My progress
Menu

Data Analyst / Engineer · Focus area 3: Analytics and dashboards

DT9: Dashboards and decisions

Goal: build dashboards that people who aren't technical read correctly and act on: start from the decision, define every metric in plain words, choose charts that fit the question, draw them honestly and accessibly, choose a BI tool sensibly, and protect people's privacy on the page.

You'll need: DT6, DT8, Module 16 (Responsive Websites), Module 79 (Privacy and Data Protection), and the SabziSathi practice kit.

Where this is used: a store manager glancing at yesterday's sales before opening; a hospital administrator checking bed occupancy; a district officer reviewing a scheme's reach. Most people meet data through a dashboard, so the dashboard decides whether the work behind it leads to good decisions.

How it differs: an analysis can explain its caveats in a meeting. A dashboard is read alone, quickly, by people who didn't build it, so its definitions, scales and warnings have to be on the page.

Time: reading about 2 hours · practice (the Try it tasks) about 3 hours · assignment about 3 hours.

Lessons

  1. 1Start from the decisionReading 15–25 minutes, plus practice✓
  2. 2Choosing the right chartReading 15–25 minutes, plus practice✓
  3. 3Honest and accessible chartsReading 15–25 minutes, plus practice✓
  4. 4BI tools and a single set of definitionsReading 15–25 minutes, plus practice✓
  5. 5Privacy, freshness and trust on the pageReading 15–25 minutes, plus practice✓

🛠️Assignment: A dashboard for SabziSathi's store managers

This is Project 3's preparation.

  1. Write the dashboard brief: readers, decisions, frequency and three headline metrics with definitions.
  2. Choose a chart or table for each question and say why.
  3. List every problem on the starter's dashboard: numbers, definitions, scales, accessibility and privacy.
  4. Rebuild the dashboard to read only from the marts, with zero-based bars, a table after each chart, definitions, suppression and a freshness line.
  5. Ask someone who isn't technical to read it and answer three questions from it; note where they hesitated.

Self-check before you finish:

✅Module check

Explain in a few sentences, in your own words:

What makes a dashboard trustworthy for people who aren't technical?

Show a good answer

A good answer mentions: starting from the reader and their decision; a few metrics, each defined in plain words with what's included and the unit; comparisons; charts matched to the question; zero-based bars, shared scales and restrained colour; titles and tables for accessibility; definitions kept in one place for every tool; small counts hidden and no personal data; a freshness line; and clear notes on incomplete, delayed or synthetic data.

📚Official references

Official documentation and guidance for this module's topics, chosen on 2 November 2026. Tools, laws and standards change, so check the current version before you rely on a detail. If a link has moved, search the same site.

📝Module quiz

10 questions · 10 marks each · 100 marks in all · pass mark 70% · retry as often as you like ·

1. A manager asks for "a sales dashboard". What's the best first step?

Why each option is right or wrong

  • “Build every chart the data allows, so nothing is missing” (Incorrect): Too many charts hide what matters and nobody reads them.
  • “Copy a dashboard from another company's website” (Incorrect): Their decisions aren't yours.
  • “Choose a BI tool, then decide what to show” (Incorrect): The tool comes after the purpose.
  • “Ask who reads it and which decisions it should support” is correct.

    The decision decides the metrics, the charts and how often it refreshes.

    • Then: three to five metrics, each tied to a decision.
    • Also ask: how often it's read.

📖 Revise: Lesson 1: Start from the decision

Question ID sp-data-09-q1 · Report a mistake in this question
2. Finance and marketing both report "revenue" for last month, ₹48 lakh and ₹53 lakh. What's the most likely cause?

Why each option is right or wrong

  • “One of the teams made an arithmetic mistake” (Incorrect): Possible, but different definitions are far more common.
  • “The warehouse lost data between the two reports” (Incorrect): Both used the same warehouse; the figures differ by design.
  • “Marketing's dashboard refreshes more often” (Incorrect): A month's figure doesn't change with refresh frequency once the month is over.
  • “Different definitions, such as whether refunds are left out” is correct.

    The same name with different rules gives different numbers. Define it once and show the definition on both.

    • Fix: one definition, shown on both dashboards.
    • Store it in: the warehouse marts.

📖 Revise: Lesson 1: Start from the decision

Question ID sp-data-09-q2 · Report a mistake in this question
3. Which chart best shows how daily orders changed over 90 days?

Why each option is right or wrong

  • “A pie chart with one slice per day” (Incorrect): Ninety slices can't be compared.
  • “A 3D bar chart with one bar per day” (Incorrect): 3D distorts lengths, and 90 bars are crowded.
  • “A gauge showing the 90-day average” (Incorrect): It hides the change the question is about.
  • “A line chart with days along the bottom” is correct.

    Lines show change over time clearly, even over many points.

    • Add: a 7-day moving average to smooth daily noise.
    • Label: the axis and the latest value.

📖 Revise: Lesson 2: Choosing the right chart

Question ID sp-data-09-q3 · Report a mistake in this question
4. A dashboard says "average delivery time: 28 minutes". Customers still complain about late orders. What should be added?

