Generalist Search Expert

$40–70/hrrate as listed

Mercor is recruiting researchers from any discipline to build and grade hard search tasks — writing questions whose answers exist online but resist a single query, then documenting the source trail that proves them. Observed pay for this work runs $40–70/hr, project-dependent and set at offer.

Summary, eligibility guidance and screening preparation are prepared by Sidequest from the platform's posting. Check the source listing before applying.

What the work involves

This is search-and-verify work, not writing work. On a typical shift you might author a question that requires stitching together three or four independent sources — a filing, a dataset, an archived page, a non-English press release — and then produce the full retrieval path: queries used, pages visited, the exact passage that settles the answer. Other tasks run in the opposite direction: a model has already produced a research answer with citations, and your job is to check every link, flag the citation that supports a weaker claim than the one it's attached to, and decide whether the reasoning chain actually closes or just looks like it does.

The common thread is provable ground truth. Tasks are rejected when the answer is contestable, when it hinges on a paywalled source nobody can re-check, or when a competent searcher can find it in one query. Expect to spend real time on difficulty calibration — the interesting band is questions that take a skilled human twenty minutes and defeat a model outright.

What the platform screens for

A verifiable research background. Mercor's screen is credential-anchored: a graduate degree, a publication record, a journalism or legal-research or analyst track record. Domain doesn't matter; depth does.
Demonstrated search fluency. Interviewers probe how you actually find things — operator use, archive and database habits, how you handle a dead end or a source in a language you don't read.
Source discipline. Can you distinguish primary from secondary, tell when a citation is decorative, and say plainly when the evidence doesn't support a conclusion?
Written precision under follow-up. The AI interviewer pushes on vague claims; specifics survive, generalities don't.

Logistics

Fully remote and asynchronous — tasks are claimed from a queue with per-task or weekly deadlines rather than fixed hours. Most contributors commit 10–20 hours a week, and availability questions in the screen are taken literally, so answer with the hours you will actually deliver. Onboarding usually includes a calibration batch reviewed against a rubric before volume opens up; pay is hourly against logged, reviewed work.

What they look for

Graduate degree or equivalent professional research record in any field
Advanced web research: search operators, archives, primary-source databases, public filings
Source evaluation — distinguishing primary from secondary and citation from support
Clear, concise technical writing in English
Comfort documenting a reproducible research trail step by step
Reliable async self-management against task deadlines

Before you apply

Would you pass this screen?

Drawn from what this listing actually requires. Answered in your browser — nothing is sent to us, and we do not store your answers.

Do you hold a graduate degree, have a publication record, or have worked for at least 2 years in a role whose primary responsibility was research, journalism, legal analysis, or consulting?
Research credential or track record · required
Can you commit to working at least 10 hours per week on this role, on a fully remote and asynchronous basis with self-managed scheduling?
Remote work availability · required
Is English your working language for reading and writing complex sources and documentation?
Fluent English reading and writing · required

0 of 3 answered

Included with this listing

Interview prep — 11 questions

Matched to what this platform's screening actually measures. Answer each out loud in 60–90 seconds: claim → concrete example → implication.

What it tests: Whether you naturally think in terms of source layers and can narrate the actual search process—not just the conclusion. Do you remember what you tried first, what failed, what you pivoted to?
A strong answer: Name a specific project, identify 2–3 distinct sources you consulted (a database, an article, an archive), and explain why each was necessary because no single source gave you the full picture. Describe at least one dead end and how you recovered.
Where people lose it: Listing sources without explaining why you needed each one, or describing the answer without describing how you found it. Vague ('I researched online') signals you don't track your own process.