Retro Logo Executive One Page Report

For this project, I focused on clarity and using limited space wisely. I didn't want a lot of crazy colors or fanfare. I didn't want multiple pages from the user's perspective, and I didn't want anything difficult to use or comprehend.

This website (https://www.geeksengine.com/article/northwind.html) gave me the impression the Northwest Traders dataset was a pretty standard Microsoft practice one, so I used the picture from the article as the logo and used the color of the grey background behind the slicers and logo/title. I kind of liked the retro Windows 95 feel of the logo, and we don't get a lot of control over things like logos, so it seemed reasonable to work with it. I didn't end up pulling any other colors into my project from that logo as they're pretty tough colors to work with, so I leaned heavily on the grey.

I made the backgrounds in PowerPoint. I tried to maximize the space used with lots of measures, measure driven text, and tooltips.

The Sales measure accounts for discounts. In many cases, I used measures as labels built specifically to appear consistent and easy for end users to understand for sales and other metrics. I used DAX formulas to control the appearance of quite a few numeric value displays in an effort to keep the format and appearance consistent. I also used custom label options to highlight the highs and lows of data over time.

I chose to focus shipping metrics on Shipping Cost per Order instead of a Total Shipping Cost because I suspect we expect to see a rise in shipping cost as we see a rise in sales since there is likely more shipping to do. I suspect knowing the proportion of shipping cost per order is more useful to our target audience because they could see how much they're paying for the same volume of shipping services than they have in the past. I would bring this up to my end user on demo and welcome tweaks to this approach. Shipping Cost as a percentage of Sales seems like a reasonable metric to monitor over time instead.

One month of data was obviously incomplete, so I wanted to give the user the suggestion to remove significantly incomplete months of data from their visualization. That said, the report should be able to accommodate more data in the future, so I came up with a formula in a calculated column that suggests when a significantly incomplete month exists that can be sliced by the user as they see fit. As a suggestion to my end user to include complete months, I published the report with the filter set.

I wanted to show the proportion of the total on charts filtered using the Top N filter option on tooltips, which required using the ALL and ALLSELECTED functions to change the filter contexts. I based all my Top 3 filters on Total Sales, which I figured the executives would probably find more valuable than order counts, but I would definitely comment that could be changed if need be during demo.

I used the methods presented by SQLBI on linestx (https://www.sqlbi.com/articles/implementing-linear-regression-in-power-bi/) to get the slope of the trend line. Since the executives are interested in trends, I built an obvious label declaring whether or not sales and shipping costs are trending negatively or positively over time.

I wanted the top 3 charts to always be in descending order of % of total and unable to be edited by the user who might hit a column header on accident. I didn't include a visible rank number, so keeping the order in place is important. The tooltips don't seem to be selectable, so nothing needed to be done for those. To block the end user from accidentally selecting the column headers of the top 3 charts on the report, I physically blocked the end user with an invisible rectangle on top of the area I don't want the user to click.

I decided not to notate a potential point of confusion for the end user, sheerly in an effort not to clutter valuable space. The order count totals on the products tooltips are occasionally less than the sum of the individual product order count. This happens when two or more of the visible products exist within the same order and is technically correct, yet seems like the kind of thing a user might find confusing without clarification. I chose simply not to use the ink on this one in the report as it seemed like a pretty in-the-weeds opportunity for confusion that could be clarified once when the report was demo'd.

I chose to name the shippers tooltip "Top 3 shippers" and apply the Top N filter by order count, but this probably wasn't necessary as there are only 3 shippers. I wanted the top 3 functionality to work like the other categories when new data is added, including more shipping companies. I'd clarify this to the audience on demo.

I wanted to show the managers in addition to the top employees on the Sales Tooltip. One limitation I see is that I can only ever show the current manager of the employee, as no historical data is present. I don't have stop and start dates for the manager employee relationship, and employee/manager relationships do occasionally change in practice. An opportunity for greater accuracy for future data would include the ability to assign a manager to an employee during the time of the sale, but that would likely only be important if the report was focused on aggregating manager level data over time.