Organisations outgrow dashboards
Dashboards are ineffective but continue to be used because they are Schelling points
One of the names I had come up with for Babbage Insight, which was quickly shot down (no pun intended) by Manu, was “Pinochet”. Because he was the original “dashboard killer” (the name was shot down, of course, because Pinochet killed much more).
The corporate dashboard as we know it was initially invented by the socialist government of Salvador Allende that took power in Chile in 1970. If you think about it, a company is similar to a centrally planned economy - a bunch of guys at the top monitoring and making decisions for large numbers of people. And this bunch of guys needs to know what is happening, and in the case of Allende’s government, they decided that a “dashboard” showing key numbers that the President could see daily was a good way to monitor stuff.
Of course, Allende’s government didn’t last long, being deposed in a bloody coup led by Pinochet in 1973 (you should listen to this brilliant Rest Is History podcast about Pinochet).
The dashboard, however, lives on. Like a planned economy, company leadership also needs to keep a constant tab on what is happening in different parts of the business, so that appropriate decisions be made, and in a timely fashion.
The question, however, is - is the dashboard (as we know it now) the best way to represent information, and present it to the company leadership?
(Out)growing pains
Over the weekend, I was talking to Arun Srinivasan, founder and CEO of Clarisights. He said something that caught my attention - that after trying to sell to startups, he discovered that “enterprises that have outgrown dashboards and gone back to spreadsheets” are an interesting market for his product, and he now exclusively targets them.
I was initially surprised by this, since my prior until then was that dashboards are essentially an “enterprise product”, but thinking about it more since our conversation, I now fully see merit in the fact that beyond a point dashboards stop being useful.
So why do companies “outgrow” dashboards? I can think of a few reasons, in no particular order. I should thank Arun for directly contributing a bunch of these thoughts, and sparking some others.
As the BI team grows in size, software engineering practices start being put in place. There will be a product management team to interface between BI and business. Bureaucratic friction comes in - to get anything done, you need to raise a ticket.
Related to this, distance appears between the users and the makers / maintainers of the dashboards. This means that when users want something to change, they need to go through the aforementioned bureaucratic process. Beyond a point, they give up on this and bypass the system
But then dashboards ought to change. As Cedric Chin puts it in this article on “how Amazon measures itself”, where he is talking about Bill Carr and Colin Bryar’s Working Backwards,
The authors note that if you improve your process over time, it is possible for a previously useful metric to stop yielding useful information. In such cases, it is totally ok to prune it from your dashboards.On the other hand, the reason we need bureaucratic process is that after a point lots of people will be using the same dashboards. They will have different needs, and so we need a team (BI product management) to decide what is the best way forward for the dashboards, and get that implemented.
This necessarily means that even if there is a desirable change in the dashboard, it takes time (and development “sprint cycles”) to get implemented. Again reason for people to exit the systemThe best data engineers want to design data pipelines and warehouses, not maintain them. As they go off to do something more interesting, their place gets taken by less capable data engineers, who go into “maintenance mode” (rather than “building mode”).
They stop optimising pipelines. If nothing is broken, they try to not fix anything. Soon, pipelines become slow. Dashboards that would load in 15 seconds now take a few minutes. More reason for people not to use them. More people will move away from dashboards to use Excel (iff they can get their hands on the data, that is - the bureaucracy, once established, zealously guards the data lest someone bypass it).There comes a point when there are enough “forks” (people doing their own analyses) from the dashboard that the dashboard itself stops being a single source of truth (as it should be). The process of dashboard death has begun here
Related to what I wrote earlier this week, this further marginalises the BI team (since they are now building and maintaining dashboards that nobody really wants). The team reacts by becoming more timid.
Sweet spot for dashboards
This makes me wonder - is there a “sweet spot” in terms of company size / complexity where dashboards are actually the best way for executives to consume data and insights? When the company is too small, data is insufficient, data pipelines are immature and there is not so much value in a dashboard. When it becomes too large, the bureaucracy around dashboards renders them unusable.
And if dashboards are useful only for a particular “sweet spot” of company size, are they actually effective tools at all? Are people continuing to use them simply because they are a Schelling point? (the current option is a great Schelling point, if you think of it)
If you have opinions on this, in any direction, I’d love to talk to you and listen to you. You can either reply to this email (if you’re getting this over email) or leave a comment to get started.
A few years ago, Sudheesh Nair, the now former CEO of ThoughtSpot had declared that he “wants all dashboards to die”. I can fully empathise!