Shaping dashboards into a better "what if" decision model
“Everything starts out looking like a toy” (No. 14)
This week’s toy: An AI-powered Dancing Robot that’s a Bluetooth speaker. Now, if it could teach me better dance moves, that would be a serious invention ;). Edition No. 14 of this newsletter is here - it’s October 3, 2020.
The Big Idea
Dashboards rarely provide as much information as the information cluster in a well-designed car. More often they provide too much information to decide on any one answer, or provide conflicting indicators to proceed depending upon which problem you’re trying to solve.
Photo by Nick Fewings on Unsplash
When you look at a typical dashboard in your business, do you know instantly what it’s telling you? Or do you need to dive into the underlying numbers and reports to get a fuller story? There are a few new services that are trying to address exactly this problem. The key deliverable is providing a framework to make decisions and show sensitivity in the model itself.
Model as a Service
Causal, Tactyc, and Summit are systems that demonstrate the use of variables in a planning model and show the result in real time when an input changes. This is more than the basic ability in an Excel model to show “changer cells” - it’s a way to show how business systems change in real time. But will it work to drive decisions?
The best models map to existing behavior and show the variance in decisions and results, giving some inference to the underlying “why”. For example, in a sales system you might want to know that the overall deal size for deals closed in a quarter was higher or lower. By looking at the lead sources, the length of time for deals to close, and the overall size of the deal you can make an inference about the “best lead source” for your company. Validating that inference - or hypothesis - requires enough data to fit into that system so that you can accurately predict future outcomes. A spreadsheet plus a fancy model doesn’t do this by itself.
In lots of organizations, the prediction model and the dashboard are separate. The Excel model is building sensitivity tables and posing “what if” questions and the dashboard shows current state. They need to be merged.
Driving the “What if” Discussion
A better dashboard incorporates the best aspect of these modeling tools, giving the ability to ask “what if” during a meeting or while viewing the dashboard alone. In our hypothetical model, allowing the sales rep or sales manager to set a goal for a metric, recording it, and then showing the results vs the forecast is perhaps the most powerful use for a dashboard. Recording our “what if” hypotheses and comparing them against real results is a great way to use dashboarding to compare results to predictions.
This means you need a method to annotate graphs, or to store forecasts, to compare them later with what we knew at the time when we looked at the dashboard. Looking at the current state of data provides today’s view. You also need a method of “playing the movie back” to show what information was present when forecasts were made.
What’s the takeaway? Dashboards add value when they contain the key metrics needed to drive the business forward, and when it’s possible to answer “what if” questions both in real-time and as a backward-looking check on prior predictions.
We’d like to know …
We’re all on a lot of Zoom calls these days. What kind of camera do you use to look your best on a call - the built-in one that comes with your computer, your phone camera, or perhaps a dedicated SLR?
Click the tweet to vote.
Links for Reading and Sharing
These are links that caught my eye.
1/ A notebook-style method for data - Observable allows you to create visible, runnable code to describe lots of different things. It’s somewhere between low code and “interesting code.” I especially like the style of sharing working information in context and would love to see an equivalent method to demonstrate formulas in Salesforce.
2/ Counterfeit Plastic - If you bought a LEGO set off of eBay this year, you might have bought a fake. The market in used LEGO bricks has apparently become so popular that it has attracted copycats. Now, if I can find a 924 Space Explorer set in vintage condition outside of my parents’ basement …
3/ Go for 2 - As an armchair football fan, I often wonder what method teams use to determine when to go for a 2 point conversion vs. the standard 1 point kick after a touchdown. Tl;dr -> go for it in the 4th quarter. Also, RIP 4thdownbot.
On the Reading/Watching List
It’s a bit meta to share that one of the items I spent reading this week is itself a list of items, but here goes: A list of Notion Tools. This is a compendium of many ways people are using the popular note-taking platform.
I’m also reading Michael Lewis’s The Undoing Project, on the work Amos Tversky and Daniel Kahneman pioneered on decision making and the way we think. Thinking about thinking is pretty meta as well, but there we go.
What to do next
Hit reply if you’ve got links to share, data stories, or want to say hello.
I’m grateful you read this far. Thank you. If you found this useful, consider sharing with a friend.
Want more essays? Read on Data Operations or other writings at gregmeyer.com.
The next big thing always starts out being dismissed as a “toy.” - Chris Dixon