Experiments

Overview

Experiments is a planned workspace for composing structured UX research and optimization studies on a full-viewport node canvas, beyond standard heuristic evaluations.

The direction is to add test types from a panel, connect them, set order, and branch when paths diverge. A node graph is a study. Think compose and connect, closer to a whiteboard graph than a linear wizard.

Why Experiments Is Separate From Evaluations

Heuristic Evaluations stay first-class objects. The canvas is optional composition around them: not a third tab on Evaluations, and not one tab per experiment type under Evaluations.

A Heuristic Evaluation is the math unit; the Experiments canvas is optional composition around it.

How A Study Works

A study is a node graph on a full-viewport board:

• Drag a test type from the panel onto the canvas.
• Place nodes and connect them with lines to set order and flow.
• Configure each node with what that step measures or asks.
• Add conditions where the path needs to branch.

The exact panel, connectors, and condition rules may change. A study can optionally include or attach a real heuristic evaluation, but research-only graphs never need Evaluations, and audit-only work never needs Experiments.

Interconnected With The Rest Of UXit

UXit is built so the main workspaces stay linked. You can start in Guidelines, Requirements, Evaluations, Analytics, or later Experiments, move in any order, and skip surfaces that do not help the work you are doing.

Experiments is not meant to be a silo. Data you build elsewhere should be available to pull into a study, and study outcomes that map to shared records should still be readable as familiar metrics, charts, graphs, or tables outside the canvas. From there, the intent is that you can export a single result or a broader package later, using the same interconnected model as the rest of the product.

Planned Experiment Types

These types are planned, not live in beta:

• A/B Testing: Run the same goal against two or more design variants and compare which one participants complete more successfully.
• Card Sorting: Ask people to organize labels into groups so you can see how they expect content to be structured.
• Heatmaps: Map where participants click and hesitate on a screen to reveal what draws attention and what gets skipped.
• Isolate Inconsistencies: Surface places where users take different routes or hit different outcomes in what should be the same experience.
• Surveys: Capture ratings and written feedback when behavior alone does not explain preference or satisfaction.
• Task-Based Testing: Assign a concrete goal and inspect whether people can finish it, plus where they wander or get stuck.
• Timed Mission Tests: Measure how quickly people can complete a goal under a clock when efficiency is part of the success criteria.
• Tree Testing: Test whether people can find the right place in your navigation structure using labels alone, without visual design.

As capabilities ship, this section can expand with setup and interpretation guidance for each type.

What Stays In Evaluations

In beta today, Evaluations remains:

• Heuristic audits mapped to Guidelines
• Scores plus Analytics metrics and condition details
• Inventory UI: Evaluations and Benchmarks tables

For running and scoring audits, use Evaluations. Experiments does not replace that workflow.