I have the distinct privilege of curating a newsletter called the Data Science Roundup, and in a recent issue I linked to a post called Dashboards are Dead. Here’s what I had to say about it:
There is a lot a lot to like about this post. The thing I like best is that someone is actually putting real effort into re-imagining the dashboard-based BI status quo. I’m not saying that dashboards aren’t good and valuable, but I have personally been feeling some of the pains that the author points out in the post.
That said, I’m not sure that I 100% agree with notebooks as The Solution. Notebooks are awesome! But they’ve also been around for…a while now. Why is 2020 the year where all dashboards migrate to notebook form?
The thing I do 100% agree with in the author’s solution section is that “data is going portrait mode.” My personal belief is, though, that this portrait mode is going to look a lot more like Reddit—posts with constrained content types, specific topic areas that can be subscribed to, social features… You’ll train the feed to surface the data from your org that you most care about.
I realize that might seem like a bit of a big jump, but it’s something I’ve been noodling on a lot recently. Want to tell me I’m crazy? Hit reply, I’d love to hear it :)
Data science reader and all around amazing human being Alexander Jia—who goes by Jia—wrote an awesome reply and I wanted to make sure that it didn’t die in my inbox. Here it is:
Coming out of the woodwork to reply to Dashboards are Dead - I think it points out some very real problems, but I don’t think it digs into the fundamental issues behind those problems, so I figured I’d take a crack at it:
– At its core, the reporting side of data (dashboards, notebooks, excel and PPT, etc.) is not a technical problem, it is a people problem.
– While democratization of data is certainly an awesome thing, pure democracy is anarchy (poorly curated and contextualized data shared through a bunch of channels)
– Poor curation leads to confusion (which dashboard do I use), distrust (dashboards are wrong), and waste (unused content, unnecessary maintenance)
– There is a fundamental difference between standard reports (core KPIs, consistently measured, globally defined) and ad-hoc exploration
- dashboards are standard reports
– With a few exceptions, most people in an organization will lack technical ability and data literacy - documentation and education are great, but they will not solve this problem
Therefore to properly support the data needs of a business, in other words, empower people to do something meaningful with data, we have to solve for:
– Curation - only model useful data, only create useful dashboards, establish a single source of truth that is trustworthy while actively deprecating other reporting channels
– Context - add useful descriptions for your dashboards, for your dimensions/measures, focusing on why something is important and actionable
– Complexity - use systems that have low barriers to entry for technical ability and data literacy, architect intuitive data discovery patterns from dashboards => exploration
Here’s an example of a rollout plan to go with that:
– Create a single source of truth for your data model (build a warehouse, align on business metrics, centrally define curated and contextualized dimensions and measures, etc.)
– Release curated standard reports (start with KPI dashboards that actually tie to individual performance, work hand in hand with leadership to ensure they and their team will use it)
– Support ad-hoc exploration with dashboards as a starting point (Looker type exploration which leverages the underlying curated/contextualized data model is ideal here)
– Chip away at deprecating old reports and converting them to ones that you own - highlighting the improvements in user experience, efficiency, and data quality to users
– Over time, establish a sense of trust by consistently delivering better content while partnering with leadership / stakeholders / power users (data champions if you will)
By no means do I think dashboards are absolutely the right answer, nor do I believe that I’ve got it all figured out - but hopefully this provides a bit of a different perspective on these challenges for the reporting function in data teams :)
I think Jia’s take is spot on, and very well expressed. The exchange continued a bit further, but I’ll spare you the play-by-play. The two things that I want to follow on here are:
- I am an optimist when it comes to widespread skill acquisition. Many people looked at me as if I were crazy four years ago when I said that data analysts should work more like software engineers, but we’re legitimately seeing that shift come to pass on a massive scale. What else might be possible in the next four years? Having some amount of moderate technical skills with data enables people to be better at their jobs, and didn’t we see millions of humans learn how to write Excel formulas? I personally am long on this trend and we at Fishtown Analytics are investing heavily in learning programs to bring this future into existence. Jia agreed that this may very well happen and that he’d love to live in that world :)
- There are entire classes of tooling built to solve some of the problems that the initial post and then Jia’s response point out. Data discovery, data governance, observability, more… Some of these classes of tools are only either a) available at big tech or the F500 or b) in an extremely early startup stage, so they are by no means in widespread deployment today. And I think that there is a long way to go for a lot of these products to hit the mainstream. But I believe that tooling will ultimately play a role, along with people, in solutions.
IMHO, dashboards aren’t dead. The original post was written by a company with a viewpoint to sell (which is fine). But the post did a great job of highlighting some problems with the status quo that I feel like no one’s paying enough attention to. If you’ve been using the modern analytics stack for multiple years now, you’re likely starting to feel some pain. It’s probably starting to feel a little bit like the “wild west” to you—and this may be just as true of your dbt project as your data warehouse and your BI tool. “Who made this thing?” “Is it good?” “Is this the one I should be using?” “Does anyone else use it?” “Can I delete it?” “Is the data current?” “Why does this look weird?” And on and on.
These problems aren’t going away—as an industry, we’re going to have to solve for them. Recognizing them and talking about them is the first step.
PS: Just because the future might not look like the past is not a reason to believe that current products in the BI space might not adapt to, or invent, this new version of the world! None of this post should be taken as a slight to any particular product. Rather, it’s an attempt to have a conversation about what all of our favorite products will (or should!) look like in the future.
⏰ Jia is joining me on an office hours to continue this conversation. We would love if you joined us! You can add the event to your calendar here.
Last modified on: Oct 11, 2022