Respondent finds the people your research is missing.

The assignment was modest: find ten Spanish speakers for a tequila-brand study. In his account of Respondent's beginnings, Jack Pratten says he went recruiting on the New York City subway in 2016. An advertising-agency friend needed participants. Pratten needed people who fit. Before there was a platform, there was the awkward business of asking strangers whether they would talk.

That little job contained the whole problem. Companies want to understand customers, but the customer they need is rarely whoever happens to be available. A payroll administrator knows things a finance director does not. A person who abandoned a product knows things its happiest customer cannot tell you. The interview begins long before anyone asks the first question.

The person before the question

Respondent turns that preliminary work into a marketplace. Researchers describe an audience, set screening questions, review applicants, and invite suitable people. Participants build profiles and apply to paid studies. The platform handles the machinery around the conversation: matching, scheduling, reminders, and incentives. Its current website advertises more than 4.3 million verified participants across more than 150 countries.

The customers include product teams, UX researchers, agencies, and market researchers. Microsoft's Matt Brodsky describes reviewing profiles alongside screener responses before deciding whom to book. Paylocity's published account concerns the more particular challenge of recruiting payroll administrators at smaller businesses. These are useful examples because they name the work. “Business professional” alone would be a hopelessly baggy recruiting brief.

“I'm able to get participants very quickly.”

Matt Brodsky · Microsoft
In Respondent's published customer account

The range extends beyond interviews. Teams can recruit for surveys, usability tests, concept feedback, and diary studies. Someone building a product can test an early prototype; someone selling one can investigate why a buyer walked away. Respondent occupies the recruiting layer of that work, while increasingly offering tools to conduct and summarize the research too.

Millions of people. A vocabulary problem.

In January 2026, Respondent described a surprisingly ordinary failure inside its matching system. A researcher searching for software engineers could miss developers with similar skills. Exact or fuzzy keyword matching did not reliably understand relevance. Industry, function, and seniority filters helped, but the company said some studies filled slowly and the panel could feel smaller than it was.

Its response was AI Matching: use meaning rather than literal words, learn from screening and participation history, and recommend projects similar to those a person previously qualified for. This was a change in how the company searched its existing supply. A bigger database would not, by itself, have fixed the vocabulary mismatch. The announcement promised better matching; it did not publish a controlled before-and-after performance study.

The copyable lesson is almost embarrassingly practical. Before buying more reach, examine how you describe the people you already have. Ask about tasks and responsibilities, not just titles. Keep the screener precise enough to distinguish someone who has done the work from someone who merely recognizes the vocabulary.

The price of getting someone into the room

Respondent's current moderated pay-as-you-go calculator lists recruitment at $40 per completed consumer session and $80 per B2B session. Participant incentives are additional. Annual commitments start at $7,500 and lower recruitment rates. The company says unsuccessful screeners and no-shows are not charged as completed sessions. Buyers should keep the methodology and commitment level attached to any quoted price.

That separation matters. Paying the marketplace gets you recruitment; compensating the participant buys their time. A ten-person B2B study with a hypothetical $100 incentive per person would therefore combine $800 in recruitment fees with $1,000 in incentives. Cheap-looking research becomes less mysterious when someone writes down both numbers.

In March 2026, Respondent addressed another mundane obstacle: some projects need the same person more than once. Multiple Payments allows up to five payouts within one project, with the recruitment fee charged on the first payment. The announcement explicitly traced the change to researchers wanting to reuse participants for several tasks. The workflow followed the study, rather than forcing the study into one payment.

A professional profile has an expiry date

Specific audiences create a second problem: credibility. Respondent says it reverifies professional email addresses every 90 days, checks phone numbers and location signals, looks for contradictory screener answers, and restricts sudden profile changes. High-risk applicants can face government-ID and live-selfie checks. It reports fraud below 1%, a company metric rather than a promise that every conversation will be clean.

Those controls establish useful signals; researchers still need to judge fit. A verified employee may have no involvement in the purchasing decision under investigation. A willing panel participant may differ from a reluctant customer. For a narrow audience, confirm feasibility before committing the whole study. For population-wide conclusions, the sampling design needs attention beyond identity checks.

The panel can travel without the brand

Respondent also sells access through an API. The November 2025 Dscout integration lets researchers screen and recruit its participants inside Dscout. Great Question and Hubble are other partners. That puts Respondent alongside research platforms as a supplier, even where their products overlap. User Interviews, Askable, Prolific, traditional panels, and recruitment agencies offer alternative routes to participants.

The careers page describes a business that expects engineering, design, and commercial teams to run studies themselves. That is a sensible discipline for a company selling access to research participants: staff encounter the same recruiting decisions as customers. It also makes an interesting test for buyers. Try a small, clearly defined project first. Review the quality of the applications, the scheduling experience, and the eventual conversation before treating a large panel count as evidence of suitability.

Its newer offerings include AI-assisted project creation, AI moderation, and recruitment for model evaluation and training. The company now helps automate more of the work around the interview. Yet the subway assignment remains a useful test of the business: can it find the particular people whose experience makes the next question worth asking?