Compare

Compare provides side-by-side analysis of model-generated reactions to products, concepts, or versions. Use it to rank hypotheses and prepare a real-customer or live-market test. It is not a replacement for a human A/B study.

Comparison Types

Product vs Product

• model preference distribution
• attribute-by-attribute reactions
• archetype-level differences
• stated switching objections

Before vs After

• changes between two supplied versions
• shifts in model-generated scores or themes
• differences across defined archetypes

These differences describe the configured model runs, not measured changes in a market over time.

Concept Comparison

• model-generated ranking
• clarity and value-proposition reactions
• implementation questions
• risks and objections to validate

Running a Comparison

1. Go to Compare in the app.
2. Select the comparison type.
3. Choose the items to compare.
4. Configure the audience context and criteria.
5. Run the model comparison.

Configuration Options

| Option | Description | | --- | --- | | Archetypes or iterations | Model perspectives or mathematical repetitions, not respondents | | Matched context | Use the same defined contexts on both sides | | Attributes | Dimensions supplied for comparison | | Segments | Defined archetypes or model-output groupings to review |

Understanding Results

Preference Matrix

The matrix can show the model-generated split, archetype breakdowns, and model-simulation ranges. It does not establish population-level statistical inference about people.

Key Differences

AI-generated summaries may highlight score differences, reasons, and candidate recommendations. Review them against the underlying responses and supplied assumptions.

Use Cases

Product Iteration

1. Capture the current product or message.
2. Create a proposed revision.
3. Compare both under the same model context.
4. Select a candidate for customer or live-market validation.

Competitive Exploration

1. Supply accurate product and competitor context.
2. Run a model comparison.
3. Identify strengths, gaps, and questions.
4. Validate with customer, sales, or market evidence.

Best Practices

1. Keep the audience and evaluation context matched.
2. Compare one major change at a time where practical.
3. Treat rankings as directional model output.
4. Inspect archetype-level reasoning and sensitivity.
5. Validate the selected candidate with real evidence.

Related Articles

• Discovery - Initial audience exploration
• Surveys - Survey-style model responses
• Model Diagnostics and Statistical Methods