Dashboards & Presentations Playbook v1

A focused companion to the Effects Reference Matrix and the Email CTA Playbook. Covers BI dashboards, analytics interfaces, and presentation decks — with platform-specific notes for the marketing reporting stack (GoHighLevel, Braze, Emarsys, Iterable, HubSpot, Klaviyo).

This doc serves two audiences:

  1. The builder agent — design rules for generating dashboards and decks as builder outputs
  2. The team's internal work — best practices for the actual reporting pipelines and client decks

Part 1 — Dashboards / BI / Analytics

Why dashboards are different

Dashboards are the only output type in the matrix where the design loses if it's pretty but doesn't answer the question. Every other channel (social, email, landing pages) has decoration as part of the job. Dashboards do not. The job is: enable a decision in 5 seconds or less.

Two layers of rules apply:

  1. Information design rules — what data, in what order, with what hierarchy
  2. Visual design rules — how the chrome, type, color, motion, and effects support (or undermine) the information

Most dashboards fail at layer 1 long before layer 2 matters.

The 5-second test

If an executive can't grasp the dashboard's headline in 5 seconds, the design has failed regardless of how beautiful the charts are.

Test method: show the dashboard to someone unfamiliar with the data for 5 seconds, then close it and ask "what did you see?" If they can't name the primary KPI and its direction (up/down/flat), redesign.

Implication for the builder: every dashboard output needs a hero KPI region at the top — the single most important number, with delta vs comparison period, in the largest typographic weight on the page.

The action test for every KPI

Every metric on a dashboard must answer: "If this number changes, does someone take an action?"

If no, remove it. "Nice to know" metrics are clutter that dilutes the actionable ones. This is the single highest-leverage cull when auditing existing dashboards.

Common metrics that fail the action test in marketing dashboards:

Common metrics that pass:

Dashboards vs reports — architecture rule

Per EPC Group's enterprise BI guidance, the correct architecture is:

Reports first. Dashboards second. Build comprehensive reports with full slicer interaction, drill-through, bookmarks, conditional formatting. Then pin the most important visuals from those reports to a dashboard that serves as the single executive entry point.

For the builder: outputs should be structured as dashboard + linked report, not a single sprawling dashboard. The dashboard is the headline; the report is where you investigate.

Information hierarchy on the canvas

Reading patterns dictate placement:

The grid below works for ~80% of marketing performance dashboards:

┌────────────────────────────────────────────────────────────────┐
│  Hero KPI 1        Hero KPI 2        Hero KPI 3        Hero KPI 4  │
│  + delta           + delta           + delta           + delta      │
├────────────────────────────────────────────────────────────────┤
│  Filters / Date range                                              │
├──────────────────────────────────┬─────────────────────────────┤
│                                  │                                │
│  PRIMARY CHART (line, trend)     │  Secondary chart (breakdown)   │
│                                  │                                │
├──────────────────────────────────┼─────────────────────────────┤
│  Detail table                    │  Tertiary chart / list         │
│                                  │                                │
├────────────────────────────────────────────────────────────────┤
│  Data freshness · last updated · data source                       │
└────────────────────────────────────────────────────────────────┘

Chart selection — the working taxonomy

The single biggest mistake in dashboard design is picking the wrong chart for the data. Below is the working decision tree.

What's the question?

Question Best chart When NOT to use
What's happening over time? Line chart Stop if you only have 2–3 time points — use bar instead
How do categories compare? Vertical bar Stop if you have >10 categories — switch to horizontal bar
How do many things rank? Horizontal bar Stop if data is time-series — switch to line
How does the whole break down? Stacked bar (over time) or treemap (single point) Avoid pie/donut beyond 5 segments
How do two variables relate? Scatter plot Stop if relationship is purely categorical — switch to grouped bar
Where's the density / intensity? Heatmap Avoid for <5×5 cells — too sparse
How does the funnel convert? Funnel chart Stop if all stages don't share the same denominator — use Sankey instead
How are users retaining? Cohort heatmap This is the only chart that does this well — use it
What's the single most important number right now? KPI card (big number + delta) Always pair with comparison period
Where did revenue come from / go? Waterfall chart Avoid if you have >7 contributing buckets
What's the distribution shape? Histogram or box plot KPI averages hide distribution — show the shape
What's the micro-trend in tight space? Sparkline Avoid if exact values matter — sparkline reads shape only
How are A and B paths flowing? Sankey diagram Avoid for >4 stages — gets unreadable
What's actual vs target? Bullet chart Better than gauge; avoid gauges entirely
What's the geographic breakdown? Map (choropleth) Only when geo matters — bar chart of countries usually communicates faster

Charts to mostly avoid:

Marketing dashboard core patterns

Five patterns that should be baked-in templates for the builder:

Pattern 1: Executive summary (the "hero dashboard")

Pattern 2: Channel performance

Pattern 3: Cohort retention

Pattern 4: Campaign / flow performance (Klaviyo-style)

Pattern 5: Send-time / engagement heatmap

KPI card anatomy

The most important component in any dashboard. Six elements:

  1. Label — what the metric is (small caps, light weight)
  2. Big number — the current value (large weight, ~48–72px)
  3. Delta — change vs comparison period (colored, arrow icon, % or absolute)
  4. Comparison reference — what the delta compares to ("vs last week" / "vs target")
  5. Sparkline — micro-trend over recent period (subtle, behind or below number)
  6. Threshold indicator — color-coded background or border when above/below target
┌─────────────────────────────────┐
│  REVENUE                        │  ← label
│                                 │
│  $42,318                        │  ← big number
│  ↑ 12.4% vs last week           │  ← delta + reference
│                                 │
│  ╱╲╱╲___╱╲___╱╲                │  ← sparkline
└─────────────────────────────────┘

Optional: a "click to drill down" affordance (subtle arrow icon, hover state).

Color systems for data visualization

Three palette types are required for any dashboard system:

Semantic palette — fixed meanings, never repurposed

Sequential palette — single-hue progression (light → dark) for ordered data

Categorical palette — 5–8 distinct colors with no ordering implied

Color rules that always apply:

Dashboard motion & effects

Motion in dashboards must serve comprehension, never decoration. Eight effect patterns that pass the test:

Effect When to use Implementation note
Number counter animation on load KPI cards on first paint 600–800ms ease-out, snap to value, never loop
Sparkline draw-in Sparklines accompanying KPI cards stroke-dasharray animation, 400–600ms
Skeleton loading All async-loading components Subtle shimmer gradient, never spinner
Smooth transitions on filter change Bar reorder, chart updates 300–400ms ease-out, use FLIP technique
Hover tooltips Charts where exact values matter Anchor to data point, follow cursor on dense data
Period overlay Comparison charts Ghost/dotted line for previous period at 40% opacity
Anomaly highlight Outlier values on charts Subtle pulse (1.5s loop, very low intensity) or color shift
Drill-down breadcrumb After clicking into detail Animated breadcrumb appears, smooth transition between views

Effects to avoid:

Loading, empty, and error states

These are the three states dashboards always botch. Spec each:

Loading state

Empty state

Error state

Mobile responsiveness for dashboards

Most BI tools botch mobile. The rules:

  1. Don't shrink desktop dashboards — re-layout entirely
  2. Stack KPI cards vertically at narrow widths, not 2-up squashed
  3. Hide secondary charts behind a "More" expand
  4. Always show the hero KPI above the fold on mobile
  5. Drill-down via tap to a dedicated detail view, not a hover tooltip
  6. 48px minimum tap targets (matches mobile button standards)
  7. Sticky filter bar stays visible during scroll

Dark mode for dashboards

Dashboards are the channel where dark mode is most expected (especially for tech audiences). Defaults:

Light mode equivalents:

Trending in BI 2026

What's actually getting deployed in 2026 (verified across Gartner, EPC Group, Power BI roadmaps, B EYE):

  1. AI-generated insights inline — Power BI Copilot, Tableau Pulse, Looker Studio Gemini integration. Natural language summaries of what changed and why, surfaced next to the chart.

  2. Natural language query interfaces — "show me revenue trends by region for Q3" generates the chart. Power BI Copilot replaces the older Q&A feature in December 2026.

  3. Dashboard-optional analytics — analytics shows up in Slack/Teams alerts, embedded in product UI, in spreadsheets, in chat interfaces. Dashboards become one delivery format among many, not the default.

  4. Embedded analytics — white-label dashboards inside SaaS products. Klaviyo, HubSpot, Iterable all moved toward this in 2025–2026.

  5. Real-time vs near-real-time honesty — explicit "last updated 2 min ago" timestamps; many "real-time" dashboards were actually 5–15 min lag, which becomes a trust issue when decisions get made.

  6. Annotation/storytelling layers — Tableau Story, Hex notebooks, Mode reports — dashboards that walk the reader through a narrative rather than presenting a wall of charts.

  7. Cohort retention as standard — every product/marketing dashboard now expects a cohort view. Used to be optional.

  8. Mobile-first — Linear and Vercel dashboards set the bar for mobile-first analytics UX in 2025.

  9. Less chart junk, more whitespace — Tufte-influenced minimalism returning. Removing gridlines, axis ticks, and chart borders is now common in modern dashboards.

ESP-specific notes (the platforms you actually ship from)

Klaviyo

HubSpot

Braze

Emarsys

Iterable

GoHighLevel

Builder presets to ship for Bi channel

Dashboard templates (10)

  1. Executive summary (4-KPI hero + trend + breakdown)
  2. Channel performance (multi-channel attribution)
  3. Cohort retention (heatmap + curve overlay)
  4. Campaign/flow performance (Klaviyo-style)
  5. Send-time / engagement heatmap
  6. Funnel analysis (conversion stages with drop-off rates)
  7. A/B test results (variants, lift, significance, revenue impact)
  8. Pipeline / lifecycle (HubSpot-style stage progression)
  9. Revenue waterfall (attribution by source over period)
  10. Real-time monitor (live metrics, refresh every N seconds)

Component presets (12)

  1. KPI card (number + delta + sparkline + threshold color)
  2. Trend line chart (with comparison overlay)
  3. Stacked bar (time-series composition)
  4. Horizontal bar (ranking)
  5. Cohort heatmap
  6. Sankey diagram (flow visualization)
  7. Funnel chart
  8. Scatter plot (correlation with bubble sizing)
  9. Bullet chart (actual vs target)
  10. Sparkline (inline micro-trend)
  11. Geographic choropleth (when geo matters)
  12. Data table with conditional formatting, sortable columns, frozen header

State presets (3)

  1. Skeleton loading (matches final layout)
  2. Empty state (with action prompt)
  3. Error state (with retry + freshness fallback)

Part 2 — Decks / Presentations

Why decks are different

Presentations are simultaneously the most-built and most-poorly-built design output. The trap: every PowerPoint, Keynote, and Google Slides ships with templates that encourage 6 bullet points per slide and 12-point font. Most decks die at the template level.

The job of a good deck: deliver one idea per slide, with the slide title carrying the takeaway, supported by visual evidence. Everything else is decoration.

Two layers of rules:

  1. Narrative design rules — what slides, in what order, with what arc
  2. Visual design rules — typography, color, charts, motion that support the narrative

The narrative test

Show the deck with the body content hidden but the titles visible. If the titles alone tell the story, the deck is well-structured. If they read as a list of topics ("Q3 Results," "Next Steps," "Challenges"), the deck is broken at the structural level.

The fix: title-as-message. Each slide title is a sentence that delivers the takeaway:

The "show, don't tell" rule for data slides

For data-driven presentations (Marita's monthly performance decks):

Most data decks fail by reversing this — body text restates what the chart shows.

Deck taxonomy

Twelve deck types worth recognizing as builder templates:

Deck type Length Tone Primary goal
Investor pitch 10–15 slides Confident, data-led Funding decision
Sales deck 8–12 slides Persuasive, customer-centered Demo-to-close
Monthly performance report (Marita's main) 15–25 slides Evidence-led, candid Client retention via demonstrated value
Internal proposal 5–10 slides Recommend-led Decision authorization
Webinar deck 30–60 slides Educational, paced Audience engagement + conversion
Conference talk 20–40 slides Story-led Audience impact + brand
All-hands / team update 8–15 slides Direct, informational Team alignment
Onboarding deck 20–40 slides Friendly, structured New-hire enablement
Training deck 30–60 slides Methodical, step-based Skill transfer
Case study deck 8–15 slides Narrative, evidence-led Marketing/sales asset
Product demo 10–20 slides Show-and-tell Product clarity
Strategic review 25–40 slides Analytical Plan approval

Slide type taxonomy (15 patterns the builder needs)

1. Title slide

Cover. Client/deck name + period + presenter. One visual anchor (logo, hero image, gradient).

2. Section divider

Visual pause. Single phrase or section number. Use these — they create rhythm. Marita's monthly decks need at least 3.

3. Single-stat hero

One enormous number + one-line context. Use for the headline KPI ("$1.2M revenue this month, +18% MoM").

4. Two-column compare

Before/after, this/that, problem/solution. 60/40 or 50/50 split.

5. Three-column feature

Three benefits, three features, three findings. Equal-weight grid.

