Let's talk about how to improve your visualizations. First let's start by figuring out what kind of analysis you're doing. I want to make sure you're spending your effort in the in the place where it makes the most sense to. And if you're just doing exploratory analysis where you are the primary audience, then the polish level can be low, right? Basically all you're trying to do is answer the question, what deserves follow up? Or what do I want to highlight going forward? If you're doing reporting, presumably you've done the work already to make sure that your metric is something that facilitates a good comparison. And for reporting, it's presumably over time. Here the polish level is medium. And it is totally fine to have boring charts as long as they are dependable. The purpose of this kind of chart is to answer the question, how's it going? And should we react? And then for storytelling, this is the place that can potentially deserve more of your effort, right? The polish level is high, especially if the audience is leadership or if the audience is a really broad, a high number of people. And here you really want to be able to focus your visuals on answering the question, is there enough evidence of this hypothesis, of this thing that I'm trying to answer in the data? So let's talk about what you can do with exploratory stuff. My first recommendation is just the thing that you will get as a result of a SQL query, which is a table. This will get you pretty far if you're just trying to answer the question, which category is biggest? But if you're looking at something over time, easy and default time series and bar charts are really helpful. And then the fanciest thing I will suggest is a scatter plot if you're trying to figure out how multiple categories or variables relate to each other. Moving on, I think you can get most of the way to any report or storytelling with a pretty simple selection of charts, either a line or bar chart with a single metric. Or if you've got a few segments to track, you could do grouped bars or multiple lines. And then occasionally, a 100% stacked bar chart will help you focus on changes in composition rather than overall number. So my advice, at least for the context of this class, is to keep it simple and to really focus on what the data says. But let's go through some examples of things we can do, even with the basics, just to clean up. So here's one kind of finishing touch, which is just label cleanup. Maybe I am trying to elevate this chart into something that is appropriate for reporting. So it's okay if it's boring, but I want it to be dependable and easy to read. And the thing that's really hard to read here is the chart title, the axes, labels. I don't really... It takes me a while to figure out what's being measured here. I think we're talking about counting users with orders, and I think we're talking about it in a rolling seven-day look back. So I might just change this to users with an order, seven-day rolling window. And because I have moved some of the explanation of the metric up into the title, that lets me keep the vertical axis simple. I'm just counting users. It's clear from the title what else I'm applying in order to restrict the users that qualify to be included in this. Another trick that I often use, and I'm not going to show, is I will often change the title to be a question, and then the data answers the question. So for this kind of chart, it might be, how many users placed an order in the last seven days? And then, again, I can use users as the vertical axis title, and it becomes clear what is measured here because the number here answers my question. I especially like to do that when I'm transforming exploratory work into something that's a little bit more polished for storytelling, because I've presumably already thought deeply about the question that I want to answer.