What is synthetic research?

Synthetic research uses AI-powered personas that simulate how real buyers, customers, or market segments would respond to your research questions. It produces directional signal — insights to sharpen your hypotheses — before you invest in real primary fieldwork. Gather Synthetic runs AI-moderated interviews across any audience archetype in minutes, not weeks.

How is synthetic research different from real research?

Synthetic research generates directional signal using AI personas, while real primary research collects data from actual people. Synthetic pre-research is faster (minutes vs. weeks), cheaper, and ideal for hypothesis generation and stakeholder alignment. It is not statistically valid and should be used to focus and prioritize real fieldwork, not replace it. Gather offers both synthetic pre-research and real AI-moderated interviews with actual respondents.

What types of research can Gather Synthetic run?

Gather Synthetic supports 14 study types: Brand Perception, Product Feedback, Concept Testing, Competitive Intelligence, CX/NPS Deep Dive, Messaging & Claims Testing, Win/Loss Analysis, Market Segmentation, Pricing Research, Ad & Creative Testing, Customer Journey Mapping, Feature Prioritization, Churn & Retention Analysis, and Audience Discovery.

How much does synthetic research cost?

The first 5 synthetic interviews are free. After that, B2C audience interviews cost $0.50 each and B2B audience interviews cost $5.00 each. There is no subscription required.

How accurate is synthetic research?

Synthetic research is directional, not statistically accurate. Confidence scores reflect internal AI response consistency, not statistical significance. Quantitative projections carry a ±15–20% margin of error. Treat findings as educated hypotheses to validate with real primary research, not as definitive measurements.

← All brands

AI-assisted customer service resolution · PUBLIC RESEARCH BRIEF

ZendeskWhich support case should AI resolve, assist, or hand back to a person?

Zendesk's AI page describes unified service knowledge, AI agents for multi-step workflows, Copilot recommendations and actions, and AI-based quality monitoring. This brief studies the boundary among automation, agent assistance and human ownership; it does not assume that automation rate, handle time or a generated answer reflects a resolved customer problem.

Sources checked 2026-09-24 · Simulation results not yet generated

Test a question in Gather ↗Explore fictional demo →

CHANGE ONE THING. LEARN WHAT MATTERS.

Three questions for the GTM team.

For a common Zendesk support request, would an AI agent attempt resolution immediately or first show the customer the knowledge sources and intended action produce a safer, more complete outcome? Score against verified policy and the customer's actual need, not containment alone.

Set up this study →

When Zendesk Copilot recommends a next action to a human agent, would showing the evidence and confidence before the recommendation or presenting the recommendation first lead to better acceptance, correction and escalation decisions? Include ambiguous cases where no action is appropriate.

Set up this study →

For a quality-review queue, would prioritizing interactions by predicted risk or by explicit policy triggers surface more consequential failures without overwhelming reviewers? Use a blinded, independently adjudicated sample and compare missed issues as well as false alarms.

Set up this study →

PROPOSED AUDIENCE

Who should weigh in?

North American customer-service organizations evaluating or operating Zendesk across messaging, email or other support channels. Include frontline agents, team leads, knowledge owners, quality reviewers and administrators handling both routine and ambiguous cases. Use adult customer scenarios and authorized employees. Proposed audience; no resolution or cost outcome is implied.

TWO TIME HORIZONS

Trial today. A habit tomorrow?

Near term · 0–90 days

Over 0–90 days, run controlled cases with synthetic or approved de-identified records, current knowledge and a fixed escalation policy. Measure factual accuracy, policy compliance, correct resolution, unnecessary automation, safe handoff, agent corrections and customer effort. Do not allow the study agent to issue refunds, alter accounts or contact customers.

Longer term · 3–12 months

Over 3–12 months, follow approved workflow variants across comparable case types. Examine reopened contacts, downstream corrections, knowledge gaps, escalation quality, agent workload, customer outcomes and maintenance required as policies change. Cost or satisfaction claims require observed operations and should account for work shifted to customers, reviewers or other teams.

What would make the result actionable?

Use versioned knowledge, current business rules, de-identified or synthetic cases and independent resolution adjudication. Protect customer and employee data, log model and workflow versions and assign a human owner to every consequential action. Simulated containment or agent preference is not evidence of real resolution, savings or satisfaction.

A Gather simulation returns hypothetical customer reactions. Quantifying revenue, traffic or retention needs actual business inputs and validation against observed behavior.

Public sources

Zendesk AI platform for knowledge, AI agents, Copilot and quality assurance ↗Current product page; checked 2026-09-24