If Chapter 2 was about what charts are for, this chapter is about how to build them so they look, feel, and behave like McKinsey charts. The goal is not art. The goal is disciplined communication: every line and label on the page exists to help a senior client grasp the message in seconds.
You will see a consistent theme: start from the main message, then make a small number of deliberate choices about chart type, scale, color, and labeling. Do that well, and even under time pressure your charts will look coherent, professional, and “of a piece” with the Firm’s standard.
3.1 Zelazny’s legacy: simplicity, clarity, and focus on the main message
Gene Zelazny’s influence on the Firm can be reduced to a deceptively simple idea: a chart is a sentence, not a picture of data.
From that idea come three principles you should internalize.
First, simplicity. Show the minimum necessary to make the point. If one series is enough, do not show three. If the last five years are enough, do not show fifteen. Simplicity is not an aesthetic preference; it is respect for your audience’s time and attention.
Second, clarity. The reader should understand what they are looking at without having to decode anything. Axes are labeled in plain business language, units are obvious, categories are ordered in a meaningful way, and the chart type is familiar. Clarity also means avoiding clever but obscure visuals. If a smart, rushed client has to pause and think, you have probably chosen the wrong form.
Third, focus on the main message. The main message is not implicit; it is explicit. It is written in the title as a complete sentence and reinforced by how the chart is constructed. Zelazny’s test was straightforward: if you cover up the data and only read the title, do you still get the point? If not, your title is too weak or too neutral.
A “Zelazny-compliant” chart, in practice, passes three quick tests:
- You can state its message in one sentence.
- You can explain its structure in one breath.
- You can remove at least one visual element without losing the point.
If you keep removing and the message gets sharper, you are moving in the right direction.
3.2 Choosing the right chart: a practical decision tree
Most confusion in charting comes from starting with the tool (“I’ll insert a line chart”) instead of the question. A simple mental decision tree will save you time and rework.
Ask yourself, in order:
- Is time central to the question?
- Yes → you are in the world of trends: line charts, column-over-time, area charts.
- No → move to the next question.
- Is the question mainly about size or ranking across categories at one point in time?
- Yes → you are in cross-section comparisons: bars, columns, dot charts.
- No → move on.
- Is the question about how a total breaks into parts?
- Yes → you are in composition: stacked bars/columns, 100% stacked, Mekko, occasionally pies/doughnuts.
- No → next.
- Is the question about variation or spread, not just the average?
- Yes → you are in distributions: histograms, box/range plots, occasionally population-type charts.
- No → next.
- Is the question about how two (or more) variables move together, or which “bucket” things fall into?
- Yes → you are in relationships and segmentation: scatter plots, bubble charts, quadrants, heat maps, portfolio maps.
- No → next.
- The question is “How did we get from A to B?”
- Yes → you are in bridges and contributions: waterfalls, decomposed stacks.
In most cases, one of these branches will clearly dominate. If you find yourself trying to satisfy three branches at once—trend plus composition plus distribution—split the story across multiple pages. One chart, one job.
You do not need an elaborate flowchart taped to your monitor. You do need to build the habit of asking, every time: “Which of these six questions am I answering?” The later chapters on chart families will plug into this same structure.
3.3 Scales, axes, and baselines: avoiding distortion
Nothing undermines credibility faster than a chart that feels “off,” even if unintentionally. Most of that feeling comes from how you treat scales, axes, and baselines.
For bar and column charts, the rule is simple: start the axis at zero. Bars and columns encode value by length; truncating the axis exaggerates differences and is visually misleading. If starting at zero would compress interesting variation (e.g., all values are between 93% and 97%), consider a different chart type or use a table instead of distorting the scale.
For line charts, you have more flexibility because the story is about change over time rather than absolute magnitude. You may not always need to start at zero, but you still must:
- Make the range explicit through clear axis labels.
- Avoid hyper-compressed scales that turn trivial noise into dramatic swings.
- Keep the same scale when comparing multiple similar charts across a sequence of slides.
Consistency is critical. If “Revenue growth” is shown on three slides, the y-axis definition should not silently change from “0–20%” to “5–15%” unless there is a compelling reason, called out in a note.
