: https://www.linkedin.com/pulse/fun-fact-department-ancient-rome-dmitry-gudkov-zbvlc)
This is why tools like Tableau (and later Alteryx and tools like EasyMorph) drove the data renaissance, they gave the user back their agency with data and birthed a plethora of tools that drove the self-serve data movement.
The Renaissance
The historical renaissance was a chaotic time, marked by rivalries between states, each with their own laws and customs. While the renaissance marked a period of extraordinary creativity, the cities and streets remained crowded, dirty and dangerous, with no clean water and sanitation.
Our data renaissance, marked by the implementation of a BI platform, as opposed to the printing press, faces similar challenges.
At the risk of stretching our analogy, our erstwhile data leader, having implemented a self-serve BI platform, plays the role of the modern-day Medici, empowering the organisation by funding and fostering new tools and platforms. Like a Renaissance patron, they may not build everything themselves, but without their vision and support, no one else could.
So, as the architect of our renaissance, so our Medici must also accept the price, the void of structure, the chaos that comes with our upheaval of the natural order.
From order comes progress. From chaos comes innovation.
- Simon Sinek
In our case chaos comes in the form of innovation and learning. This innovation doesn’t come from writing a cheque for the BI tool itself, implementing modern BI tools isn’t innovative in and of itself. Users though, given agency, learn the tools and create dashboards and datasets and innovate locally, causing a small ripple of chaos as the familiarity of the existing order is changed. Successful and useful dashboards create an arms race that soon runs our of control, with business units trying to catchup and build reporting that matches their peers. In this way, as adoption grows, so does the value. Not in measured way, but in a haphazard, uncontrollable way.
Data literacy increases as users learn about databases and data structures. They learn about aggregations and levels of detail. Users learn visual best practice; converts declare pie charts evil and rage against red and green. Support for an all powerful “church” of IT diminishes as intellectual and scientific exploration flourishes.
In the Tableau world, we even see users embracing their creative side, creating artistic dashboards with data and publishing them Tableau Public.
The Enlightenment
And so we come back to where we started this journey, at end of our renaissance period, with fresh eyes.
Our Renaissance has been punctuated by learning, and literacy. Our users understand technology and data more than ever. They have explored APIs thanks to low and no code tools. They have sourced data and built dashboards. They understand what they need from dashboards and they can articulate their requirements better than they ever good before.
They also understand their limitations. They crave order now, their new found knowledge leads them to build or seek out consistent metrics, and governed data sources and dashboards. They also seek performant dashboards, dissatisfied with the performance from their home spun data sources. They cannot drive this themselves though, and so Data and Analytical Engineering Teams must be put into place alongside other undercurrents like Data Governance and Master Data Management.
And by the way, the product of this creativity, lie abandoned; unused dashboards and data sources have been replaced or forgotten. The necessary fruits of labour, no longer needed as requirements have shifted and changed.
Rome has fallen, our Medici has left, and our new leader arrives like much like Newton. They strive to create systems and foundational principles to help guide the new age of enlightenment, cutting through the noise to bring clarity and order.
I hope you’ll forgive the rather forced analogy, but the truth is that our data renaissance, punctuated by some chaos, is essential to help drive change and growth within the business. Without the Renaissance, there can be no Enlightenment.
In much the same way as the Renaissance, focussed on classical learning and culture, laid the foundations of the Enlightenment, so the learnings and culture changes of our BI implementation lead to acceptance of the logic and reasoning of our more mature data platform.
The Lens of Success and Failure in BI
Much like the Renaissance, outwardly a success in terms of creativity and advancement, but also rife with underlying chaos, inequality, and incomplete reform; our Data Renaissance is a success or a "failure" dependent on the lens through which you examine it.
It’s therefore critically important, in any BI implementation, to have a view of the success criteria and a vision of how the implementation may develop. We’ve talked about chaos being a helpful learning tool, but there’s a gulf of difference between uncontrolled chaos and deliberate “allowed” chaos. Chaos can be created by simply taking away central sources of support and help with data, and throwing a self-serve tool into the business without a plan, that’s not the chaos we are look to create. This needs to a supported change, that we acknowledge may pull away the rug from the established order for a while and lead to change and upheaval.
Data teams introducing new self-serve BI tools can take several steps to plan and get ahead of any issues during their “data renaissance” and avoid uncontrolled chaos. That’s really what this Substack is about, and here are some practical steps I’ll be covering over the next few weeks:
Be clear with business units involved in the roll-out about how we expect a change, a void of order, and ensure it’s clear that structure and order will need to follow at some point (and what form that may take)
Create a clear and structured roll-out plan for enablement and training
Ensure enablement strategies include best practice discussions
Promote “Gold Standard” content to areas they can be found to promote data quality and best practice, and demonstrate “what good looks like”
Undertake regular content reviews and work with business units to clear unused content
Perform risk reviews on frequently used content to ensure ownership and maintenance, and avoid data quality issues
Above all, we need to acknowledge the journey and the role of a data renaissance with the overall context of a wider evolution of a business. We also need to ensure senior leadership are aware of the challenges, costs and steps further along the line to help build out a more developed, and mature data ecosystem as we reach our enlightenment.
Moreover, I’ve seen an increasing push back against self-serve and dashboarding generally in the data community. Let us be clear, modern BI tools have driven a massive increase in data literacy and change over the last ten years. If we look back at this era and choose only to see the chaos then there’s a danger that the resurgence of traditional IT in owning data turns into a counter-enlightenment. It would be a shame if this lead to a rejection of self-serve philosophies, creativity and tools that have brought real value and exploration, and a return to the businesses having a lack of agency over their data.
And so a plea, if you do step into a role where you inherit a “failure”, take the time to step back and understand the history of the rollout, and be aware you’re simply picking up the platform at the next step of its journey. Please don’t be quick to declare that BI tool _________ is the problem, it rarely is if you’ve chosen a modern BI tool that allows exploration and gives business users agency. Changing a BI tool will likely lead to success, but if it does it’s likely it’s only because of the journey with the old tool.