Exercises — Chapter 30: Slide Design: Visual Communication for Presentations
Slide design is learned by redesigning slides. Most of these ask you to produce or fix actual slides — sketch them on paper or build them in your slide software; describing a slide in words (as this book does for its figures) is fine where building isn't practical. Aim to apply the chapter's tests, not just recall them.
Selected solutions and rubrics:
appendices/answers-to-selected.md. Where a task is open-ended, use the self-assessment rubric given here.
Part A — Analyze This ⭐
For each, name what's working or broken in this chapter's terms (assertion–evidence vs. topic-and-bullets, one point per slide, the six-word ceiling, the back-row test, animation, template-not-theme, 10/20/30).
A1. A slide's title reads "Results" and its body is a five-item bulleted list of findings. Name two distinct reasons this slide will fail in a talk.
A2. A slide's title reads "Our latency dropped 60% after the caching rewrite," and its body is a single before/after bar chart with the two bars labeled directly. The presenter spends the slide explaining why caching helped. Diagnose: is this assertion–evidence, and is the presenter using the slide correctly?
A3. A title reads "System Architecture, Data Flow, and Security Considerations" (8 words). Without seeing the body, predict the problem and the fix.
A4. A presenter has set every bullet on every slide to fly in from the left one click at a time, explaining, "It keeps people from reading ahead." Identify the real problem (it isn't the animation) and the correct fix.
A5. A chart on a slide uses six rainbow colors distinguished only by a legend in the corner, has heavy gridlines and a 3-D effect, and starts its y-axis at 40 (not zero). List every Chapter 9 / Chapter 10 violation you can find, and say which one is an honesty problem, not just a clarity problem.
A6. A slide shows a single large number — "$2.4M" — with a one-line label beneath it, "annual cost of the current manual process," on an otherwise blank slide. The presenter pauses, then explains the breakdown verbally. Is this a good slide or a lazy one? Justify.
A7. A pitch deck has 38 slides for a 15-minute slot, each fairly clean (one point, one visual). Run the 10/20/30 rule on it. Does it fail? Explain what "10 slides" should really be read as, and what you'd actually check.
A8. A presenter's spoken words on each slide are a word-for-word reading of that slide's sentence title, and nothing more. The slides themselves are clean assertion–evidence slides. Is the deck the problem, or the delivery? Name the specific mistake.
Part B — Revise This ⭐⭐
Rewrite or redesign each. Give the new slide (title + described body), and one line naming the principle you applied.
B1. Convert this topic-and-bullets slide to assertion–evidence. Title: "Q3 Performance." Bullets: "• Revenue up 12% quarter-over-quarter • Customer acquisition cost down 8% • Churn flat at 4.1% • Net revenue retention 108%." (Hint: you can't make four points on one slide. Pick the one point, or split into more slides.)
B2. Fix this title: "Overview of the Proposed Migration Plan, Timeline, Resource Requirements, and Anticipated Risks." Decide how many slides it should be and give each new title.
B3. A presenter's "Methodology" slide lists eight procedural bullets. Rewrite it as a single assertion–evidence slide and list which details you've moved into the spoken narration instead.
B4. A slide titled "Recommendations" lists five recommendations as equal bullets. The presenter's real message is "do one cheap thing first, it has the biggest payoff." Redesign the slide to carry that message.
B5. This chart-on-a-slide is described: "A 3-D pie chart of support tickets by category, eight slices in eight colors, percentages in a legend, on a slide with the company logo in all four corners." Redesign it for projection. (Hint: Chapter 9 has opinions about pies; the reader needs to see one comparison, not eight.)
B6. A presenter has 60-point decorative text, a stock photo of a lightbulb behind the text, and the actual content — a key statistic — in 16-point gray in the bottom corner. Re-allocate the slide's attention correctly.
B7. A slide encodes project status with color only: three project names, each in red, yellow, or green text, no other label. Redesign it so a color-blind viewer and a grayscale-handout reader both get the status.
B8. A deck opens with an "Agenda" slide listing six sections. Replace slide one with something that earns the opening, and say where (if anywhere) a roadmap belongs.
B9. A presenter's closing slide is the single word "Questions?" centered on a blank slide. It's on screen the entire Q&A. Redesign the closing slide to earn that screen time, and explain what it should reinforce.
