getdeck
← Back to listcomparison-experiment-readout
✦layout.md

Comparison - Experiment Readout.

An evidence-grade comparison for reading out an A/B experiment. The hypothesis takes the title slot, a meta strip carries the window, traffic split and primary metric, and the control and variant values stand left and right with their sample sizes while lift, probability of winning and expected loss stack inside the empty center gutter. Below a full-bleed rule sit the verdict badge, the next action and two guardrail rows.

🔗 Markdown URL

https://getdeck.co/api/md/comparison-experiment-readout--cobalt-grid.md?lang=en&key=YOUR_KEY

Fetching your key...

⚡ MCP install command

Client

claude mcp add --transport http getdeck https://getdeck.co/api/mcp --header "Authorization: Bearer gd_..."

Fetching your key...

Paste this into your terminal to connect your local AI agent to getdeck.

Upgrade to Pro for more powerful MCP access →

02

Preview

01

Overview

Best for: Recurring experiment readouts for product and growth teams, decision forums that need the verdict and its evidence together, and any readout that has to survive the questions that follow.

Use: When reporting an A/B experiment and handing over a ship, kill or inconclusive verdict with its evidence. Sample sizes, window, probability of winning and guardrails all belong on the same slide so the readout survives Q&A.

Avoid: For persuasive stat duels use comparison-stat-face-off; for a before-and-after narrative on one population use comparison-before-after; for weighted subjective scoring use comparison-decision-matrix.

CategoryComparison
DensityMedium
Text Budget330 chars
EmphasisStat
PositionMiddle
03

Theme

Based on the currently selected theme Cobalt Grid. Switching the theme in the sidebar updates this section too.

SchemeLight
Best forAcademic reports, trend reports, design research, agency capabilities
Avoid forCasual, colorful presentations
04

Slots (24)

eyebrow
Label
Top label (ex: EXPERIMENT READOUT)22 chars
hypothesisRequired
Aa
The hypothesis in one line; it takes the title slot (ex: merging checkout into one screen lifts conversion)72 chars
meta_periodRequired
Label
Experiment window (ex: 2026-09-01 ~ 09-21)20 chars
meta_trafficRequired
Label
Traffic split (ex: 50 / 50)16 chars
meta_metricRequired
Label
Primary metric name (ex: checkout conversion)22 chars
control_labelRequired
Label
Control arm name (ex: Control A)14 chars
control_valueRequired
128%
Control primary metric value (ex: 3.10%)7 chars
control_nRequired
Label
Control sample size (ex: n = 24,180)14 chars
variant_labelRequired
Label
Variant arm name (ex: Variant B)14 chars
variant_valueRequired
128%
Variant primary metric value; rendered in the accent color (ex: 3.74%)7 chars
variant_nRequired
Label
Variant sample size (ex: n = 24,066)14 chars
lift_valueRequired
128%
Relative lift with its sign (ex: +20.6%)8 chars
lift_labelRequired
Label
Lift caption (ex: Relative lift)12 chars
confidence_valueRequired
128%
Probability of winning or significance level (ex: 94%, p = 0.03)8 chars
confidence_labelRequired
Label
Caption for the probability value (ex: Probability variant wins)20 chars
downside_value
Label
Expected loss if the call is wrong (ex: -0.2%p)12 chars
verdict_badgeRequired
Label
The verdict; ship, kill and inconclusive are all expressible (ex: Ship, Kill, Retest)10 chars
verdict_noteRequired
What happens next, in one line (ex: roll out to all traffic next release)86 chars
guardrail1_name
Label
Guardrail metric 1 name (ex: Bounce rate)18 chars
guardrail1_delta
Label
Guardrail metric 1 change (ex: -0.1%p)10 chars
guardrail2_name
Label
Guardrail metric 2 name (ex: Response time)18 chars
guardrail2_delta
Label
Guardrail metric 2 change (ex: No change)10 chars
companyRequired
Label
Company name20 chars
page
Label
Page number5 chars
05

Style Knobs (3)

Verdict badge emphasisvariant
보더 배지글자만
data-variant보더 배지 / 글자만
Control / variant value sizefontSize
Aa
font-size80–120px (step 4)
Center guttervariant
세로선 동반빈 거터
data-variant세로선 동반 / 빈 거터
06

Used in these deck templates

Not used in any deck template yet.