Before you decide which chart to use, you need to be clear on what job the chart is doing in your story. Most consulting pages are answering a small set of recurring questions about the data:
- How is this changing over time?
- How big is this compared to that?
- What is this made of?
- How is this spread out?
- How do these things relate to each other?
- How do we get from here to there?
This chapter walks through those six core purposes. For each one, I will focus less on the mechanics of chart types and more on the underlying question: what the chart is trying to show, when you should use that “lens” on the data, and what to watch out for. The later chapters then map each purpose to specific chart families.
2.1 Showing change over time (trends, cycles, seasonality)
“Is this getting better or worse?” is one of the most common questions you will be asked. Charts that show change over time answer that question far more effectively than tables of period-by-period numbers.
When you care about trends, the time axis is the spine of the chart. Time should almost always be on the horizontal axis, moving left to right in a natural sequence. The viewer’s eye will trace that line or column sequence and quickly register whether it is rising, falling, flat, volatile, or cyclical.
You use a time-series chart when:
- The question is about direction (up, down, stable) rather than a single point estimate.
- You want to show before vs after an event: policy change, acquisition, price increase.
- You want to separate short-term noise from structural trends.
Typical chart families for time:
- Line charts: the default for continuous time series (monthly revenue, daily sign-ups).
- Column/column+line: good for discrete periods (annual results, quarterly volumes).
- Area charts: emphasize the total over time, especially for stacked series.
Common mistakes:
- Showing too many series on one chart, turning it into spaghetti.
- Using irregular time intervals without signalling it clearly.
- Over-zooming on a short time window that exaggerates a small effect.
When in doubt, ask: “If I removed the axis labels and only kept the shape, would the story still be obvious?” If the answer is no, you may have the wrong chart or the wrong level of time granularity.
2.2 Comparing quantities across categories (cross-section comparisons)
Sometimes the client does not care about time at all. They want to know: “Who is the biggest?” “Which segment is underperforming?” “Where should we focus?” Those are cross-section questions. You are comparing values across categories in a single moment or period.
In these cases, the categories themselves—business units, countries, customer segments, products—are the key structure, not time. Your job is to make it effortless to see which category is larger, smaller, or different.
You use cross-section charts when:
- Ranking matters: top 5 markets, largest cost buckets, highest NPS drivers.
- You are contrasting peer groups: us vs competitors, A vs B vs C.
- The decision is about allocation: where to invest, cut, or pilot first.
Typical chart families:
- Bar charts (horizontal) and column charts (vertical) for simple comparisons.
- Clustered bars/columns when you compare multiple metrics within each category.
- Dot plots or “lollipop” charts for cleaner comparisons when you have many categories.
Good practice includes:
- Ordering categories by value (descending or ascending), not alphabetically, unless the sequence itself matters.
- Limiting the number of categories in the main chart and moving the long tail to an appendix.
- Using consistent colors for the same entity (“Client” vs “Competitor A”) across the entire deck.
If you find yourself adding a time axis and multiple series to a chart that is supposed to answer a simple ranking question, you may be mixing purposes. Split the story: one page for trend, one for cross-section comparison.
2.3 Showing part-to-whole composition (percentage breakdowns)
Executives often ask “What is this made of?” or “Where does the money go?” That is a composition question: you are explaining how a total is split into components.
Composition charts are powerful because they shift the conversation from the size of the total to the structure of that total. For example, instead of just saying “Opex is $500M,” you show that 60% is labor, 25% is third-party spend, and 15% is overhead—and that mix is different from peers.
You use part-to-whole charts when:
- The share or percentage matters more than the absolute number.
- You want to highlight concentration (e.g., “three SKUs drive 70% of revenue”).
- You are setting up a rebalancing or reallocation recommendation.
Common chart families:
- Stacked bar or stacked column charts (absolute or 100% stacked) for clean comparisons across multiple entities or time periods.
- Mekko/Marimekko charts when you want to combine share of total with another dimension (e.g., market size × share).
- Pies and doughnuts for very simple, single-point-in-time mixes with few categories.
Pitfalls to avoid:
- Too many slices or segments; composition charts break down when you have more than 5–6 meaningful components.