6. Bento grid summary

Asymmetric grid showing multiple data points or features at once. Replaces the dreaded "executive summary slide with 8 bullets."

7. Quote / testimonial

Single big quote, attribution below. Soft shadow card on subtle background.

8. Process / timeline

Horizontal flow of stages. Linear arrow connection between steps.

9. Org chart / team

Photo grid with names and roles. Avoid hierarchical org charts in client-facing decks (reads as bureaucratic).

10. Pricing / plans table

3-column comparison with recommended plan highlighted.

11. Roadmap

Quarterly or timeline-based future view. Avoid specific dates unless you can commit; use Q1/Q2/Q3/Q4 framing.

12. Chart-led slide

Title carries takeaway, chart fills the canvas. The most important pattern for monthly performance reports.

13. Quote + chart

Customer quote on one side, supporting data on the other. Bridges narrative and evidence.

14. Q&A / discussion prompt

Used at section breaks in long decks or as the close. "What questions do you have about this?" or "Three things to discuss next."

15. Closing / contact

Final slide. Contact info, next steps, calendar booking link.

Typography for decks

The single most-controlled variable in presentation design. Rules:

  1. One display font + one body font, max two total
  2. Display font: high-contrast, bold weight available, distinctive enough to anchor the design but readable at 40–72pt
  3. Body font: high readability at 18–24pt, system or web-safe to avoid font-loading issues on export
  4. Font scale:
    • Slide title: 36–48pt
    • Section title: 28–36pt
    • Body: 18–24pt
    • Caption / footnote: 12–14pt
  5. Line length: never more than 60 characters per line on a slide
  6. Line height: 1.3–1.5 for body, 1.1–1.2 for display

Suggested pairings for 2026:

Color systems for decks

Apply the same rigor as data viz:

Deck design trends 2026

Verified across Slidesgo, SlideBazaar, SlideSalad, Gamma trend reports:

  1. Editorial-magazine layouts — heavy serif headlines, asymmetric grids, big imagery
  2. Big bold typography as the design — type doing the heavy lifting instead of illustration
  3. Soft layered shadows on cards — Linear-style depth
  4. Subtle mesh gradients as section backgrounds
  5. Real product screenshots > stock illustrations
  6. Bento grids on summary slides
  7. Number counter animations (Tome / Gamma-style on slide reveal)
  8. Dark mode for tech decks — Anthropic, Linear, Vercel set the pattern
  9. Charts with annotation layers — callouts, arrows, highlighted regions on data
  10. Retro-futurism — chrome type, sci-fi gradients (specific to tech / launch decks)
  11. Tech monochrome — single-hue palettes with one accent, very disciplined
  12. Organic flowing shapes — soft blob shapes as background accents (Slidesgo's "Analog Dreams" trend)
  13. Luminous Fade — soft gradient backgrounds with high-contrast type (Slidesgo's 2026 named trend)
  14. AI-generated illustrations — used carefully; stylized, not generic

Trends to skip in decks

Motion in decks

Restrained motion patterns that pass the "doesn't distract" test:

Pattern When to use Avoid
Cross-fade between slides Default transition for cohesive decks Push, cube, reveal — all read as 2010
Mask wipe Section divider transitions Use sparingly, max 2x per deck
Sequential build (stagger reveal) Reveal complex chart in steps Don't animate every bullet — kills pacing
Number counter on reveal Single-stat hero slides Snap to value, never loop
Highlight pulse on annotation Drawing attention to chart callout Single pulse, low intensity, on slide reveal only
Background gradient shift Mesh gradient slides Very slow (10–15s loop), subtle
Pre-recorded animations (Lottie/Rive) Product demos, process explanations Heavy; only when illustration value is clear
Scroll-style transitions (Tome/Gamma) Web-native decks Don't try in PowerPoint exports — breaks

The monthly performance report template (Marita's bread-and-butter)

A specific, opinionated 20-slide template optimized for monthly email performance reports across Klaviyo, HubSpot, Braze, Emarsys, Iterable, GHL clients.

