Integration developer | Founder @ Portable | Fixed-price, reliable, ELT
I always recommend a simple playbook for building and refining dashboards: 1. What's happening 2. Why is it happening 3. How do we fix it 4. Fix the problem, a bunch of times 5. Automate the problem away 6. Delete the dashboard Why not just skip to step 5 (automating stuff)? Because you really need to get step 1 correct, otherwise steps 2-5 will be a waste. Let's drill down into each step: ___ 1. What's happening Most dashboards start with a bar chart and a table. A bar chart of a key metric over time, and a table below it showing the raw data (to double click into stuff). The goal of this step is to identify either an output metric (like revenue, sign ups, etc) or an input metric (emails sent, candidates reached out to) and watch it move. 2. Why is it happening Don't skip to step 2 too early. First make sure that looking at step 1 (what's happening, is actually worth double clicking into). For some KPIs, all you need is a bar chart and a table. But when you need to understand the why, I tend to start with a drill down into a row of the table from row 1. This could be a single customer view, an employee view, etc. Get more details (tables, charts, summaries) and make them available to users to try and figure out why things are happening. Don't try and skip to a solution. Just throw a bunch of raw data into one place. 3. How do we fix it Over time, you'll add data points in step 2 and remove them. The layout of the dashboard will change and evolve. This is because you're iterating towards a clear path to fix the problem in a repeatable way. The goal of this step is to find a model that flows naturally and works in a repeatable way to fix the problem. 4. Fix the problem, a bunch of times Now that you have a working approach, start using the dashboard to solve the problem. Use it again. And again. Make tweaks any time the solution is not perfect. Add toggles, optimize the layout etc. Make sure it flows, and works for edge cases. 5. Automate the problem away Now you know you're solving a real problem, you found the main data points to identify and address the issue, you've created a step-by-step workflow to resolve the issue, and you've battle tested the solution. At this point, start figuring out if there's a way to automate the solution. It might involve engineering effort. It might involve an automation tool or RPA solution. But just imagine. Once you automate the solution, you can finally... 6. Delete the dashboard This is always the best part :) If you find a solution to the problem, it's time to move onto the next problem. ___ Everything in the business doesn't warrant all of these steps. I've built dashboards that get to step 1, we build a bar chart and a table, use it to measure progress for a few months, and delete the whole thing when our priorities change. Priorities always change. Make sure you're only going deep on the problems that are absolutely critical to your business RIGHT NOW.