Presentation Designer

Most decks fail before the first slide is built. The designer opens PowerPoint, stares at a blank slide, and starts typing. A title goes in. Some bullets follow. A chart appears on slide 6. By slide 12, no one — including the presenter — knows what argument is being made. The result is what Edward Tufte called a slideument: a document pretending to be a presentation, delivering the worst of both worlds. Too thin to read independently, too dense to absorb live.

This skill refuses that workflow. It builds decks from the argument outward: story spine first, slide structure second, visual layer last. Every decision — what goes on a slide, how data is displayed, how type is set, whether an animation fires — is evaluated against a single question: does this help the audience receive the one idea this slide is meant to deliver? If not, it doesn't belong.

When to use / when not to

Reach for this skill when you need to design, structure, or overhaul a slide deck so it communicates and persuades — not just displays information. Covers pitch decks, sales decks, conference talks, internal strategy decks, and stakeholder updates. Handles the full workflow: narrative structure first, slide-by-slide design, data visualization, visual hierarchy, typography, speaker notes, and template/system thinking.

Do not use it when the job is purely fundraise-narrative content strategy (defer to pitch-deck-framework for investor story arc and content), purely chart-type selection for a standalone analysis (defer to visualization-chooser), or generalist graphic design with no presentation context.

Modes

Context Loading Gates (Required Before Any Output)

Before designing a single slide or writing a story spine, pull all three of these. Missing one produces a deck that fits neither the argument nor the room.

1. The Message Gate

• What is the single thing the audience must believe, decide, or do differently after this presentation?
• If you cannot state it in one sentence, the deck cannot be built yet. Work on this first.
• What is the secondary belief or context they need to hold that belief? (This is typically the second-to-last or penultimate beat of the narrative.)

2. The Audience Gate

• Who specifically is in the room? (Role, seniority, domain expertise)
• What do they already believe about this topic? (Starting sympathy, skepticism, neutrality?)
• What objection will live in the back of every skeptic's mind for the first three slides? (Name it explicitly — the structure must address it proactively.)
• Are they reading this as a leave-behind, or is a live presenter controlling the pacing?

3. The Setting Gate

• Projected in a dark conference room? Virtual screen share? Printed one-up? Sent as a PDF with no speaker?
• Time allowed?
• Does the presenter speak over slides, or is this a self-navigating deck?

The answers to these three gates determine: narrative structure, slide density, font size minimums, whether slides can carry more text (leave-behind) or must be stripped down (live presenter), and which visual vocabulary will land vs. confuse.

Methodology

The Story Spine: Argument Before Slides

The single most common failure mode in presentation design is starting with slides. Even experienced designers do this. The deck becomes a collection of individual statements rather than a sustained argument that moves an audience from one position to another.

The story spine is a pre-slide artifact. It exists entirely in words. It defines the logical chain that your slides will visualize. You cannot skip it — or if you do, you will rebuild it later at three times the cost.

The SCQA Spine (for most business decks)

McKinsey, Bain, and BCG converge on a structure derived from Barbara Minto's Pyramid Principle. The entry is the SCQA frame:

Situation — What is the current, uncontroversial state of affairs that the audience already accepts? (One to three sentences. Nothing in the Situation should be arguable.) This grounds shared reality.

Complication — What has changed, is changing, or is broken within that situation that creates a problem or opportunity? The Complication is the reason this deck exists. It introduces tension. Without it, there is no reason for the audience to keep listening.

Question — What question does the Complication logically raise? This is often implicit but should be named explicitly in the designer's spine document. "So what should we do about [the complication]?" or "Is this a real problem, and if so, how bad is it?"

Answer — The deck's central thesis, stated as a complete assertion. This is the answer to the Question. In consulting decks, the Answer appears early — sometimes in the first substantive slide — because the audience wants to know the conclusion before evaluating the evidence. This is the Pyramid Principle: lead with the conclusion, support below.

Critically: the Situation and Complication should contain nothing the audience would contest. The entire load-bearing argument lives in the Answer and its supporting evidence. If the Situation or Complication is controversial, the audience spends the opening slides arguing internally rather than absorbing the frame you're setting.

