Four densities, rich formatting, image systems, data stories, and useful interaction.
Detailed contents · select any page or press M anywhere
Choose the exact depth you need.
System overview
One fixed canvas. Many reading speeds.
4Density modes
Big idea through paper-like research.
9Image patterns
Distinct geometries with explicit purposes.
12+Data forms
Charts, flows, tables, and sequences.
3Interactions
Explore, allocate, and simulate.
Density system · exact scale comparison
Default
Clear argument
The everyday projected slide. It explains one claim in enough detail to understand immediately, while preserving room for a meaningful visual, diagram, or comparison.
Display 5.15 · Title 3.65Body 1.46 · 20-45 words
Big idea · Inspire
Big idea.
A keynote beat with very few words. Use it when the audience should feel, remember, or repeat one statement before the story moves on.
Display 7.20 · Title 5.15Body 1.62 · 0-18 words
Dense
Shared-screen analysis
Structured comparisons, operating plans, diagrams, and compact tables that still work during projected discussion. Density comes from organization, not tiny labels.
Display 4.30 · Title 2.85Body 1.18 · 45-90 words
Research
Close-reading evidence
A compact paper page with columns, citations, methods, qualifications, definitions, and backup evidence. Type remains comfortably readable while evidence volume and structural density increase.
Display 3.65 · Title 2.50Body 1.06 · 100-190 words
Type system
Hierarchy is a reading path.
Switch the whole density pack. Never rescue one crowded element with an isolated tiny font.
DisplayAa5.15
TitleArgument3.65
LedeSupporting proposition2.05
BodyReadable explanation1.46
SmallLabels and secondary copy1.06
CaptionSource · method · credit.78
Format expansion
Emphasis has more than color.
Give weight, italic, underline, foreground color, background highlight, and palette roles stable jobs. Combine at most two treatments on one phrase.
Weight
Regular · Medium · Bold
Voice
Roman · Italic contrast
Underline
Persistent emphasis
Marker underline
Draw the eye
Foreground color roles
AccentInfoPositiveWarningDanger
Color highlight
Accent field
Accent text depth
PaleBaseStrong
Accent ramp
PaleSoftBaseStrong
Motion vocabulary · slow demonstration loop
Move, settle, hold, repeat.
FadeCalm continuity · 900ms
PushNarrative progression · 900ms
RiseSection reveal · 900ms
ScaleReflective beat · 900ms
420Count-up metric
Bar growth
Line drawing
Ring sweep
Row groups
Text motion vocabulary · slow demonstration loop
Text can move without becoming theatre.
FadeQuiet support
RiseDefault entrance
ScaleMeasured emphasis
PopMetric or node
WipeShort headline
MaskLine reveal
FocusReflective beat
TrackEyebrow or label
HighlightLoad-bearing phrase
Image system · nine distinct locked patterns
Hero
One full visual · atmosphere or decisive proofSplit
What precise behavior or outcome is being explained? Define the population, period, unit of analysis, inclusion rules, exclusion rules, and what evidence would reverse the current conclusion. Separate the business question from the measurement proxy.
Population and period
Primary and secondary measures
Decision threshold
Known blind spots
Finding and counterevidence
State the observed pattern separately from interpretation. Preserve the denominator and uncertainty. Record the strongest competing explanation, the best disconfirming observation, and whether missing data could change the ranking.
Primary result and effect size
Counterexample and sensitivity
Source quality and confidence
Missing-data treatment
Implication and next test
Explain what changes now, what remains unresolved, and which follow-up most efficiently distinguishes the competing explanations. Name the owner, timebox, kill condition, and the date when confidence will be reassessed.
Action and accountable owner
Timebox and stopping rule
Next discriminating evidence
Assessment date
Evidence ledger · what moved confidence
E1 · Direct observation
Repeated behavior across three cohorts; moderate source quality; supports the direction, not the mechanism.
E2 · Instrumented result
Largest measured effect; denominator stable; one missing period lowers precision but not ranking.
E3 · Counterexample
One cohort improved without the intervention; competing explanation remains plausible under low friction.
E4 · Artifact review
Versioned output confirms the workflow changed; it does not prove the change caused the outcome.
E5 · Sensitivity test
Ranking survives ±10% value changes and complete-case removal; confidence stays medium.
E6 · Open evidence
Pre-register the next comparison and inspect whether the effect persists outside the original operator group.
Source: reader-consultable primary evidence · method note · access date · last assessed date.
Appendix · detailed table
Backup analysis keeps comparable units, denominators, and confidence together.
