Tableau Connections, Views & Calculated Fields - Interview Revision Guide
Why can two Tableau charts built from the same sales table show different profit margins? Usually, Tableau is not wrong - the analyst has connected, aggregated or calculated at the wrong level of detail.
- Connections decide how Tableau reads data - live for freshness, extract for speed and portability.
- Views are built by placing dimensions and measures on shelves, then controlling marks, filters, aggregation and sort.
- Calculated fields create new fields using formulas - but the grain of the calculation matters more than the syntax.
- Use relationships for flexible multi-table analysis; use joins when you need a fixed row-level table.
- The most common margin formula in Tableau is
SUM([Profit]) / SUM([Sales]), not average of row-level profit ratios. - A strong Tableau answer moves in this order: business question - data connection - data model - view design - calculation - validation.
Big Picture
Tableau is easiest when you see it as a pipeline: connect to trustworthy data, shape the model, build a view, add calculations, then validate the business story before dashboarding.
Core Explanation
The core idea: Tableau turns data fields into visual encodings. Dimensions usually slice the view, measures usually get aggregated, and calculated fields let you create business logic that does not already exist in the source data.
1. Connections - how Tableau reads the data
A connection is Tableau's link to a data source such as Excel, CSV, SQL database, cloud warehouse, Google Sheets or a published Tableau data source.
For Indian business analysis, a practical example is connecting Tableau to RBI Database on Indian Economy exports for macro indicators and NSE bhavcopy CSVs for market data. The connection decision is not cosmetic: a one-time MBA project can use extracts, while an analyst tracking live risk or treasury exposure would prefer governed, refreshed sources.
2. Views - how Tableau turns fields into charts
A view is the visual result created in a worksheet after fields are placed on shelves and encoded through marks. The most important pieces are:
A good view starts from a business question. "Which region sells the most?" needs a bar chart. "How has monthly revenue moved?" needs a line chart. "Which categories are high sales but low profit?" needs a scatter plot or 2x2 performance view.
3. Calculated fields - how Tableau creates business logic
A calculated field is a new field created from a formula using existing fields, constants, functions or logic. Use it when the source data does not directly contain the metric or classification you need.
Worked Example - the Profit Margin Trap
Suppose your data has two orders:
If you calculate profit ratio row by row and average it, you get (20 percent + 10 percent) / 2 = 15 percent. But the correct business margin is total profit / total sales = ₹110 / ₹1,000 = 11 percent.
For ratios in Tableau, prefer aggregate numerator divided by aggregate denominator: SUM([Profit]) / SUM([Sales]).
Metrics That Show a Tableau Workbook Is Ready
When you are asked how you would evaluate a Tableau workbook, do not say "it should be clean". Use measurable checks.
Definitions
- Data connection: The link Tableau uses to read data from a file, database, cloud app or published source.
- Live connection: A connection where Tableau queries the underlying source whenever the view requires data.
- Extract: A compressed snapshot of source data stored in Tableau's Hyper format for faster analysis.
- View: A worksheet visual created by placing fields on shelves and encoding them through marks.
- Calculated field: A new Tableau field created from a formula using fields, functions, constants or logical conditions.
- Aggregation: The summarisation of row-level values into measures such as sum, average, minimum, maximum or count.
Lenovo: Making Tableau Work Across Complex Business Functions
Lenovo has used Tableau to support self-service analytics across business functions where teams need fast, governed views of complex operational data.
Situation: A global technology company such as Lenovo works with high-volume data across sales, supply chain, finance, products and geographies. The challenge is not only "make a chart"; it is to let different teams answer their own questions without creating multiple conflicting versions of truth.
The move: Tableau becomes valuable when the company standardises data sources, publishes governed datasets, and lets business users build views on top of trusted fields. Connections handle the source logic, views translate business questions into visual analysis, and calculated fields capture metrics such as margin, attainment, ageing or product mix.
Outcome or lesson: The primary driver is governed self-service analytics - business teams can explore without waiting for every report from central IT. Supporting drivers include published data sources, reusable calculations, role-based access and a shared visual language across functions.
The strategic takeaway: Tableau scales only when visual freedom is supported by data governance. Otherwise, it becomes a faster way to create inconsistent reports.
How AI Changes Tableau: Connections, Views & Calculated Fields
AI is changing Tableau less by replacing analysts and more by reducing the friction between a business question and a usable first analysis.
Student workflow: Load a sample data dictionary, the company annual report and your Tableau notes into NotebookLM. Ask it to generate five business questions, the required fields, likely calculated fields and validation checks. Then build one view in Tableau and verify every formula manually.
Interview Relevance
"You have sales data in Excel and customer data in a SQL database. How would you connect them in Tableau, build a regional profitability view and create a margin calculation?"
In a Tableau answer, say the word grain. It signals that you understand why the same formula can behave differently at row level, aggregate level and view level.
Common Mistake
The single biggest mistake is creating calculated fields without checking the grain - especially averaging ratios instead of dividing aggregated totals. It costs candidates because the chart may look polished while the business number is wrong. One-line fix: always ask, "Is this calculation row-level, aggregate, or view-dependent?" before trusting the result.
What to Revise Next
Next, move from basic Tableau construction to advanced analytical control. Revise these in order: