Not all responses are created equal. Deepfield helps you assess response quality so you can focus on high-quality data for your analysis.
Why Quality Matters
High-quality responses:
Provide accurate, thoughtful answers
Contribute meaningful insights
Make analysis more reliable
Low-quality responses:
May contain random or rushed answers
Can skew your results
Add noise to your data
Quality Scoring
Deepfield automatically assesses response quality based on several factors.
What's Evaluated
Factor | What It Measures |
Completion time | Was the study completed in a reasonable time? |
Response length | Are open-ended answers sufficiently detailed? |
Consistency | Do answers make logical sense together? |
Engagement | Do responses show thoughtful engagement? |
Audio/Video quality | Is media clear and audible? |
Quality Indicators
Responses may be flagged for:
Speeding: Completed unusually fast
Straightlining: Same answer for all matrix questions
Gibberish: Nonsensical open-ended responses
Poor media: Inaudible or unclear recordings
Viewing Quality Scores
In the Response Table
The response table shows a quality indicator for each response:
High quality: Meets all quality standards
Medium quality: Some concerns but usable
Low quality: Significant quality issues
Individual Response Details
Click on any response to see:
Overall quality score
Specific quality flags
Details about any issues detected
Quality Factors Explained
Completion Time
Too fast: Participant may have rushed through without reading
Typical flag: Completed in less than 1/3 of average time
Too slow: Participant may have been distracted
Less concerning than speeding
May indicate thoughtful responses
Open-Ended Response Length
Too short: Brief answers that don't provide insight
Example: "good" or "idk"
Misses the value of qualitative questions
Ideal: Responses that answer the question with some detail
Complete sentences
Specific examples or explanations
Straightlining
When participants select the same option for every row in a matrix:
May indicate disengagement
Could be valid if all items truly rate the same
Review in context
Consistency
AI checks if answers make logical sense:
Does Q5 answer align with Q3?
Are there contradictions?
Do responses tell a coherent story?
Media Quality
For video and audio responses:
Is audio clear and understandable?
Is video properly recorded?
Can the response be transcribed?
Managing Quality Issues
Review Low-Quality Responses
Before including in analysis:
Filter to show low-quality responses
Review each one individually
Decide whether to include or exclude
Exclude Problematic Responses
Options for handling poor quality:
Exclude from analysis: Don't include in reports
Flag for manual review: Review before deciding
Include with caution: Use but note the limitation
Replace Low-Quality Responses
If quality issues are significant:
Consider recruiting additional participants
Replace unusable responses with new ones
Update your quality criteria for future studies
Improving Response Quality
At the Study Design Stage
💡 Tip: Keep it reasonable length. Long studies lead to fatigue and lower quality responses.
💡 Tip: Write clear questions. Confusing questions get confusing answers.
💡 Tip: Mix question types. Variety keeps participants engaged.
At the Recruitment Stage
💡 Tip: Target the right audience. Engaged, relevant participants give better responses.
💡 Tip: Set expectations. Let participants know what's involved.
At the Collection Stage
💡 Tip: Monitor early. Check the first responses for quality issues.
💡 Tip: Act quickly. Address problems before collecting many bad responses.
Quality in Analysis
Filtering for Analysis
When generating reports, you can:
Include only high-quality responses
Set quality thresholds
Exclude flagged responses
Reporting on Quality
Your analysis may note:
Total responses collected
Responses meeting quality standards
Any exclusions and reasons
Common Quality Questions
What's a typical quality pass rate?
Generally 80-95% of responses meet quality standards. Below 70% may indicate study issues.
Should I exclude all low-quality responses?
Review them first. Some may still contain valuable insights. Others should definitely be excluded.
Why is quality low across the board?
Possible causes: Study too long, questions confusing, wrong audience, poor incentive alignment.
Can I improve quality after launching?
Limited options once launched. You can recruit more participants or adjust criteria for new responses.
Quality Checklist
Before analysis, verify:
Reviewed overall quality distribution
Checked low-quality responses individually
Decided on inclusion/exclusion criteria
Documented any quality-related decisions
Have sufficient high-quality responses for analysis