Illustrative cohort evidence · values demonstrate research density only
Cohort
N
Signal
Cost
Latency
Retention
Missing
Confidence
Alpha
48
54
88
11 d
61%
4%
Medium
Beta
72
63
74
8 d
68%
2%
Medium
Gamma
96
72
58
4 d
79%
1%
High
Delta
36
69
67
5 d
74%
8%
Low
Epsilon
28
77
81
3 d
82%
12%
Unproven
Zeta
65
59
70
7 d
66%
3%
Medium
Eta
44
64
62
6 d
71%
5%
Medium
Theta
31
73
79
4 d
76%
9%
Low
Decision row. Gamma combines the strongest supported signal, lowest cost among proven cohorts, shortest supported latency, and minimal missingness.
High-value uncertainty. Epsilon has the highest raw signal and retention, but small N and 12% missingness make the ranking unstable.
Counterevidence. Delta approaches Gamma's signal with weaker confidence; inspect whether its lower sample hides a repeatable subgroup effect.
Next analysis. Re-estimate after one complete period, publish the confidence rubric, and test whether cost remains comparable across cohorts.
Method: cohort eligibility fixed before analysis; cost indexed to baseline 100; latency measured calendar days; confidence combines source quality, sample, missingness, and counterevidence.
Appendix · source gallery
Show evidence at inspectable size, then state what each artifact can and cannot prove.
A · ContextB · SynthesisC · ArtifactD · Outcome
A · Context. Observation setting and participants; cannot establish causality.
B · Synthesis. Analyst interpretation; trace claims to underlying records.
C · Artifact. Tangible prototype state; date and version are required.
D · Outcome. Follow-up observation; declare selection and survivorship limits.
Generated illustrative imagery, not factual evidence. Real artifacts need owner, source URL or file, capture date, access date, and permission status.
Appendix · method
Make the analytical path reproducible enough for another reader to challenge it.
CollectSources, dates, inclusion rules, exclusions, and custody.
Freeze the eligible source list
Record owner and access date
Separate primary from derived evidence
Log excluded items and rationale
Preserve version or capture hash
NormalizeComparable units, denominators, labels, and missing values.
Keep raw and transformed values
Declare currency and time basis
Resolve duplicate identities
Define missingness categories
Document every conversion
TestPrimary claim, counterexample, sensitivity, and uncertainty.
Predeclare the primary comparison
Run the strongest alternative
Inspect missing-data sensitivity
Check subgroup instability
Record what would reverse the claim
DecideThreshold, implication, owner, stopping rule, and reassessment.
Separate evidence from risk tolerance
Name the decision owner
Set a timebox and kill condition
Publish unresolved assumptions
Schedule confidence reassessment
Collection check. Another reader can locate every included source and understand why excluded evidence did not qualify.
Normalization check. Every displayed value can be traced back to its original unit and transformation rule.
Testing check. The conclusion survives declared sensitivity ranges or is labeled unstable when it does not.
Decision check. The action, owner, threshold, stopping condition, and next assessment date are explicit.
Method version 1.0 · illustrative workflow · replace with the actual procedure, code, assumptions, and source trail.
Appendix · definitions
Define the terms that could otherwise become silent disagreements.
Signal
A measured pattern that survives the declared threshold and is relevant to the decision being made. A signal describes an observation, not its presumed cause.Example: retention exceeds the predeclared baseline in three consecutive cohorts.
Evidence
A source-linked observation that can raise or lower confidence in a specific claim. Evidence retains its date, owner, method, boundary, and known limitations.Example: a versioned event log linked to the metric calculation and inclusion rule.
Confidence
The assessed strength of a claim after source quality, sample, method, missingness, sensitivity, and counterevidence. It is not the same as business importance.Example: medium confidence because the direction is stable but the mechanism remains unresolved.
Decision threshold
The minimum combination of evidence strength and risk tolerance required to choose an action. Different actions may use different thresholds against the same evidence.Example: a reversible pilot may proceed at medium confidence; a one-way commitment may not.
Latency
Elapsed time between observation, decision, test, and usable feedback. State which interval is measured and whether it uses calendar or working time.Example: four calendar days from prototype release to an interpretable retained-use result.
Core
The stable category used throughout a data story; its name, color, unit, and inclusion rule remain consistent between slides and interactive views.Example: Core remains blue in orient, trend, mix, explorer, and appendix tables.
Illustrative
A diagram, value, or image created to explain the presentation grammar rather than assert a real-world fact. It cannot be cited as product or market evidence.Example: the cohort values in this sample demonstrate layout behavior only.
Research mode
A close-screen density pack for evidence, methods, definitions, source material, backup analysis, and citations. It assumes deliberate reading rather than projection-only viewing.Example: a browser-consulted appendix with narrow columns, method notes, and source traceability.
Definitions should cite their governing source when they are not local conventions created for the deck.
