APAC Command Centre — a business performance dashboard with drill-down views across Product Performance, Customer Cost/Value, and Operations & Shipping, tracking gross profit, margin, and return rate trends.
Spotify Analysis — an end-to-end music streaming analytics build. Raw data cleaned and joined in Alteryx (900K+ play events, dedup, date/boolean standardization, a custom macro), visualized across an Executive Overview and Content Performance dashboard with parameter-driven Top-N views and a custom dark theme.
Would appreciate feedback on:
- Visual design — readability, contrast, anything that feels off
- Chart choices — anywhere you'd have visualized something differently
- Navigation/drill-down flow — intuitive, or confusing anywhere?
- Anything that reads as unfinished — direct critique welcome
Links below — thanks to anyone who takes a look!
APAC Command Centre -
https://public.tableau.com/app/profile/aviral.trivedi/viz/APACCOMMANDCENTRE/HubOverview
Spotify Analysis -
https://public.tableau.com/app/profile/aviral.trivedi/viz/Spotify_Analysis_17884476660320/ExecutiveOverview
#Tableau Public #Reports & Dashboards #Tableau Desktop & Web Authoring #Trailhead Challenges
We all start somewhere, and I think this is a good effort at making an interesting and structured dashboard, with your navigation being good, and your main page having a good number of KPIs, so not overloading the user. My points below are mostly based around deciding what the question is that you're trying to answer with each chart, but I think there are a few things wrong with individual elements, that you need to be careful of (especially Top N and the Order of Operations). Below is my (personal opinion) feedback, which you're welcome to ignore, should you so desire, sorry it's long.
I find trend (line) charts that combine multiple years confusing. Your chart suggests an amazing trajectory of sales, no matter which year(s) you select. It doesn't illustrate that, every January, there is a catastrophic drop in sales (compared to December). In my opinion, your chart doesn't show a trend when you have all years selected, you're comparing "Octobers" (plural) with "Septembers" (and the other months), which is a discrete date comparison and I'd plot as a bar chart. It's only a trend when one year is selected, because then all the months are in sequence, but because you're only showing your users Jan to Dec, you're misleading them into believing that sales just keep going up and up. Your chart omits this line at the end of every year, which feels kind of important:
For business dashboards, I would stick to one method of telling the user the values on the chart. On some of your bar charts, you have a y-axis, you label every bar, and you colour the bars according to their height. While colour is a personal/customer preference (I usually keep them grey, and pick out key points with colour), your y-axis, at least, is redundant, because I'm not going to read the exact(ish) value of 5.9m, and then look across to the y-axis to check it also says about 5.9m. Ask yourself what you can remove from any chart and it not lose any information:
On your discount vs. profit, it's not clear whether you're trying to show a correlation, or an indication of spread. You also have two significant outliers that massively increase the required area of your chart. I would consider categorising the average discount into Low/Medium/High, and then using a box-plot to see whether there was a lot of difference. You'd get much more visual separation, as the outliers would 'disappear' in a measure of spread.
Your filter settings on your Top N by Sales may not be correct. You've got Top 5, but the chart is only showing 1. This generally suggests you haven't accounted for the Order of Operations, where Top N is calculated before Dimension filters, I can't download your workbook to be more specific, though.
On your map, you've highlighted that Mongolia has the lowest sales at $ÂŁ3.1K, but the mark on the map is NULL (no sales), and if I mouse over Papua New Guinea, it has sales of $1.9K, suggesting that after you've labelled the chart you've changed the filters. Make sure any annotations are correct, because they're pretty static
On your Performance per Sub Category chart, I don't know what a red bar and a green bar mean: it's not the size of the bar; if it's the profit margin, you have the number labels displayed as red (which you appear to be using to mean "bad"?), but against green bars (and is a profit margin of 15.3% "bad"? in the industry I'm in we'd bite your hand off for that). Equally, the actual profit from phones and accessories is about the same, so if the message you want to convey is about good versus bad profit margin, I'm not sure this chart makes that point clear. The main message of a chart shouldn't be one of the secondary preattentive attributes of a chart.
On your heat map of delays, your commentary says delays have stayed relatively constant, but you've used a colour range that maximises the visual difference (a min to max colour range). Also, the colour isn't related to the value in the cell, and there's no legend to tell me what it actually is. If you wanted to convey a relatively constant amount, then perhaps a scale that was from an exceptional level (0 days) to a terrible level (10 days?), which would then support your message. Having a target might be more appropriate to show where you've done well. If this was a printed report, the colour bears no relation to the value displayed:
If I mouse over, I can deduce that it's the record count, but your user isn't told that anywhere.
Finally (sorry), on your shipping delay hotspots, you have the same colour meaning multiple different values, becuase you're using a dynamic legend. If you change the year, the dark red means anything between 5 and 7 days, but visually, Pakistan in 2023 (5 days) is as bad as Cambodia in 2025 (7 days), despite Cambodia having 40% longer delays, on average. For this chart, I would use a static range, where anything beyond my target, e.g. 3 days, is red. As it stands, if the business was transformed, and the longest delay was 1 day, that would still be considered "very bad red".
My two top tips would be: always know the question you're answering - if you want to say something is good or bad, you have to define good and bad, and these will often be static values or targets; when you think a chart is finished, look to see what you could remove without losing any information.
