You will be able to choose the right chart for each number and lay out a page that reads in under a minute.
The furniture shop owner looked at Jun's first draft of the dashboard and pointed at a pie chart in the middle of the page. It showed sessions by channel in nine slices, four of them thin slivers in similar shades of blue. "Which one is email?" she asked. Jun had to hover over each slice to find out. Then she asked whether Paid Search was bigger than Organic Social. They were 14 percent and 12 percent, and on the pie they looked the same.
The data was right and the chart was the wrong shape for it. This lesson covers how to match each number to a chart that answers its question at a glance, and how to lay out a page that someone can read in under a minute.
Most marketing dashboards need only three kinds of chart. Picking among them is a matter of asking what the reader wants to know about the number.
For a headline number, use a scorecard: a single figure in large type, such as sales this month or enquiries this week. On its own, a scorecard just states a number. Add a comparison, which Looker Studio's scorecards support directly: set a comparison date range, such as the previous period or the same period last year, and the scorecard shows the change in percent with an up or down arrow. That answers "are we doing better or worse?" with no chart at all. Four or five scorecards across the top of the page make a good opening row.
For a trend over time, use a line chart, with time running along the bottom. Lines show direction and seasonality better than anything else: whether sales are climbing, flat or dipping, and whether this year's December looks like last year's. Keep it to a few lines at most. Two lines, such as this year and last year, are easy to read. Eight lines become a tangle.
For comparing categories, use a bar chart: sales by channel, enquiries by landing page, cost by campaign. People judge lengths along a common baseline far more accurately than angles or areas, which is exactly why the pie chart failed the owner. Sort the bars from largest to smallest so the order itself tells the story, and use horizontal bars when the labels are long, like campaign names.
Tables still have a place, for detail the reader may want to look up rather than see at a glance, such as the top ten landing pages with their key events. Put them lower on the page.
Pie charts work only when there are two or three slices and the point is a rough share, such as "about two-thirds of sales come from returning customers". With more slices, readers cannot tell similar ones apart, and the legend sends their eyes back and forth. A bar chart of the same data is almost always clearer.
3D effects make every chart worse. Tilting a pie or adding depth to bars distorts the sizes, so slices or bars at the front look bigger than equal ones at the back. Looker Studio offers few of these effects, but charts pasted in from other tools sometimes carry them. Leave them out.
Other habits that cost clarity: too many colours, where each colour should mean something; dual axes, where two different scales share one chart and invite false comparisons; and decimal places nobody needs, such as an engagement rate shown as 61.8734 percent.
Looker Studio titles charts with the metric name by default, or not at all. "Sessions by session default channel group" tells the reader what was measured. "Which channels brought visitors this month?" tells them why they should look.
Jun rewrote every title as the question from his sketch in lesson 6.1. "Purchases by channel" became "Which channels brought sales this month?" "Sessions by landing page" became "Are people reaching the new sofa range pages?" The owner said it was the first time she understood what each chart was for without asking.
A question title also keeps you honest. If you cannot write the question a chart answers, it probably does not belong on the page.
Readers often want a different period from the one you set: last quarter, the Chinese New Year weeks, the month of a big promotion. Without a control, they have to ask you or edit the report, and editing is how reports get broken.
Add a date range control to the top of the page. It is a small dropdown that lets the reader pick a period, and every chart on the page follows it, including the comparison periods on scorecards. Set its default to the period most readers want, such as last month or the last 28 days.
If your report has filters people use often, such as a channel or a product range, add a dropdown control for those too. Two or three controls are plenty.
People read a dashboard the way they read a page: top left first, then across, then down. Put the scorecards with comparisons in the top row, the main trend line below them, then the bar charts that explain the trend, and any detail table at the bottom. Keep the controls at the very top where they are found first.
Line charts up so the edges align, use one colour for one meaning throughout, and leave white space between sections. A page that fits on one laptop screen without scrolling gets read far more often than one that runs on.
Most existing reports can be improved without rebuilding them. Retitling a few charts as questions and swapping the one with the wrong shape is often enough to change how a reader uses the page.
Rewrite the titles of three charts in an existing report as questions, and replace any chart that does not suit its data.
Junxiong-WFG Organisation is an authorised representative of AIA Financial Advisers Private Limited (Reg. No. 201715016G).