The Problem-Insight-Solution Spine (for pitch and product decks)

For decks that need to sell a belief rather than advise a decision:

1. Context — What is the world this problem lives in? What has shifted recently that makes this moment significant?
2. Problem — Who is suffering, and how? Make the audience feel the friction, not just understand it intellectually. Specificity here is load-bearing — generic pain statements slide past.
3. Insight — What did you see that others missed? Why hasn't this been solved? The Insight is your differentiator at the argument level, before you even name a product.
4. Solution — How does your approach uniquely resolve the Insight gap?
5. Evidence — What validates the solution? (Data, results, case example, traction)
6. Ask — What specific action, decision, or belief shift does the audience leave with?

Spine quality check

Before translating the spine into slides, run these tests:

• Read the spine aloud in 90 seconds. Does it make a coherent argument even without visuals? If not, the visual layer cannot rescue it.
• Does each beat raise a question that the next beat answers? If there are logical jumps, a slide will feel arbitrary.
• Could a skeptic sitting in the third row articulate the objection that the deck must address? Have you placed that objection's response before slide 8?
• Is the last slide asking for something specific — a decision, an approval, a follow-up — or does it trail off into "Next Steps" vagueness?

Slide Structure: The Assertion-Evidence Model

Once the spine exists, each slide gets exactly one job: deliver one beat of the argument. The Assertion-Evidence model, developed by Michael Alley for engineering communication and widely validated for business presentations, is the operating principle for every substantive slide.

Anatomy of an Assertion-Evidence slide

The title is the assertion, not the topic. Traditional slide design uses titles like "Q3 Revenue" or "Customer Feedback." These are topics, not arguments. An assertion title states what you want the audience to conclude from this slide: "Q3 revenue grew 34% despite market contraction" or "Customers leave because of onboarding friction, not product quality." The audience reads the title and knows immediately what to think about what follows.

The assertion title should be:

• A complete declarative sentence
• The conclusion the evidence supports (not a question, not a label)
• Specific enough that the evidence either confirms or denies it
• No longer than two lines at the slide's heading size

The body is evidence, not more text. Below the assertion, the slide carries the visual that proves or demonstrates the claim: a chart, a diagram, a photograph, a direct quote, a before/after comparison. The body content is almost never a bullet list. If the assertion is "Customer churn is concentrated in the first 14 days," the body is a retention curve chart — not four bullets explaining the retention curve in words.

The one-idea rule

Each slide carries one idea. Not one topic. One idea — a single claim the audience can accept or reject. If a slide requires two paragraphs to set up before the chart makes sense, that's two slides: one for the setup claim, one for the chart and its assertion headline.

The practical test: cover the title. Can you still explain what the chart proves? If you need to re-read the title to know what conclusion to draw, the evidence isn't strong enough or the chart needs a different assertion. Cover the chart. If the title alone already conveys the point without the visual, the visual is decoration.

Visual Hierarchy and the Squint Test

Every slide has a visual hierarchy — an intended reading order that guides the audience's eye from the most important element to the least important. The problem is that most slides have an unintended hierarchy, where color, size, and position have been assigned arbitrarily, and the audience's eye lands on the logo in the corner before it lands on the data.

The Squint Test

Squint your eyes at the slide until it blurs. Ask: what is still visible? What has disappeared?

What survives the squint is your dominant visual element. In a well-designed assertion-evidence slide, the assertion headline and the primary data point (the key number, the tallest bar, the trend line's direction) survive. Supporting labels, axis text, and footnotes dissolve — as they should.

If the slide's logo, a decorative background pattern, or a large block of body text dominates the squint, the hierarchy is inverted. The thing that draws the eye first is not the thing that delivers the message.

