split-panel

StatementPanelStructure

Featured left panel + supporting right zone — one prominent claim beside the points that substantiate it.

Variant

Use when one prominent element (a heading, a hero number, a pull-quote, a phase) deserves a dedicated panel and the right side carries the supporting points. The default anchors a heading; variants reshape what the panel features: metric (hero number, light-left), pullquote (pull-quote), steps (numbered step-timeline), watermark (accent panel + letterform + meta footer), proof (a scenario signal + two paired proof cards, for a claim that must be demonstrated, not just supported). Add capstone on top of proof for that sequence's climactic entry: the signal becomes a quoted card, the checkpoints become plain top-rule pillars. For a binary decision with a verdict, reach for split-compare. A run of proof slides is a SEQUENCE, so the engine tints each one from the theme's categorical palette by its position in the deck — authors write split-panel proof and nothing else. capstone implies proof, so the sequence's final slide is just split-panel capstone. An explicit cat-1…cat-8 overrides the assignment for one slide.

summaryboard-deckhero-numberpull-quotetakeaway

When to use

One feature, supporting points. When a single prominent element — a thesis heading, a hero number, a quote — deserves its own panel and the right side substantiates it. The panel is the anchor; the right is the evidence.
Pick the variant by what the panel features. Heading (default), hero number (metric), pull-quote (pullquote), phase + numbered steps (steps), an accent panel with a letterform watermark and a metadata footer (watermark), or a scenario + two proof cards (proof).
A claim that must be demonstrated, not just supported. proof pairs the panel's claim with one scenario paragraph ("you know you're here when…") and two named proof cards. Use it when the audience needs evidence the claim actually holds, not just a list of reasons to believe it.
Points carry a title + body. Each right-side item leads with a bold title (lifted automatically) and a nested one-line body. Three or four points read best; more crowds the panel.

When not to use

A binary decision with a verdict. If the slide weighs two options and lands a recommendation, use split-compare — its right zone is a 2-option grid + a verdict card, which split-panel does not provide.
Co-equal halves. split-panel is asymmetric — a featured panel beside supporting detail. For two co-equal options side by side, use compare-prose.
A list with no feature. If there's no prominent left-panel element, a plain list or cards-stack serves better — the panel earns its place only when one element leads.

Common mistakes

Using a ## heading in the left panel under pullquote instead of a > blockquote. Under pullquote the transform builds the left panel from ONLY the blockquote and its citation — an h2 never lands in the left panel at all; it gets swept into the right column above the supporting points instead, leaving the featured panel blank.
Adding a third line to watermark's trailing metadata footer, expecting a third labeled row. The footer's "Audience ·" / "Intent ·" prefixes are hard-coded to the first two list items only — a third item renders with no label prefix at all.
Under proof, omitting the lede paragraph between the heading and the ### signal label. Always write a lede under proof — omitting it lets the extractor mistake the signal paragraph for the lede, hoisting it into the left panel and leaving the right zone's label empty.
Under proof, authoring more or fewer than exactly two checkpoint items. The two-column card grid assumes a pair. One item leaves an empty column; three crowds a third card half-width.
Applying the same cat-N to every slide in a sequence, or picking numbers out of order (cat-1, cat-4, cat-2). The point of cat-1…cat-8 is that each slide in the SET reads as its own step — assign them in order, one number per slide, not the same number repeated or shuffled.

Authoring

<!-- _class: split-panel -->

`Eyebrow context`

## Headline that anchors the panel.

One-sentence framing paragraph explaining what the points cover.

- First point
  - Supporting detail explaining the first point.
- Second point
  - Supporting detail explaining the second point.
- Third point
  - Supporting detail explaining the third point.

Slots

Slot	Selector	Required	Description
`eyebrow`	`p:first-of-type > code`	no	Optional inline-code label above the feature (the phase number under `steps`, the unit under `metric`).
`heading`	`h2`	yes	The featured element in the left panel — a heading by default; a hero number under `metric`; the phase name under `steps`. (Under `pullquote`, use a blockquote instead — see the variant.)
`lede`	`p`	no	One-sentence framing paragraph under the feature.
`points`	`ul > li`	yes	Right-side supporting points. Each li's lead is the point title — it renders bold automatically (no `…`); follow it with a nested `- body` line. Under `proof` there are exactly THREE items and the FIRST is the scenario signal (its lead is the label, e.g. "You know you're here when"); the other two render as the paired proof cards.

Anatomy

┌─────────────────────────────────────────┐
│  header                                 │
│  ┌────────────┐  FINDINGS               │
│  │ BRIEF      │  │ Finding title        │
│  │ heading    │  │ body detail          │
│  │ + lede     │  │ Finding title        │
│  │            │  │ body detail          │
│  └────────────┘                         │
│  footer                           1/19  │
└─────────────────────────────────────────┘

Variant decision rule

default (no modifier) — A thesis heading deserves the panel and the right column substantiates it with prose points — the plain briefing look.
metric — A hero number is the featured element — the panel flips light and the number becomes the display type.
pullquote — The featured element is a verbatim quotation — author a blockquote in the left panel instead of a heading.
steps — The panel anchors a numbered phase rather than a heading, and the right column is a numbered sequence rather than loose points.
watermark — You want a decorative accent panel — an oversized letterform behind the heading — plus an optional two-line Audience/Intent metadata footer after the points.
mirror — Same anatomy, but the deck's reading rhythm wants the featured panel to land on the right instead of the left.
qr — A URL to scan supplements the panel — a bullet tagged qr (or a bare URL) auto-resolves into a QR figure appended to the RIGHT (supporting) column; the left panel keeps its normal required heading/lede, it doesn't become the QR.
cat-1 — Rarely — the tint is ASSIGNED automatically from the slide's position in the run of proof slides, so you normally write nothing. Reach for an explicit cat-N only to pin one slide to a specific hue (or to repeat a hue deliberately). A pinned slide still counts in the sequence, so it does not shift the slides after it.
capstone — This is the SEQUENCE's last, most-earned entry — the claim is proven, not just illustrated. Write split-panel capstone: it implies proof (same three-item shape) and joins the same automatic tint sequence. Swaps the signal callout for a quoted card and the checkpoint cards for top-rule pillars.

Variants

`metric` — metric

Light panel behind one hero number.

<!-- _class: split-panel metric -->

`split-panel metric`

## 16<em>wpi</em>

Words per item — the budget this layout holds its supporting column to.

- The panel flips light
  - metric flips the panel light behind one hero number.
- The number claims
  - Keep support brief — two items beside a figure.

When to use

When not to use

Common mistakes

Authoring

Slots

Anatomy

Variant decision rule

Variants

metric — metric

pullquote — pullquote

steps — steps

watermark — watermark

proof — proof

capstone — capstone

mirror — mirror

qr — qr

cat-1 — cat-1

cat-2 — cat-2

cat-3 — cat-3

cat-4 — cat-4

cat-5 — cat-5

cat-6 — cat-6

cat-7 — cat-7

cat-8 — cat-8

Related

`metric` — metric

`pullquote` — pullquote

`steps` — steps

`watermark` — watermark

`proof` — proof

`capstone` — capstone

`mirror` — mirror

`qr` — qr

`cat-1` — cat-1

`cat-2` — cat-2

`cat-3` — cat-3

`cat-4` — cat-4

`cat-5` — cat-5

`cat-6` — cat-6

`cat-7` — cat-7

`cat-8` — cat-8