Dashboards, Actions, Stories and Publishing in Tableau (Capstone)

Plan before you build

People read a screen in a Z-pattern: top-left first. Put the KPI cards across the top, the main trend or map in the middle, and the detail (tables, Top N) at the bottom. Keep to about 5–7 views and one colour meaning per dashboard (for example orange = loss everywhere).

Building the dashboard

KPI cards

Each card is a sheet: the measure on Text, Mark type Text, a formatted label such as ₹<SUM(Sales)> cr, and a small YoY calculation in the second line with ▲/▼ and colour.

import pandas as pd, numpy as np
df = pd.read_csv("superstore_sales.csv", parse_dates=["order_date"])
y = df.groupby(df.order_date.dt.year).agg(sales=("sales", "sum"), profit=("profit", "sum"), orders=("order_id", "nunique"), customers=("customer_id", "nunique"))
cur, prev = y.loc[2025], y.loc[2024]
for k in y.columns:
    unit = 1e7 if k in ("sales", "profit") else 1
    val = f"₹{cur[k] / unit:,.2f} cr" if unit > 1 else f"{cur[k]:,.0f}"
    print(f"{k:10s} 2025: {val:>12s}   YoY {cur[k] / prev[k] - 1:+.1%}")
print(f"margin     2025: {cur.profit / cur.sales:.1%}   (2024: {prev.profit / prev.sales:.1%})")

sales      2025:     ₹7.84 cr   YoY +19.8%
profit     2025:     ₹0.75 cr   YoY +22.1%
orders     2025:        1,232   YoY +13.5%
customers  2025:          623   YoY +5.6%
margin     2025: 9.6%   (2024: 9.4%)

Dashboard actions

Dashboard → Actions → Add Action:

Choose Run action on: Select / Hover / Menu and what happens when clearing the selection (show all values is the usual choice). Use Use as Filter (the funnel icon on a sheet) for the quick version.

Stories

A Story is a sequence of Story Points, each a captioned snapshot of a sheet or dashboard with its filters. Use it for a presentation: "1. Sales grew 20% in 2025 → 2. Growth came from Technology in the South → 3. Tables and Bookcases lose money → 4. Recommendation".

Publishing and sharing

Before publishing: give every sheet a clear title, hide unused sheets (right-click the dashboard tab → Hide All Sheets), add tooltips in plain language, set the default filters, and test on the phone layout.

Capstone 1: Superstore India executive dashboard

Build a one-page dashboard for the Head of Sales:

1. KPI row: Sales, Profit, Margin, Orders for the selected year with YoY change.
2. Trend: monthly Sales (bars) and Margin (line) on a dual axis.
3. Map: filled map of profit margin by state (diverging palette).
4. Category → sub-category bars with the drill-in set action; loss-makers in orange.
5. Top 10 customers table with a Top N parameter.
6. Filters: Year (single value), Region, Segment, applied to all sheets.

d25 = df[df.order_date.dt.year == 2025]
sub = d25.groupby("sub_category").agg(sales=("sales", "sum"), profit=("profit", "sum"))
sub["margin %"] = (sub.profit / sub.sales * 100).round(1)
print("2025 loss-making sub-categories:"); print(sub[sub.profit < 0].round(0))
st = d25.groupby("state").agg(sales=("sales", "sum"), profit=("profit", "sum")); st["margin %"] = st.profit / st.sales * 100
print("\nHighest and lowest margin states:", st["margin %"].idxmax(), f"{st['margin %'].max():.1f}%", "|", st["margin %"].idxmin(), f"{st['margin %'].min():.1f}%")

2025 loss-making sub-categories:
                  sales    profit  margin %
sub_category                               
Bookcases     3918268.0 -169018.0      -4.0
Tables        5297064.0 -286188.0      -5.0

Highest and lowest margin states: West Bengal 13.7% | Telangana 7.5%

Write three insight sentences under the dashboard, for example: what grew, what lost money, and what you recommend (discount caps on loss-making sub-categories, focus on the highest-margin states).

Capstone 2: Regional video-game sales (the class case study)

The same business case you solved in Excel in class: how do video-game sales vary across North America, Europe, Japan and other regions, and where are the opportunities?

vg = pd.read_csv("video_game_sales.csv")
print(vg.shape, "| years", vg.year.min(), "to", vg.year.max())
regions = ["na_sales", "eu_sales", "jp_sales", "other_sales"]
best_genre = vg.groupby("genre")[regions].sum().idxmax()
print("Best genre by region:", best_genre.to_dict())
vg["decade"] = (vg.year // 10 * 10).astype(str) + "s"
print((vg.groupby("decade")[regions].sum()).round(1))

(2400, 11) | years 2010 to 2025
Best genre by region: {'na_sales': 'Sports', 'eu_sales': 'Sports', 'jp_sales': 'Role-Playing', 'other_sales': 'Sports'}
        na_sales  eu_sales  jp_sales  other_sales
decade                                           
2010s      312.5     249.4     136.8        108.9
2020s      188.1     148.9      83.8         62.6

Tableau tasks: 1. Pivot the four regional sales columns into Region / Sales on the Data Source page (select the columns → Pivot), so Region becomes one dimension. 2. Top 5 games per region: Region on Rows, Name on Rows, SUM(Sales) on Columns, a Top 5 by Sales computed per region using RANK(SUM([Sales])) with Compute Using = Name, filtered ≤ 5. 3. Regional sales by decade: a calculated field STR(INT(YEAR([Year])/10)*10) + "s" (or ([Year] // 10) * 10 if Year is a number) and a stacked bar coloured by Region. 4. Best genre in each region: highlight table of Genre × Region with the maximum per region marked using WINDOW_MAX. 5. A stacked bar of regional distribution over time with a clear title, then assemble all four into a dashboard with a Region filter action.

Summary