Why each option is right or wrong

  • “A larger font for the average” (Incorrect): Size doesn't add information.
  • “The average for last month as well” (Incorrect): Two averages still hide the slow deliveries.
  • “A pie chart of deliveries by rider” (Incorrect): It doesn't show how long deliveries take.
  • “A histogram or percentiles to show the slow tail” is correct.

    An average hides spread; the 90th percentile or a histogram shows how many deliveries take far longer.

    • Show: the share of deliveries over 45 minutes, for example.
    • Keep: the average, with its spread.

📖 Revise: Lesson 2: Choosing the right chart

Question ID sp-data-09-q4 · Report a mistake in this question
5. A bar chart of two stores' revenue starts its axis at ₹9 lakh. Store A has ₹9.5 lakh and B ₹10 lakh. What will readers think?

Why each option is right or wrong

  • “That the two stores earned about the same, which is roughly right” (Incorrect): The cut axis makes them look very different.
  • “That the chart is wrong and should be ignored” (Incorrect): Most readers trust the picture.
  • “Nothing, since readers look only at the labels” (Incorrect): Readers compare bar lengths first.
  • “That B earned about twice A, though it's about 5% more” is correct.

    From ₹9 lakh, A's bar is 0.5 and B's is 1.0 tall. Bars should start at zero.

    • Fix: start bars at zero.
    • Lines: may start above zero if clearly labelled.

📖 Revise: Lesson 3: Honest and accessible charts

Question ID sp-data-09-q5 · Report a mistake in this question
6. A dashboard marks good stores green and poor ones red, with no labels. What's the accessibility problem?

Why each option is right or wrong

  • “Green and red are too bright on screens” (Incorrect): Brightness can be adjusted; meaning carried by colour alone is the problem.
  • “Screen readers read colours aloud too slowly” (Incorrect): Screen readers don't read colours of chart elements at all without text.
  • “Red should mean good in financial dashboards” (Incorrect): Conventions vary; the issue is relying on colour alone.
  • “Some readers can't tell red from green” is correct.

    About 1 in 12 men has a red-green difference. Add labels or patterns, or use a safer palette.

    • Fix: labels or patterns as well as colour.
    • Check: a colour-blind simulator on the page.

📖 Revise: Lesson 3: Honest and accessible charts

Question ID sp-data-09-q6 · Report a mistake in this question
7. Three dashboards each compute "same-store sales" with their own SQL, and the numbers drift apart. What's the best fix?

Why each option is right or wrong

  • “Ask every analyst to copy the SQL from the latest dashboard” (Incorrect): Copies drift again.
  • “Delete two of the dashboards” (Incorrect): The other teams still need the number.
  • “Round all three figures to one decimal place” (Incorrect): Rounding hides the difference without fixing it.
  • “Define it once for every report to use” is correct.

    One definition in the warehouse marts or a semantic layer, used by every tool, makes every report agree.

    • Where: dbt marts, or a semantic layer such as LookML or dbt's.
    • Then: certify one dataset.

📖 Revise: Lesson 4: BI tools and a single set of definitions

Question ID sp-data-09-q7 · Report a mistake in this question
8. A team picks a BI tool after a great demo. Later they find every viewer needs a paid licence. What did they skip?

Why each option is right or wrong

  • “Checking that the tool has a dark mode” (Incorrect): Nice, but not why the budget failed.
  • “Asking the vendor for a second demo” (Incorrect): Another demo wouldn't change the licence terms.
  • “Trying the tool on a laptop first” (Incorrect): Licences apply however the tool is tried.
  • “Pricing licences for their real users” is correct.

    Licence models differ between viewers and editors. Counting real users first avoids surprises.

    • Count: editors and viewers separately.
    • Compare: licence models for those counts.

📖 Revise: Lesson 4: BI tools and a single set of definitions

Question ID sp-data-09-q8 · Report a mistake in this question
9. A dashboard shows customers by area. One area has 2 customers. What should it show?

Why each option is right or wrong

  • “2, since hiding numbers makes the dashboard less accurate” (Incorrect): Accuracy matters, but a count of 2 can point to individuals.
  • “0, so the area doesn't stand out” (Incorrect): That's false.
  • “The two customers' names, so the count can be checked” (Incorrect): That puts personal data on the dashboard.
  • “"Fewer than 5", and check no total reveals it” is correct.

    Suppressing small counts protects people; a total plus the other cells can reveal the hidden one.

    • Threshold: often 5 or 10.
    • Watch: totals that let a hidden cell be worked out.

📖 Revise: Lesson 5: Privacy, freshness and trust on the page

Question ID sp-data-09-q9 · Report a mistake in this question
10. Last night's pipeline failed. What should this morning's dashboard do?

Why each option is right or wrong

  • “Show zeros for yesterday, since no data arrived” (Incorrect): Zeros look like a real result and mislead.
  • “Copy the day before's numbers into yesterday” (Incorrect): That invents data.
  • “Remove the dashboard until the pipeline is fixed” (Incorrect): People lose access to the last good numbers too.
  • “Keep the last good version, marked delayed” is correct.

    Readers still see trusted numbers and know yesterday isn't in yet.

    • Show: "Data delayed: yesterday not yet included".
    • Avoid: zeros or copied numbers.

📖 Revise: Lesson 5: Privacy, freshness and trust on the page

Question ID sp-data-09-q10 · Report a mistake in this question

ProLearners Assist