What the work involves

Each environment reproduces a marketing org's tool surface and populates it with documents, dashboards, personas and deliberately planted problems. Your job is to design the analytics tasks that live inside it: a brief drawn from work you have genuinely owned, a clear statement of what the task is really testing, and the specific artifacts — a GA4 property with a broken UTM convention, a Salesforce opportunity table that disagrees with the ad platform's conversion count, a Looker dashboard with a filter someone left on — that make the problem findable but not obvious.

The scope is anomaly detection, channel-performance synthesis, cross-source reconciliation, attribution and incrementality reading, and reporting that supports an actual decision. Half the work is writing the reference answer and the grading criteria; the other half is reading AI agent attempts and judging them against your own standard. The hardest part is the plausible-but-wrong response — the agent that produces a clean, confident channel report while silently double-counting a conversion window. Your criteria have to catch that, in writing, in a way another reviewer could apply.

What the platform screens for

  • Hands-on ownership. Mercor is explicit that managing analysts is not the same as having done the work. Expect follow-ups that go to the level of specific metrics, date grains, and how you actually resolved a discrepancy.
  • Tool fluency you can demonstrate: GA4, a product analytics tool such as Amplitude, a CRM such as Salesforce, and ad platform reporting.
  • Written reasoning. Much of the deliverable is prose explaining why a decision is right, not the decision itself.
  • Prior work building AI training environments, RL environments or simulated case studies is strongly preferred, as is Rubric Academy or Rubric Bootcamp certification. Neither is stated as mandatory.

Applicants complete a short multiple-choice knowledge screener specific to marketing analytics before human or AI review.

Logistics

Remote and largely asynchronous, structured as contract contribution rather than employment. Observed pay for this listing is $60–100/hr, which varies with the depth of the sub-domain and the review tier you are placed in; it is not a guarantee. Volume is typically flexible and negotiated in blocks — practitioners who can commit a predictable 8–15 hours a week tend to get sustained assignment flow rather than one-off tasks.