Working With Scales And Axes

Why this matters

As a Business Analyst, you often present trends, comparisons, and distributions. The same data can look clear or misleading depending on the scale and axes you choose. Correct choices help stakeholders see real change, compare fairly, and trust your insights.

• Forecast review: choose time and value scales that reflect seasonality without exaggerating small bumps.
• Executive dashboards: align axes so KPI movements are comparable across charts.
• Experiment results: center around zero for lift/impact charts to highlight positive vs. negative effects.

Concept explained simply

A chart is a mapping from data (domain) to pixels on the screen (range). Scales define this mapping. Axes show viewers how to read that mapping with ticks, labels, and gridlines.

Mental model

Think of a scale as a translator: it takes data values (like dates or dollars) and translates them into positions on the screen. The axis is the subtitle that explains the translation so everyone can follow.

Core concepts

• Scale types:
  • Linear: equal steps in data are equal steps visually. Default for most continuous data (revenue, counts).
  • Logarithmic: compresses large values; equal ratios look equal. Use for multiplicative growth or wide ranges (10–10,000). Avoid zeros/negatives.
  • Time/Date: handles days, months, quarters; respects irregular intervals.
  • Ordinal/Categorical: for named categories; order matters if natural (Mon–Sun), else treat as unordered.
  • Band/Point: for bar/column charts (band allocates width, point for discrete markers).
  • Diverging: emphasizes deviations around a meaningful center (e.g., 0, target, budget).
• Domain: data min–max (or custom limits). Consider outliers: include with context, transform, or clip clearly.
• Range: pixel space; set margins so labels don’t collide.
• Ticks and gridlines: 4–6 major ticks per axis is usually enough; use light gridlines to aid reading without clutter.
• Zero baseline:
  • Bars/columns encode length; start at zero to avoid exaggeration.
  • Lines encode position; zero baseline optional—prefer zoom to trend detail when appropriate.
• Units and formats: show units in axis title or labels (USD, %). For large numbers, use K/M/B consistently.
• Outliers: consider log scale, breaking the axis with a clear marker, or annotating outliers.
• Dual axes: risky. Only use when series share a common baseline or after normalizing (indexing). Otherwise, separate charts.

Worked examples

Example 1: Monthly revenue 2023–2025

Data: monthly revenue from $120K to $210K with seasonal peaks.

• X: Time scale (months Jan 2023–Dec 2025).
• Y: Linear $ scale, domain slightly padded (e.g., $110K–$220K) to avoid clipping labels.
• Ticks: X every quarter; Y 5 ticks (e.g., 120K, 150K, 180K, 210K).
• Bars or line? Line for trend, bars for discrete monthly reporting. If bars, include zero baseline.

Example 2: API response times with outliers

Data: Most requests 100–400 ms; rare spikes at 5,000 ms.

• Option A (overview): Y linear 0–6,000 ms; annotate outliers; use median and IQR to summarize.
• Option B (detail): Y linear 0–800 ms for main chart; add a small inset or a broken-axis indicator for outliers.
• Option C (if multiplicative patterns): Y log scale; make sure viewers know zero/negatives can’t be shown.

Example 3: Campaign impact vs. baseline

Data: Weekly lift from A/B test, ranging −4% to +9%.

• Y: Diverging linear scale centered at 0% (e.g., −10% to +10%).
• X: Time scale (weeks).
• Gridline at 0% emphasized; color scheme switching at 0 (e.g., red below, green above).

How to choose scales and axes (step-by-step)

1. Identify data type: continuous, time, categorical, or ratios across magnitudes.
2. Pick a scale: linear (default), time, ordinal/band, or log/diverging if your story requires it.
3. Set domain: min–max with padding; handle outliers intentionally (annotate, transform, or separate).
4. Define tick strategy: 4–6 major ticks; use helpful units (%, K/M/B).
5. Confirm integrity: zero baseline for bars; avoid misleading truncation.
6. Test readability: mobile-friendly labels, rotated only if needed, abbreviate numbers.

Quick checklist

• [ ] Is the scale type appropriate for the data?
• [ ] Is the domain set to include important points (zero, target, center)?
• [ ] Are tick marks few but informative (4–6)?
• [ ] Are units/percentages clearly labeled?
• [ ] If bars: does the axis start at zero?
• [ ] If log scale: are there no zeros/negatives and is it clearly indicated?
• [ ] Are outliers addressed (annotated, inset, or transformed)?
• [ ] Did you avoid confusing dual axes (or normalize if used)?

Practice: Exercises 1–2

Try these before viewing solutions. Then compare with the official answers provided below each exercise.

Exercise 1 (mirror: ex1)

Pick the right scale, domain, and axis choices:

1. Quarterly sales by region (bars), values 0.8–2.3M.
2. Daily active users over 2 years, seasonal peaks and a big spike (marketing campaign).
3. CSAT change vs. last month (−5 to +5 points).

Decide: scale type, zero baseline need, tick strategy, and any outlier handling.

Exercise 2 (mirror: ex2)

Debug this chart description: "We used a column chart with Y starting at 70, monthly data points omitted for quiet months, and a second Y-axis for conversion rate alongside revenue." List at least three issues and propose fixes.

Common mistakes and how to self-check

• Truncated bar axes: bars not starting at zero exaggerate differences. Self-check: does bar length reflect the whole value from zero?
• Too many ticks: crowds labels. Self-check: 4–6 major ticks only.
• Wrong number format: mixing K/M/B. Self-check: pick one convention and stick to it.
• Using log without telling viewers: causes confusion. Self-check: add a note or axis title indicating log scale.
• Dual-axis confusion: unrelated units on the same plot. Self-check: can someone misread correlation? If yes, split or normalize.

Mini challenge

You must present ad spend (0.9–1.2M) and ROAS (0.7–1.5) trends for the last 18 months.

• Design 1: Single chart? If so, how will you avoid dual-axis confusion?
• Design 2: Two small-multiple charts? What scales and ticks will you use?

Who this is for

• Business Analysts preparing dashboards and reports.
• Anyone turning raw metrics into stakeholder-ready visuals.

Prerequisites

• Basic chart types (bar, line, scatter).
• Comfort with metrics, units, and percent change.

Learning path

1. Working with Scales and Axes (this lesson).
2. Color and Emphasis for Clarity.
3. Choosing Chart Types for Questions.
4. Designing Dashboards that Compare Fairly.

Practical projects

• Rebuild a KPI dashboard: align axes across monthly charts, set consistent tick counts, and center diverging charts at zero.
• Outlier story: create two versions of a latency chart (overview vs. detail) and explain the trade-offs.
• Indexing demo: show three markets indexed to 100 to compare growth paths fairly.

Next steps

• Apply the checklist to one of your current charts and iterate.
• Take the quick test below to confirm understanding.

Quick test info

Anyone can take the test for free. Only logged-in users will have their progress saved.