The SPECS levers

Five controls set hierarchy on a slide:

• Size — The assertion headline is the largest text on the slide. The primary data point or key visual element is the largest graphic element. Everything else is smaller.
• Position — Audiences in Western-language contexts scan top-left first, then right and down. Place the assertion headline at the top. Place the primary insight (the number, the peak of the curve, the key comparison) in the upper-left quadrant of the visual area.
• Elements — Every element on the slide must earn its place. Remove anything that doesn't directly support the assertion. This includes: decorative borders, background textures not serving a semantic purpose, redundant labels, "and" items that belong on a separate slide.
• Color — Use color as hierarchy signal, not decoration. The primary evidence uses the brand's lead color or the most saturated tone. Everything else uses grey, muted versions, or white. One pop of saturated color per slide, maximum.
• Strength — Bold the key number or key phrase in the assertion. Use regular weight for everything else. Never underline (underline signals link on screen). Avoid italic for evidence — it signals qualification, not emphasis.

White space as a hierarchy tool

White space — empty areas — is not wasted space. It is a hierarchy signal. It tells the audience "look at what is here, not everywhere." Target at least 30–40% of the slide area as intentional negative space. Slides that feel "full" are almost always slides where everything is competing equally for attention, meaning nothing wins.

Data Visualization on Slides

A chart on a slide has a single job: make one data story visible instantly. It is not a data dump. It is not a comprehensive view of a dataset. It is visual evidence for the assertion in the headline.

Chart selection principles

Choose the chart type that makes the relationship type visually obvious, then simplify until only the story remains:

The simplest chart that accurately conveys your message is the right chart. When the data story is "we grew 3x," the slide is a large typographic "3x" with the assertion headline — not a bar chart.

Tufte's data-ink principles for slides

Edward Tufte's core principle: maximize the data-ink ratio. Every drop of ink (or pixel) on a slide should either display data or be necessary to interpret data. Everything else is chartjunk.

Remove:

• 3D effects on bars, pies, or lines (distort the visual proportions; never add information)
• Shadow effects on chart elements
• Background fills on the chart area (the slide background is already white or lightly colored)
• Gridlines that are denser than the data requires — typically two or three horizontal gridlines maximum
• Chart borders
• Legends/keys — label data directly instead. When the audience must look back and forth between a legend and the data, comprehension degrades. Use data labels on the bars or direct labels on line endpoints.

Keep or add:

• Direct labels on data series with the key values (the peak, the final point, the outlier)
• A single annotated callout arrow or circle pointing to the most important data moment
• Zero baseline on bar and column charts (never truncate the axis to make changes look bigger)
• A consistent color across the same data series throughout the entire deck

The data story headline rule

Every chart slide needs an assertion headline that states what the data proves — not a topic label. "Revenue by quarter" is a topic. "Revenue accelerated in Q3 as CAC dropped" is an assertion that tells the audience what conclusion to draw before they study the chart.

The headline and the visual should be redundant in meaning but non-redundant in information: the headline states the conclusion in words; the chart shows the evidence in data. Neither is sufficient alone.

Typography and Contrast for Projection and Screen

Typography on slides exists in a hostile environment. Projection degrades contrast. Screen sharing compresses resolution. Rooms are too bright. Viewers may be 30 feet from the screen or squinting at a laptop with a dozen browser tabs competing for attention. The defaults in PowerPoint and Keynote are wrong for this environment.

Size minimums

These are floors, not targets. Go larger whenever the slide's density permits.

• Assertion headline (title): 28–36pt minimum. Larger if the slide carries very little other content.
• Body text, chart labels, captions: 18pt minimum. This is the floor for readable projection in a standard conference room. For webinars and screen-share contexts where viewing distance varies, 20pt is safer.
• Footnotes, source citations, legal text: 14pt minimum. Never smaller. Anything under 14pt is invisible when projected.

If content does not fit at these sizes, the slide has too much content. Remove content, not size.

Font selection

• Use a single sans-serif family for body text and data. Sans-serif fonts (Inter, Helvetica, Arial, Calibri, Source Sans) render cleanly on screens and projectors. Serif fonts can work for display headlines when the brand requires them but degrade at small sizes under projection.
• Use maximum two typefaces per deck: one for headlines, one for body/data. More creates visual noise without adding hierarchy.
• Avoid decorative or script fonts anywhere that text carries a load-bearing message. They add reading friction.

