Analytics and Tracking

Chart

Also called Graph, visualisation

A visual representation of data that maps a dimension to position and a metric to length, height or area.

Quick facts: Chart

Category
Analytics and Tracking
Also called
Graph, visualisation
Level
Beginner
Affects
Report clarity, decision speed, stakeholder understanding
Where to see it
Looker Studio, GA4 exploration reports, Google Sheets
In this article4
  1. How a chart works
  2. Why charts matter
  3. Common mistakes with charts
  4. How to act on it

How a chart works

Every chart does the same thing underneath. It takes a dimension — dates, campaigns, countries, landing pages — and gives each value a position or a colour. It then takes a metric and turns it into a length, a height, an angle or an area. The reader’s eye does the comparison, which is the whole point: a table asks you to read, a chart lets you see.

Chart types differ in what the eye is good at. People judge the lengths of bars sharply and compare heights along a shared baseline well. They judge angles, areas and colour intensity poorly. That single fact explains most of the standard advice: bars for comparing categories, a line for change over time, a table when the exact figures matter, a map when geography is genuinely the question, and pie charts only when there are very few slices and the point is that one dominates.

Why charts matter

A chart is usually where the decision happens. Nobody reads the underlying rows. They look at the shape, form a view in a second or two, and act on it — so a badly chosen chart does not merely look untidy, it produces a different decision from the one the data supports.

The choice also sets what gets noticed. A metric plotted on its own draws attention to its movement; the same metric plotted beside its volume shows whether the movement was meaningful at all. A conversion rate rising while sessions collapse is a bad month, and a chart showing only the rate hides that completely.

Common mistakes with charts

Starting a bar chart’s axis somewhere other than zero exaggerates small differences dramatically, which is fine on a line chart tracking change but misleading on bars the reader is comparing by length. A pie with many thin slices is unreadable, and a stacked bar with many series is close behind.

Two axes on one chart is the persuasive one: putting spend and revenue on separate scales makes them appear to move together whether or not they do. Then there is decoration — charts added because a page looked empty. Every chart a reader cannot act on makes the ones that matter harder to find.

How to act on it

Write the question above the chart before choosing the type, then pick whatever answers that question most directly. One question per chart. If the answer is a single number, it belongs in a scorecard rather than a chart with one bar.

Label the units and the date range on the chart itself, because charts are screenshotted and forwarded without their page. Sort bars by value rather than alphabetically so the ranking is visible, limit series to the number a person can hold in their head, and pair every rate with the volume behind it. A reporting dashboard that follows those rules gets read; one that does not gets ignored, whatever it cost to build.

Do and do not

Do

  • Write the question the chart must answer first
  • Sort bars by value, not alphabetically
  • Label units and date range on the chart

Do not

  • Cut a bar chart's axis short of zero
  • Put two different scales on one chart
  • Show a rate without the volume behind it

Questions people ask about this

Which chart should I use for marketing data?

Use a line chart for anything measured over time, bars for comparing campaigns, pages or countries, and a plain table when readers need to look up exact figures. Reserve maps for questions that are genuinely about location. If you cannot say in one sentence what question the chart answers, the problem is the question rather than the chart type.

Are pie charts really that bad?

They are poor at what people usually ask of them. The eye compares angles and areas badly, so slices of similar size become indistinguishable and small ones are unreadable. A pie can work when there are only a few parts and the message is simply that one dominates. For anything else, a sorted bar chart says the same thing more clearly.

How many charts should a report have?

As few as answer the questions the reader actually has. A report of a handful of well-chosen charts gets read every month; a wall of them gets skimmed and then ignored. A useful test is to ask what each chart would change if the line moved. If nothing would change, the chart is decoration and can go.

Related terms

Found this useful?

Share it, or ask an AI to summarise it

Back to the glossary

Knowing the term is the easy part

Applying it to your own site and budget is the work. Book a call and I will tell you what actually applies to you.