📊 Complete Chart Gallery

Compare discrete categories where exact values matter. The human eye excels at comparing bar lengths, making this the most accurate chart for comparing quantities across groups.

Always start the Y-axis at zero — truncated axes exaggerate differences and mislead viewers.

Rotated bar chart that accommodates long category names while maintaining readability. Ideal for "Top N" lists where natural reading order (top to bottom) matches ranking order.

Sort bars by value (not alphabetically) unless there's a specific reason — pre-sorted data reduces cognitive load.

Compare multiple data series across categories by placing bars side by side. Best for before/after comparisons, A/B testing results, or comparing 2-4 related measures.

Limit to 2-4 series maximum — more than 4 becomes visually cluttered and hard to interpret.

Display part-to-whole relationships while also comparing totals across categories. Each bar segment represents a component; the full bar shows the total.

Only the bottom segment is easy to compare accurately — put your most important category there, or use 100% stacked to focus purely on proportions.

Colorblind caution: Limit to 3-4 segments with high-contrast colors. Add a clear legend and consider direct segment labels.

Compare entities across multiple dimensions simultaneously. The shape formed reveals the "profile" of each entity, making it easy to spot strengths and weaknesses at a glance.

Normalize all axes to the same scale (0-100) — otherwise, the visual shape becomes meaningless.

Shows how an initial value increases and decreases through a series of intermediate values to reach a final value. Essential for financial storytelling — revenue bridges, cost breakdowns, and variance analysis.

Color-code positive (green) and negative (red) bars. Include running totals as floating connectors between bars.

One of Knaflic's "12 most used visuals." Connects two time points with lines to show change direction and magnitude. Instantly reveals which items rose, fell, or stayed flat — perfect for before/after comparisons.

Highlight the lines you want to emphasize and gray out the rest. Direct-label both ends — no legend needed.

Stephen Few's fundamental rule: "Tables interact with our verbal system; graphs with our visual system." Use tables when readers need to look up specific values, compare precise numbers, or scan for data points. Use graphs when showing patterns or trends.

Right-align numbers, left-align text. Use conditional formatting as a heatmap to add pattern recognition while preserving precision.

A powerful technique from Few and Kirk: repeat the same chart for different categories with consistent scales. The human eye instantly spots differences in shape across panels without needing a legend.

Keep all axes identical across panels. Compact design — each panel should be small but readable. Label panels clearly.

Track how values change over continuous time. The connected line implies continuity between points, making it ideal for metrics measured at regular intervals where values exist between observations.

Use consistent time intervals and consider adding a trend line or moving average to smooth noise and reveal underlying patterns.

Line chart with the area beneath filled, emphasizing magnitude and volume over time. The filled area makes totals more visceral and is excellent for showing cumulative quantities.

Use stacked area charts sparingly — only the bottom layer is easy to read accurately. Consider small multiples instead.

Combines bars and lines on the same chart — typically bars for volume/quantity and line for rates/percentages. Allows comparison of two related but different-scale metrics.

Use a secondary Y-axis for the line metric. Consider if two separate charts might be clearer — dual axes can mislead.

Group continuous data into bins and show frequency. Instantly reveals whether data is normal, skewed, bimodal, or has outliers — essential for understanding data quality before analysis.

Experiment with bin sizes — too few bins hide patterns, too many create noise. Start with √n bins (where n = data points).

Display individual points spread horizontally to avoid overlap, revealing both distribution shape and actual data density. Perfect when you need to show that each point represents a real observation.

Perfect for showing stakeholders that each dot is a real customer, employee, or transaction — not just an abstract statistic.

Combines box plot statistics with a kernel density estimate shown as a mirrored shape. Width shows density at each value, revealing multimodal distributions that box plots hide.

Violin plots can reveal bimodal distributions (two peaks) that box plots completely miss — always check before choosing.

Stacked density curves that create a "mountain range" effect. Visually striking way to compare many distributions, often showing change over time (like daily patterns across months).

Order ridges meaningfully (chronologically, by median, by variance) to tell a clear story about how distributions shift.

Shows median, quartiles, and outliers at a glance. The box contains 50% of data (IQR), whiskers extend to 1.5×IQR, and dots beyond are outliers. The gold standard for statistical comparison.

Add individual points (jittered) alongside box plots when sample sizes are small enough to show every observation.

Plot two numeric variables against each other to reveal correlations, clusters, and outliers. Each point represents one observation, with position encoding two different measurements.

