Data Visualization
Data Visualization
What is data visualization?
- Turn numbers into charts and graphs
- See patterns and meaning at a glance
- Good visualization = fast understanding
Turning a table of numbers into a chart changes the kind of questions raised in a meeting. Put up a table, and verification questions like 'Where did this figure come from?' come first. Put up a single trend line, and the discussion goes straight to causes: 'Why did it turn down in March?' Time once spent reading shifts to time spent judging, and that is what visualization actually changes. With the same material, half the meeting goes to decisions rather than interpretation.
Choose a chart by the question you are trying to answer, not by taste. Use bars to compare sizes across items, lines for change over time, scatter plots for the relationship between two variables, histograms for the spread of values, and choropleth maps for regional differences. For showing shares, a pie chart with more than five slices makes ranking by eye impossible, so a bar chart is the better choice. Keep it to 5 to 7 metrics per screen; beyond that, you have a warehouse, not a dashboard. Also, an exploratory chart an analyst draws for their own use, an explanatory chart placed in a report, and a monitoring dashboard that tracks the same metrics every day are different things. Different purposes call for different pictures, even from the same data.
In practice, axes and colors are where things go wrong most often. If a bar chart's Y-axis starts at 95 instead of 0, a 2-point difference looks like a threefold gap. Get an investment approved with that slide, and you lose all credibility in front of an executive who sees the source data. If good and bad are separated only by red and green, the two bars look like the same tone to attendees with color vision deficiency, so pair color with number labels or symbols. 3D effects, gradients, and flashy backgrounds are noise, not information, and the greater danger is that the smoother a chart looks, the less people question a wrong number. Spending half a day on decoration while skipping the 10 minutes it takes to check the aggregation basis is the most common failure pattern in real work.
Jaryong KooMarketing PhD and authority in AI, data analytics, and personal branding
John KimAuthor of B2B Marketing Design and CEO of Nanum ConsultingThe history of statistical graphics goes back to William Playfair and others in the 18th and 19th centuries, and data visualization became mainstream with the advance of computers and BI tools.
Consider a hypothetical case: Company A, a food manufacturer with 520 employees and KRW 230 billion (≈ US$172,078,400) in annual revenue. Every Monday, a 38-sheet Excel performance report went to the sales meeting, and 1 hour 20 minutes of the two-hour meeting went to checking 'where did this number come from?' The overhaul began when the head of sales said, 'I don't want to look at tables anymore.'
- Manager Kim of the sales planning team went through recordings of the past 8 weeks of meetings and pulled out the 47 questions people actually asked. 41 of them clustered around just two: 'Which channel is missing?' and 'How much did it drop versus last month?'
- Based on those two questions, they standardized the formats: horizontal bars for channel comparisons and line charts for the 13-week trend. They also removed pie charts and 3D effects from the reporting template altogether.
- The first draft packed in 22 metrics, and the division head said, 'I can't tell what I'm supposed to look at.' The team kept only six on the first screen, including revenue, attainment rate, and inventory turnover, and moved the rest to drill-down detail tabs.
- Senior Manager Park on the data team fixed the Y-axis of every bar chart to start at 0, colored only the channels that missed their targets in orange, and added actual value labels at the end of each bar so the chart stays readable even for people who have trouble telling the colors apart.
- Eight weeks after the overhaul, the Monday meeting dropped from 120 minutes to 55, questions to verify numbers fell from 41 to 6, and 24 of the 31 branch managers had started opening the dashboard the day before the meeting.