You also control units and indexing. Indexing to 100 in a base year is powerful for showing relative change across series, but it must be clear in the axis label and often reiterated in the chart title (“Indexed to 100 in 2020”). For currency, always specify whether you are in millions or billions, and keep that convention consistent across the deck.
Lastly, pay attention to aspect ratio and gridlines. A chart that is too wide and flat can make strong growth look almost horizontal; one that is too tall can make moderate growth look explosive. Use light gridlines sparingly to aid reading values, but not so many that they dominate the data.
Your aim is not mathematical purity but visual honesty. A reasonable rule of thumb: if a fair-minded skeptic could accuse the chart of dramatizing or downplaying the story through its scales, fix the scales.
3.4 Color, labels, and call-outs in the McKinsey style
You have three main tools to guide the eye: color, labels, and call-outs. Used sparingly and consistently, they make charts readable even under time pressure.
Color
In McKinsey-style charts, color is functional, not decorative.
- Use a restrained, neutral base palette for most series.
- Reserve a single stronger highlight color for what matters most: “our client,” “recommended scenario,” or “target state.”
- Keep the mapping consistent across the deck. If blue is the client and gray is the benchmark, do not swap them midstream.
Avoid rainbows, gradients, and heavy fills. When everything is loud, nothing stands out.
Labels
Direct labeling usually beats legends. Whenever possible, label lines and bars directly next to the data instead of forcing the reader to shuttle between chart and legend. Keep labels short, business-like, and clearly associated with their element.
A few practical habits:
- Round numbers sensibly; executives rarely need two decimal places.
- Label only the points that matter: the start, key inflections, and the end, not every single data point.
- Make sure labels never overlap or sit on top of each other; if they do, simplify the data or use fewer series.
Call-outs and annotations
Call-outs are there to explain, not to decorate. A good call-out answers “What should I notice here?” in a phrase or two. Use:
- A single arrow or bracket on the key bar, point, or region.
- A short, meaningful annotation (“New pricing introduced here,” “Top-quartile branches,” “FX impact”).
- Consistent styles for recurring themes (e.g., the same call-out shape and tone for “one-off effects” across multiple slides).
If you feel tempted to add many overlapping annotations to rescue a confusing chart, the problem is not the absence of call-outs; it is the underlying design. Simplify first, annotate second.
3.5 Building charts efficiently from Excel to PowerPoint
All of this needs to work at 11:30 p.m. the night before a Steering Committee. So beyond design principles, you need a reliable, fast workflow.
Start with clean data in Excel. Structure your data with charting in mind: one column per series, one row per time period or category, clear headers, and no merged cells. If the data are messy, spend a few minutes tidying them; it pays off every time you need to update the chart.
Then, build the base chart in Excel before you worry about styling. Insert the simplest chart type that matches your purpose (line, column, bar, etc.). Let Excel guess initially, then adjust titles, axis labels, and series selection so the skeleton is correct. Only then move to formatting.
From there, think in terms of templates and reuse:
- Maintain a small library of Firm-standard chart templates: line over time, simple bar, stacked bar, waterfall, scatter, etc.
- Instead of formatting each new chart from scratch, paste your data into a template and adjust.
- Keep charts linked to the underlying Excel when frequent refresh is expected; when numbers are stable (e.g., historicals) and you care more about robustness, paste as an enhanced metafile or picture to “freeze” the view.
In PowerPoint, align your charts to the slide master and grid. Use consistent positions and sizes so that flipping between pages feels stable to the reader. Place titles, subtitles, and sources in their standard locations. The less you improvise layout, the more your charts will feel like part of a coherent narrative rather than a collage.
Finally, avoid heavy manual overrides that break on update:
- Do not drag individual bars to fake waterfalls; use proper waterfall chart structures.
- Avoid manually drawn shapes that mimic data; they will not move when numbers change.
- Be cautious about reshaping axes by hand; let the underlying data and defined scale drive what you see.
If you build your charts on solid data structures, use a small set of standard formats, and respect the master layouts, you will find that updating a full deck the night before a client meeting becomes an exercise in controlled maintenance, not a scramble of rework. That is how the graphics function has operated for years—and the standard this guide aims to help you reach on every engagement.
