Synthetic People
Synthetic People
Test Studio
Our Test Studio reruns real-world surveys on the Synthetic People platform and measures how closely outcomes align with human results.
The benchmarks are transparent. The methodology is visible. We measure ourselves publicly.
3
Total studies executed across scenarios.
80.2%
Average similarity between human and synthetic results.
94.9%
Alignment of directional trends across outputs.
3
Breadth of scenarios tested in the suite.
3
Industry mix represented across all studies.
One built on stated responses. One built on behaviour. Both answering the same questions.
Estimated CostEstimated using $5-$8 per response across the sample size.
$1,000-$1,600
Estimated TimeTypical fieldwork duration based on sample size.
1-2 weeks
Estimated EffortGeneral time required for cleaning, analysis, and reporting.
80-120 hours
Population Calibrated
200
Behavior Signals
18,932
relevant people's actions
Contextual Threads
62
conversations inferred
Knowledge Bank
31
sources analysed
Estimated CostAligned to human survey coverage, same depth, no fieldwork.
$2499
Estimated TimeWeeks of survey cycles, compressed into hours, same baseline.
3-4 hrs
Estimated EffortSame rigor as human analysts, without the operational drag.
1-2 hrs
Avg. SimilarityHow closely synthetic responses match human distributions across all questions, calibrated to the same survey structure, audience, and context.
84.6%
Directional AlignmentHow often both systems point to the same conclusion, even when the exact numbers differ.
97.0%
Prediction AccuracyHow reliably synthetic outputs anticipate the dominant human choice across questions within this study.
97.1%
Relationship StrengthHow consistently patterns between options hold across both datasets, not just individual answers, but how they move together.
96.2%
A signal-level view of where synthetic persona mirrors human response patterns.
Among the major behavioral patterns evaluated, Sales Adoption (90.4%), Hiring Impact (89%), HR Adoption (87%), and Customer Support Adoption (86.1%) showed the strongest convergence with human survey outcomes.
The largest variation appears in emotionally framed excitement toward AI adoption and future investment expectation questions, which fall within the preferences shift and lower consistent patterns ranges at 73.5% and 80.0% similarity respectively.
Variation is most visible in future-oriented sentiment, investment intent, and emotional framing questions, where responses are naturally influenced by internal positioning, executive signaling, recent market exposure, and rationalization in the moment.

Go beyond the summary. Inspect every question, every distribution, and every gap - side by side.
One built on stated responses. One built on behaviour. Both answering the same questions.
Estimated CostEstimated using $5-$8 per response across the sample size.
$41,250-$66,000
Estimated TimeTypical fieldwork duration based on sample size.
4-5 weeks
Estimated EffortGeneral time required for cleaning, analysis, and reporting.
80-120 hours
Population Calibrated
8,250
Behavior Signals
57,271
relevant people's actions
Contextual Threads
119
conversations inferred
Knowledge Bank
45
sources analysed
Estimated CostAligned to human survey coverage, same depth, no fieldwork.
$2499
Estimated TimeWeeks of survey cycles, compressed into hours, same baseline.
3-4 hrs
Estimated EffortSame rigor as human analysts, without the operational drag.
1-2 hrs
Avg. SimilarityHow closely synthetic responses match human distributions across all questions, calibrated to the same survey structure, audience, and context.
79.4%
Directional AlignmentHow often both systems point to the same conclusion, even when the exact numbers differ.
93.1%
Prediction AccuracyHow reliably synthetic outputs anticipate the dominant human choice across questions within this study.
96.8%
Relationship StrengthHow consistently patterns between options hold across both datasets, not just individual answers, but how they move together.
76.1%
A signal-level view of where synthetic persona mirrors human response patterns.
Among the five signal categories evaluated, Emotion (88%), Reason / Explanation (82%), and Behavior (79%) showed the strongest convergence with human survey outcomes.
The largest variation appears in purchase-stage preference and future price expectation questions, which fall within the Preferences Shift range at 61.0% and 67.7% similarity respectively.
Variation is most visible in timing, future outlook, and purchase-stage intent questions, where responses are naturally influenced by framing, recent market exposure, and rationalization in the moment.

Go beyond the summary. Inspect every question, every distribution, and every gap - side by side.
One built on stated responses. One built on behaviour. Both answering the same questions.
Estimated CostEstimated using $5-$8 per response across the sample size.
$5,155-$8,248
Estimated TimeTypical fieldwork duration based on sample size.
1-2 weeks
Estimated EffortGeneral time required for cleaning, analysis, and reporting.
80-120 hours
Population Calibrated
1,031
Behavior Signals
421,872
relevant people's actions
Contextual Threads
74
conversations inferred
Knowledge Bank
37
sources analysed
Estimated CostAligned to human survey coverage, same depth, no fieldwork.
$2499
Estimated TimeWeeks of survey cycles, compressed into hours, same baseline.
3-4 hrs
Estimated EffortSame rigor as human analysts, without the operational drag.
1-2 hrs
Avg. SimilarityHow closely synthetic responses match human distributions across all questions, calibrated to the same survey structure, audience, and context.
76.7%
Directional AlignmentHow often both systems point to the same conclusion, even when the exact numbers differ.
94.6%
Prediction AccuracyHow reliably synthetic outputs anticipate the dominant human choice across questions within this study.
97.2%
Relationship StrengthHow consistently patterns between options hold across both datasets, not just individual answers, but how they move together.
88.8%
A signal-level view of where synthetic persona mirrors human response patterns.
Among the six signal categories evaluated, Emotion (91.4%), Attribute (89.2%), Reasoning (86.4%), and Behavior (84.0%) showed the strongest convergence with human survey outcomes.
The largest variation appears in future food consumption intent and climate-related behavioral reporting, which fall within the emerging difference range at 54.3% and 63.8% similarity respectively.
Variation is most visible in future-oriented consumption questions and sustainability-related self-reporting, where responses are naturally influenced by aspiration, framing, perceived social responsibility, and rationalization in the moment.

Go beyond the summary. Inspect every question, every distribution, and every gap - side by side.