Core Concepts

Understand the key concepts behind m0rphic: zones, experiments, variants, and the AI optimization loop.

Rendering Units (Zones)

A Zone is an area of your UI that m0rphic manages. It can be an entire page or a specific component section. Zones are the atomic units that m0rphic renders and optimizes.

typescript

Note

Each zone operates independently. The AI can optimize different zones for different goals without affecting each other.

Experiments

An Experiment is a test comparing different UI variations. Experiments can be created automatically by AI or manually configured.

Experiment Types

Component Experiments

Test different component implementations (HeroA vs HeroB)

Layout Experiments

Test different arrangements of the same components

Segment Experiments

Target specific user segments with tailored experiences

Experiment Lifecycle

DraftRunningPausedCompletedArchived

Variants

A Variant is a specific UI configuration within an experiment. Each experiment has a control variant (baseline) and one or more treatment variants.

typescript

Tip

Users are consistently assigned to the same variant throughout an experiment. This assignment is done server-side to ensure consistency across devices.

AI Hypotheses

The AI generates Hypotheses based on user behavior data. A hypothesis is a prediction about why a variant might perform better.

"Users on mobile devices are 40% less likely to click the current CTA button. A larger button with higher contrast may improve mobile conversion rates."

Confidence: 78%

Hypotheses are validated through experiments. If data supports the hypothesis, the winning variant becomes the new default.

Insights

Insights are observations derived from analyzing user behavior. They highlight components that are underperforming or opportunities for optimization.

typescript

The Optimization Loop

m0rphic continuously optimizes your UI through a closed-loop system:

  1. 1

    SDK Renders UI

    Zone components are rendered based on current configuration

  2. 2

    Collect Behavior Data

    Clicks, views, scroll depth, time spent are automatically tracked

  3. 3

    AI Analysis

    ClickHouse data is analyzed to identify patterns and opportunities

  4. 4

    Generate Hypotheses

    AI creates testable hypotheses about potential improvements

  5. 5

    Run Experiments

    Variants are served to users and measured against metrics

  6. 6

    Learn & Deploy

    Winners are deployed, losers are discarded, and the cycle repeats

Metrics & Goals

Experiments are evaluated against configurable metrics:

Click Rate

Percentage of users who click

Conversion Rate

Percentage completing a goal action

Completion Rate

Percentage finishing a flow/form

Time on Page

Engagement duration

Bounce Rate

Percentage leaving immediately

Custom Metrics

Define your own success criteria

Composite Metrics

m0rphic supports weighted composite metrics. The AI can automatically infer optimal weights based on your product type (e-commerce, SaaS, content, etc.).

Auto vs Manual Mode

m0rphic supports two operational modes:

Auto Mode (Default)

  • • AI creates experiments automatically
  • • Winning variants deploy without approval
  • • Minimal human intervention required
  • • Best for high-traffic, low-risk pages

Manual Mode

  • • AI proposes experiments for review
  • • Human approval required before deployment
  • • Full control over changes
  • • Best for critical conversion flows

Mode can be configured per zone, allowing auto-optimization for low-risk areas while maintaining control over critical paths like checkout.