Contrast

• Dark text on light background is the safest choice for projected presentations. The projector degrades color fidelity and brightness; dark-on-light survives that degradation better than light-on-dark.
• For dark-background decks (which look dramatic in demos, poorly in bright conference rooms), ensure headline text is pure white or very near white. Creams and off-whites disappear fast on projectors.
• Verify contrast ratios before finalizing. A ratio of at least 4.5:1 for body text (WCAG AA standard) is the floor; for slide text under projection conditions, 7:1 or higher is safer.
• Never place text on top of a complex photograph without a semi-transparent overlay that clears the background. "It looks fine on my screen" is not sufficient — check it projected or on a lower-quality monitor.

Emphasis hierarchy

• Bold for primary emphasis. One use per slide, on the key number or key phrase.
• Italic for occasional secondary emphasis or proper terminology — not for emphasis generally.
• Color for categorical distinction in data, not for decorative emphasis in text.
• ALL CAPS in small doses for section labels or category tags — never for running text (degrades readability significantly).
• Never underline text on digital slides unless it is a genuine hyperlink. Underline signals "clickable" to any audience raised on the internet.

The Slide-as-Teleprompter Anti-Pattern

The most common failure in live presentations: the speaker puts everything they want to say on the slide, then reads it aloud. The audience reads faster than the speaker talks. By the time the speaker reaches the third bullet, the audience has already read all six and is bored, checking their phone, or mentally arguing with bullet four while the presenter is still on two.

The slide is not the presentation. The speaker is the presentation. The slide is the visual evidence that makes the spoken argument tangible.

Signs that a deck is being used as a teleprompter:

• Slides contain full sentences in the body text that the presenter reads verbatim
• Bullets are sub-sentences of each other ("First we did X. Then we did Y. Finally Z")
• The "agenda" slide lists the presentation structure in sentence form
• There is no speaker — if you remove the presenter, the deck still conveys 80% of the information (this is a leave-behind, not a presentation)
• The presenter turns to face the screen to read their own slides mid-talk

The correction: Strip the slide to the assertion headline and the single most powerful visual evidence. Move every explanatory sentence to speaker notes. The presenter knows what to say; the slide's job is to make the audience see what the presenter is saying, not read it while they wait.

This creates a division of labor: the spoken word carries the argument, the evidence, the nuance. The slide carries the image, the data, the anchor that keeps the idea in the audience's working memory during and after the talk.

Builds, Animation, and Transition Restraint

Animation is almost always a distraction. The exception: a carefully chosen reveal that controls the pace of a complex argument or forces the audience to process one piece of information before seeing the next.

Default: no animation

Every slide should be designed as if it will be printed or PDF'd. The static version must make sense on its own. If a slide only works with its animation sequence, the slide is not well-designed — it has outsourced its clarity to a performance.

When builds are justified

• Sequential argument: When a slide shows a process with three steps, and the audience must understand step one before step two makes sense, a simple fade-in reveal on step two is justified. The audience finishes processing step one, the presenter clicks, step two appears.
• Data comparison: When showing before/after data, a click-to-reveal of the "after" state lets the presenter hold the audience in the "before" for a moment of tension. But the final state must be visible before the presenter moves on.
• Diagram construction: When building a system diagram piece by piece to explain architecture or flow, each piece can fade in as the presenter names it.

What is never justified

• Entrance effects for every bullet — "Fly in from left" on six sequential bullets turns a slide into a loading screen. The audience waits for the animation to finish rather than engaging with the content.
• Slide transitions with motion — Cube rotates, page curls, and wipe transitions add visual noise between ideas. Use a simple cut or a half-second dissolve at most.
• Animations that are noticeable as animations — If the audience's attention goes to the animation rather than to what it's revealing, the animation has failed. The rule of thumb: if the motion is conspicuous, it is wrong.
• Looping GIFs or auto-advance video on data slides — Motion in the background of a data slide splits the audience's attention precisely when focused reading is required.

Speaker notes — the presenter's half of the split

Every substantive slide in a live presentation should have speaker notes. The notes carry:

• The full sentence that delivers the assertion headline (the thing the presenter will say that the headline echoes visually)
• The one or two supporting sentences that explain the evidence on the slide
• Any statistic, attribution, or nuance that matters but doesn't fit the slide
• Transitions: the sentence that bridges from this slide to the next ("That's why this matters — and it compounds with what we see on the next slide...")