Slide-by-slide structure

  1. Title

    • Client name + period (e.g., "Hansø Home · October 2025 Performance Review")
    • Date, presenter, deliverable status

  2. Executive summary (the most-read slide)

    • 4-up bento grid of hero KPIs: Revenue, Click rate, List growth, Conversion rate
    • Each with delta vs previous month + delta vs YoY same month
    • One-line takeaway in title

  3. Section divider: "What worked"

  4. Top win #1 — biggest revenue driver

    • Title = the takeaway
    • Chart = the proof
    • One-line note = the why

  5. Top win #2 — best-performing flow or campaign

  6. Top win #3 — segment or experiment win

  7. Section divider: "Where we're at risk"

  8. Risk #1 — usually deliverability, list health, or a declining flow

    • Title = the takeaway (the specific risk)
    • Chart = the trend
    • One-line note = the recommendation

  9. Risk #2 — segment or campaign underperformance

  10. Section divider: "Campaign performance"

  11. Campaign overview chart

    • Horizontal bar ranking campaigns by revenue per recipient
    • Annotation: highlight top performer + outlier underperformer

  12. Campaign deep-dive — winning campaign breakdown

    • Subject line, send time, segment, performance vs benchmark

  13. Section divider: "Flow performance"

  14. Flow overview chart

    • Stacked bar of flow revenue by flow type (welcome, abandoned cart, browse abandon, etc.)
    • Comparison to previous month

  15. Flow deep-dive — most improved or most degraded flow

    • Conversion stages, where the drop-off is, recommendation

  16. Section divider: "A/B test learnings"

  17. Test results panel

    • Tested element, variants, lift, statistical significance
    • One-line takeaway per test (what we learned)

  18. Section divider: "Plan for next month"

  19. Plan slide

    • 3 priorities in bullet form (not 8 — three)
    • Each with expected impact and owner

  20. Closing / contact

    • "Next review: [date]"
    • Contact for questions

Builder considerations

Tool landscape for decks (Marita has Gamma already)

Tool Sweet spot Limitation
Gamma AI-native, fast generation from outline, web-native sharing Web-first; PPTX export can drift visually; templates can look samey without input
Pitch Team collaboration, professional templates, sales tracking More manual; AI generation feature is secondary
Tome AI narrative generation, modern aesthetic Limited editing precision once generated
Beautiful.ai Structured templates that prevent design mistakes Manual content; less AI
Canva Asset library, design flexibility Less data integration
PowerPoint Universal compatibility, enterprise standard Default templates encourage bad design
Keynote Apple ecosystem, smooth animations Mac-only, weaker collaboration
Google Slides Browser-native, ubiquitous collaboration Limited animation, basic visual tools
Figma Slides Designer-team friendly, component-based Newer, smaller template library

For Marita's workflow: Gamma for fast generation of outline-stage decks; builder agent could output to Gamma format directly via Gamma's MCP. PowerPoint export for clients who insist on editing in PPT.

Trending styles to watch in 2026 (specific named aesthetics)

From Slidesgo's 2026 trends (verified across multiple sources):

Builder presets to ship for Dk channel

Deck templates (12)

  1. Monthly performance report (the Marita template above)
  2. Quarterly business review (longer, strategic)
  3. Investor pitch deck (10-slide standard)
  4. Sales deck (8-12 slide problem/solution/proof/close)
  5. Client onboarding deck
  6. Case study deck (problem/approach/results/quote)
  7. Webinar deck (30+ slides, paced)
  8. Conference talk template
  9. All-hands update
  10. Product demo
  11. Strategic proposal
  12. Email program audit report (specific to her email work)

Slide type presets (15)

  1. Title slide (cover)
  2. Section divider
  3. Single-stat hero
  4. Two-column compare
  5. Three-column feature
  6. Bento grid summary
  7. Quote / testimonial
  8. Process / timeline
  9. Org chart / team
  10. Pricing table
  11. Roadmap
  12. Chart-led slide (title + full-canvas chart)
  13. Quote + chart hybrid
  14. Q&A / discussion prompt
  15. Closing / contact

Style presets (8)

  1. Editorial-Magazine
  2. Tech Monochrome (dark)
  3. Tech Monochrome (light)
  4. Luminous Fade
  5. Bento Bold
  6. Analog Dreams
  7. Corporate Considered (for Co channel)
  8. Service-Trust (for Sv channel)

Output formats

  1. PDF (universal sharing)
  2. PPTX (client editing)
  3. Web link / embed (live presentation)
  4. PNG sequence (for image-based email integration)
  5. MP4 (for social repurposing)
  6. 9:16 vertical slices (for social repurposing)

Part 3 — Cross-cutting principles

When dashboard meets deck

Marita's monthly performance reports are dashboards delivered as decks. The hybrid pattern is its own thing:

Build the dashboard first. Then extract the story from the dashboard into the deck. The deck doesn't reproduce the dashboard — it tells the dashboard's story.