Add a trend line to quantify relationships, but always check the R² value — correlation doesn't imply causation.

Scatter plot with a third variable encoded as bubble size. Famous example: Hans Rosling's wealth vs health chart with population as bubble size. Powerful for strategic quadrant analysis.

Scale bubble areas (not radius) to value for accurate perception. Consider adding quadrant lines to create BCG-style matrices.

Each vertical axis represents a variable. Lines connecting across axes show individual records. Patterns emerge when many similar records create visible "bands" across dimensions.

Brushing (selecting lines on one axis to highlight across all) is essential for exploring patterns — static versions lose most power.

Nodes represent entities; edges (lines) represent relationships. Force-directed layouts automatically cluster related entities. Essential for social networks, org charts, and dependency mapping.

Node size can encode importance (degree centrality), and edge thickness can show relationship strength — don't waste these encodings.

The most recognized but most misused chart. Humans struggle to compare angles accurately. Only effective when showing one dominant slice (60%+) or comparing to 50% — otherwise, use a bar chart.

If your first instinct is pie chart, pause and ask: Would a bar chart tell this story better? Usually, yes.

Colorblind caution: Adjacent slices may be hard to distinguish. Always include a legend with patterns or direct labels.

Pie chart with center removed — use that space for a KPI total, label, or icon. The hollow center actually improves perception slightly and provides prime real estate for context.

The center is premium space — use it for the total, a delta (↑12%), or a meaningful icon. Don't waste it.

Colorblind caution: Same limitations as pie charts. Use direct labels on segments rather than relying solely on legend colors.

Nested rectangles where area represents value. Efficiently displays many categories in limited space. Supports hierarchies — categories can contain sub-categories with the same visual logic.

Add a second variable as color intensity (e.g., size = revenue, color = growth rate) to pack more insight per pixel.

Colorblind caution: Adjacent rectangles may blend. Use high-contrast color schemes or add text labels to each cell.

A 10×10 grid (100 squares) where colored squares represent percentages. Humans understand "63 out of 100" better than a 227-degree slice. More accurate than pie charts for comparing close values.

Use icons instead of squares for infographics (100 person icons, with 37 highlighted for "37% affected").

Concentric rings showing hierarchy levels — inner ring is the parent, outer rings are progressively detailed children. Interactive versions let users click to zoom into sub-hierarchies.

Sunbursts shine when interactive — static versions can be confusing. Consider treemap for print materials.

Colorblind caution: Multiple concentric rings with similar colors are hard to distinguish. Use tooltips, labels, or interactive highlighting.

Visualize how quantities decrease through sequential stages. Each horizontal bar represents a stage, with width proportional to value. Instantly reveals where the biggest drop-offs occur.

Add conversion percentages between stages. Color-code by stage health (green = good conversion, red = problem area).

Color encodes values across a matrix of two categorical dimensions. Instantly reveals patterns — like sales peaks on weekday evenings or correlation clusters in variables.

Use sequential color scales (light → dark) for single values; diverging scales (red ← white → green) when zero is meaningful.

Colorblind caution: Red-green scales are problematic. Use blue-orange or sequential single-hue scales (light blue → dark blue) for accessibility.

Flow diagram where width represents quantity. Shows how inputs split, merge, or flow through stages — perfect for tracking customers through funnels or energy through systems.

Limit to 3-5 stages — more stages make flows hard to follow. Use color to encode categories (paid vs organic traffic).

Shows flows between categories arranged in a circle. Arc length shows total volume; chord thickness shows flow size between pairs. Elegant for showing all pairwise relationships at once.

Interactive hovers are essential — static chord diagrams can be overwhelming. Consider Sankey for linear flows.

Colorblind caution: Many intersecting colors make this challenging. Ensure interactive highlighting and clear labels. Consider Sankey as a more accessible alternative.

Stacked area chart with a flowing, centered baseline. Creates an organic "river" aesthetic showing how multiple series evolve together. Sacrifices precision for visual storytelling impact.

Use for emotional impact in presentations, but provide a supporting table or simple chart when stakeholders need exact numbers.

A semicircular arc showing current value against a maximum. Intuitive for single KPIs but uses space inefficiently. Best for high-visibility dashboard tiles where simplicity trumps density.

Consider bullet charts instead — they show the same info in less space with better precision. Reserve gauges for when visual impact matters most.

Stephen Few's superior alternative to gauges. Shows actual value (bar), target (line), and qualitative ranges (background bands) in a compact horizontal space. Fits many KPIs in one view.

