Survey Issues

This guide addresses issues specific to survey creation, execution, and analysis.

Question Generation

AI Not Generating Questions

Symptoms: Question generation fails or returns empty.

Solutions:

1. Ensure product has sufficient description (100+ words)
2. Add target audience information
3. Try a different survey type
4. Manually create questions as fallback

Questions Are Too Generic

Symptoms: Generated questions don't fit your product.

Solutions:

1. Add more product-specific details
2. Include industry terminology
3. Specify what you want to learn
4. Provide competitor context
5. Edit generated questions to add specifics

Questions Don't Match Intent

Symptoms: Questions ask about wrong aspects.

Solutions:

1. Specify research goals in the prompt
2. Use the appropriate survey type:
   • Product Survey for general feedback
   • Concept Test for new ideas
   • Competitive Analysis for comparisons
3. Edit individual questions

Running Surveys

Survey Won't Start

Symptoms: Start button not working or survey fails immediately.

Solutions:

1. Verify at least one question exists
2. Check that product is selected
3. Ensure the model-iteration setting is within supported limits
4. Try reducing the iteration count
5. Refresh and retry

Survey Runs But No Responses

Symptoms: Survey completes with 0 responses.

Solutions:

1. Check targeting criteria (too narrow?)
2. Verify product description is clear
3. Ensure questions are answerable
4. Try without demographic filters
5. Contact support if persistent

Partial Responses Only

Symptoms: Fewer model responses than the requested iteration count.

Solutions:

1. Some filtering of low-quality responses is normal
2. Very narrow targeting may reduce applicable archetype contexts
3. Complex questions may have lower completion
4. Request additional model iterations if needed for the workflow; iterations are not people

Survey Taking Too Long

Symptoms: Survey not completing in expected time.

Solutions:

1. Large iteration counts take longer
2. Complex questions require more processing
3. Check job status page for progress
4. Keep browser tab open for streaming results

Result Issues

Model-Simulation Ranges Too Wide

Symptoms: Results show a wide model-simulation range.

Solutions:

1. Review whether additional mathematical iterations are appropriate; they do not add human evidence
2. Simplify questions (reduce ambiguity)
3. Add more product context
4. Target more specific demographics
5. See Statistical Methods

Results Seem Unrealistic

Symptoms: Unexpectedly high or low numbers.

Solutions:

1. Review any cited reference comparison and its limitations
2. Review question wording for bias
3. Compare to similar products
4. Consider product description accuracy
5. See Calibration

Missing Segment Analysis

Symptoms: Segment breakdown not available.

Solutions:

1. The workflow may require a minimum model-iteration count for grouping
2. Check if demographic targeting excluded variation
3. Verify clustering was enabled
4. Re-run with more model iterations if appropriate; validate segments with real customer data

Price Analysis Issues

Symptoms: Van Westendorp or price analysis failing.

Solutions:

1. Ensure price questions used correct format
2. Provide price anchors/references
3. Check currency consistency
4. Treat model price output as exploratory and validate with real purchase or suitable human-research evidence

Question Types

Scale Questions

Issue: Responses clustered at extremes.

Solutions:

• Use 5 or 7-point scales (not 3)
• Include neutral option
• Balance positive/negative anchors
• Add "Not Applicable" option

Open-Ended Questions

Issue: Responses too short or vague.

Solutions:

• Ask specific questions
• Request examples or explanations
• Limit number of open questions (3-5 max)
• Place after closed questions

Ranking Questions

Issue: Inconsistent rankings.

Solutions:

• Limit options to 7 or fewer
• Use clear, distinct options
• Consider paired comparisons for many items
• Check for overlapping options

Data Export Issues

Export Missing Data

Symptoms: Exported file incomplete.

Solutions:

1. Check all result tabs are included
2. Verify export format supports all data types
3. Try different export format
4. Export sections separately

Charts Not Rendering in Export

Symptoms: PDF shows blank where charts should be.

Solutions:

1. Wait for page to fully load before export
2. Try PNG export for individual charts
3. Use CSV for raw data
4. Check browser print settings

Best Practices for Avoiding Issues

1. Start small - Use a smaller model run while reviewing questions and setup
2. Preview questions - Review before running
3. Complete products - Higher completeness = better results
4. Specific targeting - Keep the supplied audience context relevant and plausible
5. Clear questions - Simple language, one concept per question

Related Articles

• Surveys - Survey feature overview
• Results Interpretation - Understanding outputs
• Model Diagnostics and Statistical Methods