For the builder: the same data pipeline should be able to render both. A dashboard view (live, exploratory) AND a deck view (narrative, exportable) from the same underlying data.

Data-to-narrative pipeline

The agent-buildable workflow:

  1. Pull data from ESP (Klaviyo, HubSpot, Braze, Emarsys, Iterable, GHL) via API
  2. Compute KPIs with comparison periods (MoM, YoY) and benchmark deltas
  3. Identify the story — biggest movers (positive and negative), surprising findings, anomalies vs threshold
  4. Render dashboard for live exploration
  5. Generate deck draft with auto-populated slides (executive summary, top wins, risks, plans)
  6. Human-in-the-loop edit — Marita reviews titles, adjusts takeaways, adds context the data can't tell
  7. Export to the appropriate format(s) for the audience

This is the most valuable thing the builder agent can do for her business specifically.

Accessibility for both formats

Apply across both dashboards and decks:

Internationalization

For dashboards and decks shipped to clients in multiple regions:

Quality bar checklist before shipping any dashboard or deck

10-item pre-flight checklist:

  1. 5-second test passed (dashboard) or title-only narrative test passed (deck)
  2. Every KPI / slide passes the action test — leads to a decision
  3. Comparison periods present on all metrics (no naked numbers without context)
  4. Data freshness timestamp visible (dashboards) or data period clear (decks)
  5. Color palette accessibility verified (color blindness, contrast)
  6. Dark mode tested if applicable to audience
  7. Mobile responsive verified (dashboards) or vertical export available (decks)
  8. Loading / empty / error states designed (dashboards)
  9. One typeface family + one accent (decks) or single style system (dashboards)
  10. Export formats verified — PDF, PPTX, image, embed depending on channel

Part 4 — What this means for the builder agent

The agent building outputs for this channel needs to:

For Bi (dashboards) outputs

  1. Detect data type — time-series, categorical, geographic, distributional — and select chart type from the taxonomy
  2. Compute comparison periods automatically — MoM, WoW, YoY, vs target — and show deltas
  3. Apply semantic colors — green/red/amber/blue with explicit meanings
  4. Generate KPI cards with all six anatomy elements (label, number, delta, reference, sparkline, threshold color)
  5. Detect anomalies and apply subtle highlighting
  6. Render skeleton loading states matching the final layout
  7. Output in three modes: light dashboard, dark dashboard, print-ready
  8. Honor data freshness — surface timestamp explicitly, never hide stale data

For Dk (decks) outputs

  1. Generate title-as-message titles from data — never "Q3 Results," always "Q3 revenue grew 18% driven by new flows"
  2. Select slide type from data — single-stat slide for hero KPI, chart-led slide for trends, bento grid for multi-point summaries
  3. Apply one consistent style preset across the entire deck — no mixing
  4. Auto-generate speaker notes with editable override
  5. Render section dividers at appropriate narrative breaks (every 4–6 slides)
  6. Output in all formats: PDF, PPTX, web link, PNG sequence, MP4, 9:16 social slices
  7. Provide one-click variants: vertical for social, light/dark, with-notes/without-notes

For the data-to-narrative pipeline

  1. ESP connectors — Klaviyo (REST API), HubSpot (CRM API), Braze (Currents export), Emarsys (REST API), Iterable (REST API), GHL (REST API)
  2. Data normalization layer — common schema across ESPs so the same template renders regardless of source
  3. Story detection — identify biggest movers, anomalies, threshold violations automatically; surface to human reviewer
  4. Human-in-the-loop checkpoint — never auto-publish, always present for review before export
  5. Versioning — every report version stored; ability to compare period-over-period reports

Closing notes

The Effects Reference Matrix, Email CTA Playbook, and this Dashboards & Presentations Playbook together cover the full output range of a modern marketing-design builder. Reading order for the agent:

  1. Matrix — what effects are available per channel + reality checks per channel
  2. Email CTA Playbook — conversion-specific patterns for the three email archetypes
  3. This doc — dashboards and decks, which have their own design languages and don't fully overlap with anything in the matrix

For the team's internal workflow (vs the builder output side): apply the dashboard rules to your own reporting pipelines first. The 5-second test and the action test will cull more clutter from your existing dashboards than any new tool will add. The monthly performance report template above is opinionated; treat it as a starting point and adjust per client.

Last verified against current trend reporting and platform documentation: May 2026. Companion to: Effects Reference Matrix v2.2 and Email CTA Playbook v1.