Stack multiple bullet charts vertically for a compact KPI dashboard. Add % of target as text for quick scanning.

Knaflic's advice: "When you have just a number or two to share, simple text can be a great way to communicate." A prominent number with context (vs. target, vs. prior) delivers maximum impact with minimum cognitive load.

Include directional indicators (↑↓) and color coding. Add context: "$4.2M Revenue" means little; "$4.2M Revenue (+12% vs LY)" tells a story.

Compact trend lines embedded in tables or KPI cards. Few and Wexler recommend sparklines for dashboards where space is precious — show 12 months of trend in the width of a number.

Add markers for start, end, min, max points. In Power BI, use the Matrix visual with sparkline measure or the new Sparklines feature.

Interactively break down a measure across multiple dimensions. AI can suggest which dimension to drill into next based on what explains variance best. Essential for root cause analysis.

Use the AI splits feature (lightbulb icon) to let Power BI suggest which dimension explains the most variance at each level.

AI automatically identifies which factors most influence a target metric. Shows relative impact of different variables with statistical confidence. Reveals hidden drivers in your data.

Switch between "Key influencers" and "Top segments" views. The segments view clusters similar records to find high/low performing groups.

AI automatically generates natural language summaries of your data. Highlights trends, outliers, and key takeaways without manual analysis. Updates dynamically as data changes.

Customize which insights to include and how they're phrased. Combine with other visuals for context — the narrative can describe what's shown in adjacent charts.

Visually appealing but analytically weak. Word size is hard to compare accurately, position is arbitrary, and long words appear more prominent regardless of frequency. Hides actual counts that matter for decisions.

If your first instinct is word cloud, pause and ask: Would a bar chart of top 10 terms tell this story better? Almost always, yes. Use word clouds only as visual hooks, never as evidence.

Horizontal stacked bars showing sentiment distribution across categories. Essential for customer feedback, reviews, and social listening dashboards.

Use AI Builder or Azure Cognitive Services to classify sentiment, then aggregate in Power BI. Always show sample size per category.

Shows NPS on a -100 to +100 scale with zones for Detractors (red), Passives (yellow), and Promoters (green). The score = % Promoters minus % Detractors.

Always show the breakdown (% Promoters, Passives, Detractors) alongside the NPS. Track trends over time rather than obsessing over point-in-time scores.

Diverging stacked bars centered on neutral. Shows distribution of agree/disagree responses across survey questions. Instantly reveals consensus vs. polarization.

Center on neutral to make positive/negative comparison intuitive. Sort questions by most positive or most polarized depending on your story.

Radial tree showing hierarchical relationships from a central concept. Perfect for brainstorming outputs, topic taxonomies, and showing how ideas branch from core themes.

Keep to 2-3 levels deep for readability. Use colors to distinguish branches and consider adding metrics as node sizes.

Traditional top-down tree structure for organizational charts, decision trees, and parent-child relationships. Clear visual hierarchy with connecting lines.

Add metrics to nodes (team size, budget) to combine structure with quantitative data. Consider collapsible levels for large hierarchies.

The foundation of Statistical Process Control (SPC). Shows measurements over time with Upper Control Limit (UCL), Lower Control Limit (LCL), and center line. Points outside limits or patterns signal special cause variation requiring investigation.

Control limits are NOT specification limits. They're calculated from your data (±3σ). Points within limits but showing trends, runs, or patterns also indicate problems.

Combines bars (individual values) with a cumulative line. Based on the 80/20 rule — typically 20% of causes create 80% of problems. Focus improvement efforts on the tall bars first for maximum impact.

Always sort bars descending by value. The cumulative line should reach 100%. Draw a horizontal line at 80% to quickly see which categories comprise the "vital few."

Simpler than control charts — plots data over time with a median line. Look for runs (consecutive points above/below median), trends (6+ points continuously increasing/decreasing), and shifts. No statistics needed, just counting.

Count runs (crossings of median + 1). Too few runs = data is shifting. Too many = data is oscillating. Use run chart rules before graduating to control charts.

Shows how well your process fits within specification limits. Distribution curve overlaid on spec limits (LSL/USL). Cp measures spread (can you fit?), Cpk measures centering (are you fitting?). Target Cpk ≥ 1.33 for capable processes.

Cpk = 1.0 means 3σ capability (99.73% within spec, 2700 ppm defects). Cpk = 1.33 means 4σ (63 ppm). Cpk = 2.0 means 6σ (0.002 ppm). Always report both Cp AND Cpk.