Design and stress-test simulated field marketing environments — event follow-up, sponsorship calls, regional pipeline recovery, named-account planning — that are used to evaluate AI agents. You write the briefs and planted problems, specify the CRM and tracker states they live in, and grade agent attempts against your own reference standard.

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

The work

Mercor is building high-fidelity replicas of a marketing org's tool surface: email, storage, CRM, project management, social scheduling, web and product analytics, AEO/SEO, ads, CMS and support. Your job is to populate the Field Marketing slice of that world with tasks that a real regional marketer would recognize — a post-event lead list where the scan data conflicts with the CRM record, a sponsorship renewal that looks good on registrations and bad on sourced pipeline, a region that has to recover a number with two quarters left and a fixed budget.

Day to day you will: write a task brief drawn from work you actually did, state explicitly what it is testing (the planted issue and the decision a strong practitioner reaches), specify the documents, dashboards, personas and tool states the task needs to feel real, author or review the reference answer and the criteria separating a strong response from a plausible-but-wrong one, and then read AI agent transcripts and score them against that standard.

What the screen looks for

The distinction that matters most here is ownership versus execution. The listing is blunt about it: you should have run a region against a sales number, not only executed event logistics. Screeners probe for pipeline attribution arguments you have had with sales, sponsorship decisions you killed or renewed and why, and how you handled a quarter where the region was behind. They also test evaluation instinct — can you say what makes a wrong answer plausible, not just wrong? Fluency in Salesforce (or equivalent) and a tracker like Asana or Linear is assumed and gets checked with specifics. Prior work on AI training or RL environments, simulated case studies, or Rubric Academy/Bootcamp certification is a strong plus, not a gate.

Logistics

Fully remote and largely async; you pick up task authoring and review work in blocks rather than to a fixed schedule.
Observed pay band is $60–100/hr, typically set by depth of hands-on ownership and prior environment-design experience. Bands are as observed on the platform, not guaranteed.
Applicants complete a short multiple-choice knowledge screener specific to field marketing before human or AI review.
Volume varies with environment build cycles; treat this as a substantial side engagement rather than a fixed weekly commitment.

Marketing Expert — Field Marketing (AI Agent Environments)

The work

What the screen looks for

Logistics

Interview prep — 10 questions