SlideLegend

Fix the slide

Reviewing someone else's slide is easier than reviewing your own, because you see it the way the audience does: all at once and without the backstory. This game trains that outside view.

Each of the three slides hides five faults. They are the ones that turn up again and again in real decks: a title that names a topic instead of stating the finding, a 3D pie that distorts its own slices, bars that do not start at zero, text too pale to read on a projector, a paragraph that competes with the speaker, gridlines louder than the data, a legend that sends the eye on a round trip. Some regions on each slide are fine, so clicking everything does not work.

Each fault you find gets a label and a one-paragraph explanation. When all five are marked, switch between the original and a fixed version of the same slide. The fixed versions use the same data; only the choices change. That comparison is the real lesson: most slide problems are solved by removing or rearranging, not by adding.

Three slides, five faults each

Each slide hides five faults. Click or tap the part of the slide you think is wrong; keyboard users can tab through the marked regions and press Enter. Some regions are fine, so choose with care. When all five are found, compare the slide with its fixed version.

Slide 1 of 3 · Quarterly sales

Q3 Sales Overview +12%!! Revenue share by region NorthSouthWestEast Revenue by quarter, €m 9095100105110 Q1Q2Q3 Source: finance team ledger, Q1 to Q3 3

Fixed version

Q3 revenue hit €108m, up 12.5% on Q2 Revenue share by region, Q3 North 38% South 27% West 21% East 14% Revenue by quarter, €m 94 Q1 96 Q2 108 Q3 Source: finance team ledger, Q1 to Q3 3

  1. Topic title. "Q3 Sales Overview" names the subject but not the finding. A title that states the message ("Q3 revenue hit €108m, up 12.5% on Q2") is read first and remembered.
  2. Decoration competing with data. The starburst is the brightest thing on the slide, so the eye lands there first. The number belongs in the title, not in clip art.
  3. 3D pie. Tilting a pie makes front slices look bigger than back slices of the same size. A flat, sorted bar chart shows the shares accurately.
  4. Bars without a zero baseline. The axis starts at 90, so Q3 looks four times bigger than Q1 when it is about 15% bigger. Bar length only works when bars start at zero.
  5. Low-contrast text. Very light grey on white is unreadable on most projectors. Small text needs at least a 4.5 to 1 contrast ratio with its background.

Slide 2 of 3 · Customer survey

Satisfaction climbed after the new support hours In the first quarter customers told us thatreaching support was hard, especially afterwork. From March the support desk stayedopen until 8 pm on weekdays, and a chatchannel was added on the website. The surveywas sent to 2,400 customers each month andthe response rate stayed between 31% and34%. Satisfaction rose in every channel butthe gain was largest for chat, while emailbarely moved because replies still took twodays. Next steps are under discussion andwill be presented at the next meeting. Satisfaction: 72.48613% JanFebMarAprMayJun Overall Phone Email Chat

Fixed version

Satisfaction climbed after the new support hours Support open until 8 pm since March Chat rose from 52% to 75% satisfied Email barely moved: replies still take two days 72% satisfied in June, up from 58% in January % of customers satisfied 50% 60% 70% 80% OverallPhoneEmailChat JanFebMarAprMayJun

  1. Text wall. Twelve lines of small text compete with the speaker: people read instead of listening. Keep three short points on the slide and say the rest.
  2. False precision. Five decimal places suggest an accuracy a survey does not have, and they are hard to say aloud. Round to "72%".
  3. Missing axis labels. Without values or a unit, the reader cannot tell whether the lines moved by 2 points or 20. Label the axis and say what it measures.
  4. Cluttered gridlines. Dense dark gridlines are louder than the data. A few faint lines at round values are enough, or none if the lines are labelled.
  5. Legend far away, similar colours. Four close shades of blue in a box far from the lines force the reader to match colours back and forth. Label each line at its end and highlight the one that matters.

Slide 3 of 3 · Hiring plan

Engineering hires fell behind plan in Q2 036912 Hires 200400600800 €k Platform EngineeringData InfrastructureMobile ApplicationsSecurity OperationsDeveloper ToolingQuality Assurance NOTE!!! HIRING IS VERY BEHIND PLAN!!! Please REVIEW ASAP (see chart) Hires Budget

Fixed version

Engineering hires fell behind plan in Q2 Data and Security are furthest behind. Hires made in H1 against plan, by team. Hires madePlan Data Infrastructure 4 of 8 Security Operations 3 of 6 Mobile Applications 6 of 8 Platform Engineering 9 of 10 Quality Assurance 5 of 6 Developer Tooling 7 of 7

  1. Rainbow gradients and shadows. Each bar has its own colour that means nothing, plus gradients and shadows that blur where the bar ends. Use one colour and a second only for emphasis.
  2. Dual axis. Two scales on one chart let the crossing of the line and the bars mean whatever the scales make them mean. Show the budget separately, or drop it.
  3. Rotated labels. Tilted text is slow to read. When category names are long, turn the chart sideways so the names run horizontally.
  4. Shouting emphasis. Capitals, underlining, italics, red and exclamation marks all at once cancel each other out. One calm sentence in the title or subtitle carries the point.
  5. Detached legend. The legend sits far from the chart it explains. Put a short key directly above the chart or label the elements themselves.

Questions

Why does a 3D pie chart distort the data?

The tilt puts the front slices closer to the viewer, so they look larger than slices of the same value at the back, and the visible depth edge adds area that stands for no value at all. Readers already judge angles and areas less accurately than lengths, so the perspective stacks a second error on top of the first.

Should a bar chart always start at zero?

Yes. A bar shows its value as a length, so a baseline above zero shrinks every bar by the same amount and makes small differences look large. Line charts are different: they show change through position and slope, so a baseline above zero is acceptable there as long as the axis is clearly labelled.

How much contrast does text on a slide need?

WCAG 2.2 asks for at least 4.5 to 1 between normal text and its background, and 3 to 1 for large text. A projector in a lit room washes colours out further, so a pale grey that passes on a laptop screen can become unreadable on the wall. Dark text on a light ground, or the reverse, is the safe choice.

What is the difference between a topic title and a message title?

A topic title names the subject, such as "Q3 sales". A message title states the finding, such as "Q3 sales fell after the price change". The message title tells the audience what to look for in the chart before they start reading it, which is the main idea behind assertion-evidence slides.

Back to the Lab