Speaker notes are not a script to be read verbatim. They are an anchor — a reference the presenter can glance at in presenter mode without losing their place. The goal is that the presenter has internalized the material well enough to speak naturally, using the notes only to ensure completeness.

Template and System Thinking for Recurring Decks

A one-off deck can be designed ad hoc. A deck type you will produce repeatedly — monthly business reviews, client status reports, sales proposals, conference submissions — needs a template system. The system is an investment that pays back on every future instance.

What a slide system defines

Slide Master settings (the locked layer):

• Brand fonts: heading family + body family + sizes at each hierarchy level
• Brand colors: primary, secondary, accent, neutral grey scale
• Logo position, size, and safe area (no content enters the logo zone)
• Footer content: company name, date placeholder, slide number — and their fixed positions
• Background treatment: solid, gradient, or pattern — applied once, inherited everywhere

Layout palette (the toolkit layer): A set of pre-designed slide layouts covering the recurring use cases:

• Title slide
• Section divider
• One assertion + one full-bleed visual (most common content slide)
• One assertion + two-column evidence (for comparisons)
• One assertion + three-up comparison (for option matrices)
• One assertion + data table (for performance reviews)
• Closing slide / call-to-action

Each layout has placeholder zones that anyone can populate with content without making font, size, or color decisions. The design thinking is front-loaded into the template; users execute it.

Style rules document (the decision log): A one-page reference that answers the questions contributors will ask:

• What size is the headline on a content slide? (Answer once, enforce always)
• Which color goes on positive metrics vs. negative?
• How are charts sourced? (In-line caption vs. footer citation)
• What does a well-formed assertion title look like? (One example)
• What goes in speaker notes vs. on the slide?

Locking vs. unlocking template zones

In PowerPoint/Keynote/Google Slides, the slide master controls what is locked. Lock:

• Logo and brand marks
• Footer content and position
• Slide background
• Font family (but allow size variation within the hierarchy)

Unlock for content editors:

• Placeholder text zones
• Chart/image content within placeholder frames
• Color fills within the defined palette

When contributors cannot accidentally move the logo or change the typeface to Comic Sans, the deck stays consistent regardless of who builds it. The system enforces the design.

Worked Example: Bullet Slide Redesign

Original slide (the failure)

Title: Customer Retention Challenges

Body bullets:

• Customer churn rate has increased over the past two quarters
• Main reasons include poor onboarding experience and lack of product education
• Support tickets peak in the first 14 days of subscription
• Customers who complete onboarding tutorial have 3x higher retention at 90 days
• We need to redesign the onboarding flow and create video walkthroughs
• Timeline: Q2 redesign, Q3 launch

Problems identified:

1. Title is a topic, not an assertion
2. Six bullets encoding four different ideas (problem + cause + evidence + solution + timeline = four separate slides)
3. The most powerful data point — "3x higher retention" — is buried in bullet four as a subordinate clause
4. The final two bullets are action items that belong in a separate "Recommendation" slide, not on a diagnostic slide
5. No visual evidence — data that should be a chart is paraphrased in text
6. Speaker reads all six bullets aloud; audience reads them faster and checks out

Redesigned deck section (three slides from one)

Slide 1 of 3: Title (assertion): Churn has accelerated for two consecutive quarters — and concentrates in the first 14 days

Body: Retention curve chart. X-axis: days since subscription. Y-axis: % of cohort retained. The curve drops steeply between days 1–14, then flattens. Annotation arrow at the day-14 inflection point with a direct label: "72% of churn occurs here."

Speaker notes: "What you're seeing is our retention curve for the last two cohorts. The sharp drop in the first two weeks isn't typical of mature SaaS products — it signals something is wrong specifically in onboarding, not in the product itself."

Slide 2 of 3: Title (assertion): Customers who complete onboarding are 3x more likely to stay 90 days

Body: Simple horizontal bar chart. Two bars: "Completed onboarding tutorial" — 71% retained at day 90. "Did not complete" — 24% retained at day 90. Direct labels on each bar. No legend needed.

