The second level is about making data visible. Nine lessons teach you to pick the right chart for the question: compare, see a trend, spot a pattern, show where people leave. And — more important — to notice when a chart lies. After this level you will read any chart in seconds and build honest pictures that speak for themselves.
Every lesson is a template with demo data: ten minutes to complete, one skill that stays. Open it in the Templates section of Analytics Agent and press “Use this template” to take the lesson on your own data.
Bars or Lines: Comparing vs Trend
Shows: bars answer “who is bigger”, lines answer “where it is heading”; a bar axis must start at zero, or the differences get multiplied. A live example shows the drop you can only see on an honest line.
You get: the two-chart rule covers most day-to-day questions — and you stop building “pretty” charts that lie.
Pie Charts and Their Traps
Shows: a pie is honest when there are few slices and one clearly dominates; for precise comparison, angles and areas are the eye’s weakest spot: no one can tell 6.0% from 6.25% in a pie, while in bars it is obvious; and when you need to keep the whole makeup of many parts, the pie gives way to a treemap.
You get: you stop using pies for serious decisions and know what to use instead.
Histogram: The Shape of Numbers
Shows: a histogram sorts numbers into bins and shows the shape of the data — where the mass sits, where the tail stretches. On 800 customers you see that a typical customer buys once or twice while rare fans buy up to twelve times; the mean and median are drawn right on the shape, and the bin count moves with a slider — from four, which hide the shape, to thirty, which fragment it.
You get: the answer to “what is typical” arrives as a picture, not a guessed average.
Scatter Plot: Two Numbers Side by Side
Shows: 40 stores as 40 points — the cloud shows whether budget and sales grow together; clumps and lone points get checked, not deleted.
You get: a first honest meeting with “grow together — still not cause”, which protects you from false conclusions.
Box Plot: Where the Middle Lives
Shows: level and spread at a glance: two products with the same median can have twice the spread of checks, and the points beyond the whiskers are a reason to ask, not to delete.
You get: the “how do groups differ” comparison — the one where bars of averages are powerless.
Heatmap: Color Shows Intensity
Shows: a “day × hour” matrix painted with color: on Friday evening, 3 of the 84 hours in the grid carry 11.9% of the week’s receipts, and dead Monday is visible at once. Color finds the pattern, tables confirm it with numbers.
You get: “when and where” questions get answered in two seconds and always backed by a number.
Waterfall and Funnel: From What to What
Shows: a waterfall takes money apart step by step — from revenue to profit — while a funnel counts people: visits, carts, payments. The leak shows where the chain drops, not in one overall number.
You get: you learn exactly where and by how much costs press and customers are lost.
Choosing the Right Chart
Shows: a two-question filter covers almost any task: compare — bars, trend — line, distribution — histogram, relationship — points, intensity — color. And a cheat sheet live, right away: seven mini-charts built from one table.
You get: choosing a chart stops being a matter of taste and becomes the answer to “what do I want to see”.
Honest Charts: Fixing a Misleading One
Shows: a “before and after” teardown: axis from zero, one scale per chart, one meaning per color, a label on every axis — plus a ten-second test: cover the title and ask “what do I see first?”. The truncated axis labels itself, and revenue and clicks sit on their own labeled scales, not on one shared scale.
You get: you spot misleading charts in other people’s reports and stop drawing them in your own.