this is completely wrong, because you are counting the occasions when orders were recorded, which for most flavors is once per day every day (x4 sizes, x2 states) this is why your chart is surprisingly uniform across flavors and seasons. when that happens it's a good thing to suspect something is wrong in order to count the orders, like you label the axis, you would need to set Y to "orders" and then aggregate via "sum", not via "count" like this: Plot.groupX( { y: "sum" }, { x: "shortname", y: "orders", fill: d => colorScale(d.season) } ) and then of course remove the domain on Y or make it large enough

your data doesn't have "count" column, so d.count always returns undefined and resolves to 1, then radius scale turns that into whatever single size you get could be worth swapping d.count to something like revenue or orders or whatever. additionally, you have multiple marks at the same place, which you can see by changing "fill" to "stroke" this is because you have multiple sizes for each flavor and a lot of dates too it would be possible to introduce something like a "size" column — whatever comes after "pizza" in the name. and then play around with encodings to avoid overplotting remember that you also have facets fx and fy