Speaker notes: "This tells us the product works. The customers who learn how to use it stay. The problem is not what we're building — it's whether new customers ever understand what they bought."

Slide 3 of 3: Title (assertion): Redesigning onboarding in Q2 directly addresses the retention gap

Body: Simple two-column layout. Left: Current state (icon + two-line description of existing onboarding). Right: Proposed state (icon + two-line description of redesigned flow + video walkthroughs). Arrow between columns.

Speaker notes: "The ask is Q2 resources for a redesign, with Q3 launch. The retention curve on the first slide suggests that even a modest improvement in onboarding completion would pay back within 90 days."

What the redesign did:

• One assertion per slide — three ideas became three slides
• The strongest data point is now the entire slide, not a buried clause
• Bullets became charts — the audience sees the evidence, not a description of it
• Action items were separated from diagnosis
• Speaker notes carry the spoken argument; slides carry the visual evidence

Common Failure Modes (and Their Repairs)

Quality checklist

Run this before the deck leaves your hands. The first three carry most of the weight: a deck that fails them is a document being read aloud, which is the failure this whole skill exists to prevent.

• [ ] The argument stands on its own. Read the slide titles in order with nothing else on screen: they form a coherent case, not a list of topics.
• [ ] Every slide asserts something. Each title is a claim the slide then evidences, not a label like "Results" or "Overview".
• [ ] No slide is the speaker's script. If a slide can be read silently instead of heard, the content belongs in the notes.
• [ ] The squint test passes on every slide. Blur your eyes: the intended focal point is still the first thing you see.
• [ ] One idea per slide. A slide carrying two arguments has been split.
• [ ] Every chart makes its point without narration — the takeaway is in the title, the axis is honest, and nothing decorative competes with the data.
• [ ] Type is legible at the back of the room, not just on the authoring screen, and contrast survives a washed-out projector.
• [ ] Builds and transitions earn their place. Each one reveals sequence or relationship; none is motion for its own sake.
• [ ] Speaker notes exist and carry the words the slides deliberately do not.
• [ ] A recurring deck has a template behind it, so the next edition inherits the system rather than rebuilding it.
• [ ] The deck survives being sent without a presenter — or you have deliberately made a separate leave-behind, and said which is which.

Output Format

The output of this skill adapts to the mode:

Structure-First mode:

1. Story spine document (SCQA or Problem-Insight-Solution) — complete sentences, not bullets
2. Slide-by-slide outline: slide number → one-sentence assertion → type of evidence recommended → speaker note starter
3. Design notes: typography scale, color assignments, chart types recommended per data slide

Slide Makeover mode:

1. Diagnosis: slide-by-slide inventory of failure modes identified
2. Redesigned slide specs: new assertion headline + body description (or new slide division if one slide becomes many)
3. Chart rebuild specs for any data slides (chart type, data series, labeling, color)
4. Speaker notes for each revised slide

Data Visualization Fix mode:

1. Chart-by-chart audit: current chart type, identified problems (chartjunk, wrong type, missing assertion, buried lede)
2. Rebuilt spec for each: chart type, axis treatment, data labels, annotation callout, assertion headline
3. Color and contrast check

Deck System mode:

1. Slide master spec: fonts, sizes, colors, logo zone, footer
2. Layout palette: list of slide types with placeholder zone descriptions
3. Style rules reference document
4. One worked example deck showing all layouts populated with real content

In all modes, the output distinguishes between what goes on the slide (the minimal visual) and what goes in speaker notes (the spoken argument). These are always kept separate. A slide specification that blurs the two produces the teleprompter anti-pattern the system is designed to prevent.

Related skills

• pitch-deck-narrative-architect — Sequences a fundraising or pitch deck into a persuasion arc with a slide-by-slide assertion map and evidence ladder engineered backward from the ask.
• data-deck-chart-designer — Selects the right chart type for a presentation slide's message and strips it to its readable essentials, producing a chart-slide spec sized for back-of-room legibility.
• slide-template-master-system-builder — Builds a reusable slide master library of title, section, content, and data layouts with a locked grid, type scale, and color roles.