- Mixing absolute and percentage logic on the same page without clear labeling.
- Using pies when you need to compare multiple periods or entities; stacked bars are almost always better for that.
A useful test: if the discussion is about changing the mix, you almost certainly want a stacked or 100% stacked structure rather than a simple bar or line.
2.4 Showing distributions and spread (ranges, skew, outliers)
Average values are comforting but often misleading. Many client problems are actually about variation: why some branches are much better, why a small set of customers drives most of the profit, or why a tail of long service times is hurting satisfaction.
Distribution charts help you answer: “How is this quantity spread out?” They show you not just the center (mean, median) but the shape: narrow vs wide, symmetric vs skewed, with or without outliers.
You use distribution views when:
- You need to talk about risk or variability, not just the typical case.
- You are segmenting performance: top quartile vs bottom quartile units.
- A small tail has disproportionate impact (e.g., extreme delays, top-value customers).
Relevant chart types:
- Histograms for frequency distributions over a continuous variable (order size, response time).
- Box-and-whisker or range plots for comparing spread across multiple groups.
- Population pyramids or similar range charts when age or size bands matter.
Typical consulting mistakes:
- Presenting only an average for inherently skewed data (e.g., income, deal size).
- Hiding outliers by capping axes too tightly.
- Using overly technical statistical visuals without explaining the takeaway in plain language.
Whenever the client conversation touches on “variability,” “volatility,” “tail,” or “dispersion,” consider whether a distribution chart would make the point more honestly than a single average.
2.5 Showing relationships and segmentation (correlations, clusters, portfolios)
Many executive questions are about how two (or more) things move together: “Do higher discounts actually drive volume?” “Are our fastest-growing customers also the most profitable?” “Which segments are both large and attractive?”
Here, the core purpose is to show relationships between variables and, often, clusters of similar entities.
You use relationship charts when:
- You suspect a trade-off (e.g., price vs share, speed vs cost).
- You want to highlight segments or quadrants: high growth / low margin, strategic vs non-strategic customers.
- You are building a portfolio view: which businesses or products sit where on a risk–return map.
Typical chart families:
- Scatter plots (simple or four-quadrant) for two-variable relationships.
- Bubble charts when you need a third variable (e.g., size of prize).
- Radar charts sparingly, when comparing multi-attribute profiles across a few entities.
- Heat maps when showing strength of relationship across many combinations (e.g., correlation matrices, product–segment fit).
Key good practices:
- Keep the number of points manageable on a main-page scatter; move dense clouds to the appendix or aggregate.
- Use quadrants and shading to make “good” and “bad” zones visually obvious.
- Label only the points that matter (e.g., client vs key competitors), not every observation.
If the decision is fundamentally “which of these items belong in which box?”—core vs non-core, keep vs grow vs exit—you are in the realm of relationship and segmentation charts.
2.6 Showing contributions and bridges (from A to B via drivers)
Finally, some of the most powerful consulting charts are about explaining a change: not just that profit fell from $200M to $150M, but how and why it moved. This is the job of contribution and bridge charts.
These charts decompose the movement from a starting point (A) to an ending point (B) into a series of drivers or effects. They let you say: “Volume added +$30M, price added +$10M, mix subtracted –$40M, and cost inflation subtracted –$50M.”
You use bridge-style charts when:
- The client is asking “What happened?” or “Where did the money go?”
- You need to quantify the impact of individual levers (volume, price, mix, FX, one-offs).
- You are setting up a value-creation or savings storyline: how your proposed actions move the metric from its current level to the target.
Typical chart families:
- Waterfall charts (bar or column) for step-by-step decomposition from A to B.
- Stacked bars across two periods when drivers are naturally grouped into a small number of categories.
- Combination charts where a line shows the total while columns show individual effects, in more advanced cases.
Watch-outs:
- Be clear about the logic of decomposition (e.g., volume–price–mix vs rate–volume–other).
- Sign and direction matter: gains vs losses should be visually distinct and consistently treated across slides.
- Do not overload a single waterfall with tiny steps that the audience does not care about; aggregate minor items into “Other.”
A well-constructed bridge is often the single most remembered page in a deck. It connects analysis to action by showing which levers matter most and how much each contributes to the outcome.
