What the work actually involves

You write and evaluate the artifacts a pricing team ships: a willingness-to-pay analysis built from conjoint or Van Westendorp output, a packaging and tiering recommendation deck aimed at a CRO, a discount governance spreadsheet with approval thresholds and margin floors, a revenue waterfall reconciling list to net to realized, a competitive benchmarking summary, a pricing experiment readout with confidence bounds and a go/no-go recommendation. Some tasks are generative — you produce the gold-standard version from a prompt and a scenario. Others are evaluative — the model produced a deck or a workbook and you say precisely where it goes wrong: the elasticity coefficient applied outside the range it was estimated on, the waterfall that double-counts a channel rebate, the tiering deck that recommends a fence customers can trivially arbitrage.

The written rationale matters as much as the artifact. A finding of "the margin math is wrong" is worth little; a finding that names the line, the correct treatment, and why a pricing committee would reject the output is what trains a model.

What the platform screens for

  • Ownership, not adjacency. Screeners push on models you built and decisions that followed from them, including ones that were overruled or that didn't work.
  • Depth under follow-up. Expect to be asked how you estimated elasticity, what identification problem you had, and what you did when the data was observational and price changes were endogenous.
  • Spreadsheet and slide craft. Structure, auditability, assumption tabs, and whether a deck lands a recommendation on page one rather than page eleven.
  • Evaluation judgment. Whether you can distinguish a stylistic preference from an error that would misprice a product line, and rank findings by consequence.

Logistics

Fully remote, fully asynchronous, no standing meetings. You pick up task batches when you have time; 5–20 hours a week is typical and more is usually available. The screen includes an AI voice interview covering background, two or three domain probes, and availability. $90/hour is the rate observed on this listing, not a guarantee — rates vary by task type, batch, and reviewer tier.