B10. A "Comparison" slide shows a five-column, seven-row table in 11-point type comparing four tools across seven criteria. It's unreadable from the back, but the presenter insists "people need to see the full comparison." Resolve the tension: what becomes the narrated slide, and what happens to the full table?
Part C — Write/Build This ⭐⭐–⭐⭐⭐
Produce real slides (sketched, built, or described in full).
C1. Redesign these 5 bad slides. Here is a short, genuinely bad five-slide deck for a fictional internal pitch ("we should adopt automated testing"). Redesign all five as assertion–evidence slides. For each, give the new sentence title, the single visual you'd use, and one line on what you cut. 1. Title: "Introduction." Body: "• Our team ships software • Bugs are a problem • This presentation will discuss automated testing • Agenda: problem, solution, cost, benefits, recommendation." 2. Title: "The Problem." Body: "• 23% of our releases required a hotfix in the last quarter • Hotfixes average 6 engineer-hours each • Customer-reported bugs up 15% year over year • Manual QA takes 3 days per release • Engineers spend ~20% of time on bug triage." 3. Title: "Solution." Body: "• Automated testing • Unit tests, integration tests, end-to-end tests • CI pipeline runs tests on every commit • Industry standard practice • Many tools available (Jest, Pytest, Selenium, etc.)." 4. Title: "Cost/Benefit Analysis." Body: A 3-D bar chart, six bars in six colors, y-axis starting at 20, legend in the corner, comparing "estimated setup cost," "ongoing maintenance," "hotfix hours saved," "QA days saved," "bugs prevented," "engineer time freed." 5. Title: "Conclusion." Body: "• Automated testing is a worthwhile investment • Setup cost is recovered within two quarters • Recommend approval to proceed • Questions?"
C2. Convert this bulleted slide to assertion–evidence. Here is one dense slide from a data presentation. Convert it to a single assertion–evidence slide (or split it if it holds more than one point, and say why).
Title: "Survey Results." Body: "• 1,200 customers surveyed (38% response rate) • 67% rated onboarding as 'difficult' or 'very difficult' • 'Difficult' onboarding correlated with 3× higher cancellation • Top requested improvement: guided setup (cited by 54%) • Net Promoter Score for difficult-onboarding cohort: −12; for easy cohort: +41."
C3. Take a deck you have (or one you've sat through). Pick its single worst slide and redesign it as assertion–evidence. Submit before-and-after (described or built) and a one-paragraph diagnosis.
C4. Build the opening three slides of a five-minute talk on a technical topic you know. Slide 1 must open on your point (not an agenda); slides 2–3 must each be assertion–evidence. Include the sentence titles and the visual for each.
C5. Design one slide that requires a build animation — a diagram with at least four parts and a flow between them. Describe the build sequence (what appears on each click) and the sentence you'd say as each piece appears. Then describe what the audience would experience if you showed the whole diagram at once instead.
Part D — Synthesis & Critical Thinking ⭐⭐⭐
D1. Find the flaw. A presenter argues: "Assertion–evidence is fine for simple topics, but my material is complex, so I need dense slides to capture all the nuance." Evaluate this claim. Is complexity a reason to abandon assertion–evidence, or a reason to lean harder on it? (Connect to Chapter 18's threshold concept and to Chapter 1's "writing is thinking.")
D2. Translate for audience. You have one finding — "our model's accuracy improved from 91% to 96% after adding the new feature." Design the single assertion–evidence slide three ways: (a) for a technical peer audience at a research talk, (b) for a non-technical executive deciding whether to fund deployment, (c) for a general-public science talk. The data is identical; the assertion title and the chosen visual should differ. (This is Chapter 2 — audience is everything — applied to a slide.)
D3. Chapter 9 said "a figure does not speak for itself," and this chapter said "the slide's title is its caption." Explain how these are the same claim, and what it implies about the most common slide-design failure (a bare chart with a topic title).
D4. Some organizations (famously Amazon) have banned slide decks for important meetings in favor of written narratives read silently at the start. Using this chapter's threshold concept, explain why that policy can be rational — and explain when slides remain the better choice. (Hint: read/listen channel; the deck-that's-secretly-a-document.)
D5. The 10/20/30 rule's exact numbers are arbitrary, yet the rule improves most decks. Explain the general principle this illustrates about rules of thumb (connect to the "Why Does This Work?" box on §30.6), and give one other rule of thumb from an earlier chapter that works for the same reason.
