Dashboards are the problem, Not the solution

Some might describe me as bothersome, and they’d be right.
Those who disagree with me would say that I nitpick, and they would be right.
Others would say I have a funky body odor and I would be like, what do you mean, we never met in person, what do you want from me and please stop messaging me from fake profiles, I’m so tired of all this… Sorry.

Back to business.

The main thing I wanted to convey is that I do not do dashboards in any way, shape, or form. My opinion on dashboards is that they are a misguided idea, and as a result, I am against them, and like what others have already assumed about me several times, I do not have any expertise in either Tableau or Looker or Power BI, or any other tool that can produce dashboards.

Why are you against dashboards?

Let’s start with some basic assumptions.

Is there a role a dashboard should and can play?

  1. An aid to monitoring KPIs for potential malfunctions (such as a sharp decline in purchases, sessions, or conversions). In other words, it is a tool that promotes action.
  2. An instrument to measure KPI’s – to provide the manager, or end client, with a sense of where we stand in regard to achieving our main performance indicators (such as purchases, number of sessions, etc.)
    This can be viewed as a way to relieve analysts from the burden of conducting banal analyses.
  3. Data Transparency – A means of enhancing transparency between the analytic side and the management one (i.e., eliminate the feeling that the analyst is less than perfectly accurate).
  4. A means of shortening response time.

Whenever you create a dashboard for an end user, there are a number of tensions that can form.

The dashboard-end user-analyst triangle of death

  1. The spectrum of possibilities — a host of binary questions that multiply the complexity of the dashboard exponentially (should I give the client the option of changing dates? What about access channels? Age? Gender?)This spectrum
  2. directly produces an additional tension, which is (drum rolls for the second paragraph): How accessible vs. flexible is the data. If the data is very flexible, that is if we have a lot of options — the end client can play with the data as much as they like and start coming to conclusions. But then the information is inaccessible, the end client can’t just come, look and get their answer.
  3. A significant part of this tension is (drum-roll for paragraph number three): The tension between an analyst and analyzing — if the manager has a dashboard to play with, what is the analyst even there for? What is the manager there for? It’s not an issue of better or not as good, it’s a matter of fundamental assumptions.
  4. There are analytic mistakes an analyst would tend to make less often than a manager, because that’s their job (say, as a silly example, if I see half of my clients are women and half of them are from the north, that doesn’t mean these halves are the same, but it’s much easier to jump to these conclusions if we see them one next to the other without mediation).

What actually happens?

The dashboard is very quickly abandoned. Why?

Managers request a dashboard because they want to keep their finger on the pulse. But the same managers don’t look at sales daily — why would they?

  • Managers who are “data driven” can try and play around with the dashboard, but when it comes to it, a large part of the work put into a dashboard is thinking how to organize it in a way that gives the most information and one very specific* perspective
    * More than that will get messy.
  • Managers who want a succinct “one page” dashboard get a very generic product which doesn’t have the room to go into details.
  • Managers who want a very extensive dashboard get a three to four “pages” product, but as they can’t see everything in one place with a through-line narrative they get the (completely correct and justified) feeling that the dashboard is not what gives them the value.

What’s my practical offer?

Following my basic assumptions:

  1. If we have to monitor potential malfunctions we have to be able to define them. Visualizations are all well and good, but they’re not an index.
    We have to define acceptable ranges for specific KPIs and keep them few. (imagine getting 500 notifications every day for 30 different KPIs which were really really really important to someone, so you get notified on every little change. When I get that, everything very quickly finds itself in the spam folder. It’s 2022 and I’m not about to start filtering unnecessary information).
  2. Measuring KPIs — automated weekly reports, agreed on in advance by management (that receives them) and the analyst (that sends them). No meetings — brief, focused reports that give a sharp status on those same KPIs. If they still have some questions — excellent! We have phones, mails, zoom, random coffee breaks and carrier pigeons.
    Ok, the way you take load off is by automating things, not by losing responsibility and authority. That, I would say, actually has the opposite effect. If an analyst sends their manager a dashboard they might feel — wrongly — that they’d done their job, because they gave a starving man a fishing rod or something like that.
    This transparency, between an analyst (and their analyses) and management, in my experience there’s only one tool that can create that, and that’s disclaimers.
  3. Before every serious analysis I make I write down all of my basic assumptions, every action I take (say, deals from the last five years with a purchase volume of over X for clients buying through Y who are in the same country (or not necessarily)) — this transparency gives the other side the very important say on what information they like to consume. They also give doubters (a harsh word to write) the tools to reproduce the analysis.
    There are no shortcuts to building trust, no shortcuts to transferring responsibility.

That’s it for now, the comment section is probably going to host a lively debate about dashboards, which are definitely considered an important part of the analyst’s job (and you know I’m a bad-boy with a pleather jacket who doesn’t follow rules and gets results my way or something like that), but maybe I’m wrong and there won’t be a single comment, oh well.

One important note: I ONLY discussed dashboards, not visualizations. I consider visualizations to have a very high value, depending on the recipient and the use, but that’s perhaps for another time.


Hope that help you with your journey of becoming the best analyst you can be 🙂

More to explore

Well-defined processes and deserts

This post emphasizes the significance of clear definitions in data analysis. It provides examples of poorly defined questions and suggests ways to improve their clarity for accurate analysis. Additionally, it offers an interesting answer to the question of “Where’s the world’s biggest desert?”

Statistics and analysts

Let’s talk about the importance of statistics for analysts Since this is a major topic, and statistics (and math in general) has

Analysis Paralysis

Analysis Paralysis

Don't be a stranger!
Contact me

right here!

Leave a Reply

Your email address will not be published. Required fields are marked *