At what point does "Self-Service Analytics" just become an excuse for unmanaged Technical Debt?
I’ve been looking into how large-scale analytics teams manage their stack, and I keep hitting a paradox that I’d love some senior perspective on.
We talk a lot about "Self-Service," but in practice, it seems to lead to a massive sprawl of 50+ page dashboards where 90% of the tabs are just slightly different filters of the same broken logic.
The pattern I’m seeing:
Business asks for a "quick view."
Data is siloed across 3-4 legacy platforms, so the analyst writes a "temporary" Python script or custom SQL to bridge them.
That "temporary" bridge becomes the foundation for a permanent dashboard.
Repeat until you have a maintenance nightmare where nobody knows the true lineage of the data.
Coming from a CS background, this looks like classic Technical Debt—we’re building UI (Dashboards) to hide the fact that the underlying data interoperability is broken.
My question is: Is it actually possible to move away from this "Dashboard Sprawl" in a complex org? Or is the manual work of "stitching" data together just an unavoidable cost of business because vendor tools are fundamentally built to be silos?
I’m trying to figure out if the future of the field is in perfecting the Viz layer, or if we’re due for a massive shift toward the "headless" / automated infra layer.