Part M — Mixed Practice (Interleaved) ⭐⭐–⭐⭐⭐
These mix this chapter with earlier ones, so you must choose the right tool.
M1. You're handed a 4-page written data-analysis memo (Chapter 27) and asked to present it in a 10-minute meeting. Walk through what you do: what stays a document, what becomes slides, how many slides, and how you decide each slide's assertion. (Chapters 27 + 30, and the "separate the artifacts" idea.)
M2. A colleague's slide has a chart, but you're not sure it should be a chart at all — it shows two exact dollar figures the audience will want to quote. Apply Chapter 9's "figure vs. table vs. prose" decision to a slide. Should this be a chart, a big-number slide, or just spoken? Justify.
M3. You wrote an excellent assertion–evidence title, but it runs 22 words and wraps to three lines on the slide. Two earlier-chapter skills are in tension here (conciseness, Chapter 3; one-idea-per-unit, Chapter 6/8). Resolve it: how do you keep the claim and fix the length?
M4. A teammate sends you their deck for review (Chapter 12 — editing; Chapter 23 — collaborative feedback). Give three specific, kind, actionable comments on their slides, in the constructive-feedback style this book teaches. Make at least one comment about assertion–evidence and one about the back-row test.
M5. You're presenting a system design (Chapter 32 previews this). You have one architecture diagram with seven components. Decide: one slide with a build animation, or several slides each showing part of the system? Give your reasoning in terms of one-point-per-slide and the build-animation rule.
M6. Your slide deck must also serve as the leave-behind document afterward (a real constraint your manager has imposed). Reconcile this with the "deck that's secretly a document" warning. What's your move? (There's more than one defensible answer — argue yours.)
M7. You're giving the same 12-slide assertion–evidence talk twice: once live in a conference room, once as a recorded webinar people watch alone at their desks. Do the slides change? Should anything? Reason about the read/listen channel in each setting (Chapter 2 — audience; Chapter 18 — the live listener) and decide what, if anything, you'd adjust for the recorded version.
M8. A reviewer (Chapter 12) tells you your strong assertion title — "Our new caching layer cut p99 latency from 2.1s to 0.7s under peak load" — is "too long for a slide." Is the reviewer right? Apply the six-word-ceiling rule correctly and defend or revise the title. (Watch the trap: the ceiling is for topic titles, not assertions.)
Part E — Extension ⭐⭐⭐⭐ (optional)
E1. Take a 20+ slide deck (yours or a public one — many conference talks post their slides) and perform a full slide audit: for each slide, record (i) is the title an assertion or a topic? (ii) how many points does it make? (iii) would it pass the back-row test? (iv) is any animation earning its place? Tabulate, then write a one-page diagnosis: what's the deck's dominant failure mode, and what are the three highest-impact fixes?
E2. Write a one-page "slide style guide" for a team — the rules a colleague could follow to produce consistent, legible, assertion–evidence slides without having read this chapter. (This is Chapter 23's style-guide skill applied to slides.) Include the non-negotiables and the "it depends" exceptions.
Self-Assessment Rubric (for open-ended items: A6, B1–B8, C1–C5, D1–D5, M1–M6, E1–E2)
Rate each on a 0–2 scale:
- Assertion, not topic (0–2): Does each title state a claim in a full sentence, not name a topic? (0 = topic phrases; 2 = every title is a point.)
- One point per slide (0–2): Does each slide make exactly one point, with multi-point slides split? (0 = crammed; 2 = clean.)
- Visual as evidence (0–2): Is the body a single visual that proves the title, labeled directly, stripped to data-ink — not bullets or chartjunk? (0 = bullets/pasted chart; 2 = clean rebuilt visual.)
- Legibility (0–2): Would it pass the back-row test — big type, high contrast, no meaning on color alone? (0 = unreadable from distance; 2 = legible from the back.)
- Audience & restraint (0–2): Is the slide designed for the listener (your voice carries the rest), with no gratuitous animation and the template's structure overridden? (0 = document-performed-aloud; 2 = a true visual aid.)
10 = publishable. 7–9 = solid, tighten one dimension. Below 7 = redo, targeting the lowest score. For B1, B4, C1, C2: the single most important check is whether you chose one message per slide rather than reproducing every bullet — reproducing the bullets in a different layout is the most common failed attempt.