What the work involves

You receive a synthetic deal document pack for a consumer, leisure or multi-site services transaction — think fitness clubs, owned-versus-franchise estates, multi-site roll-ups — and build a working financial model from it. No template, no shell, no pre-wired tabs. The deliverable is an editable Excel workbook with live formulas throughout, explicit source references back to the pack, and an assumptions section where your judgement calls are stated rather than buried. Expect roughly 12–15 hours on the build.

The second half is peer review. You get two anonymized models from other contractors who worked the same transaction, and you assess them against a fixed criteria set, writing a short rationale for each judgement and completing a review scorecard. This is where the project earns its pay band: the platform is capturing how an experienced associate distinguishes a defensible modelling choice from a sloppy one, and why. Reviews that only flag broken links and formatting are worth little; reviews that catch an unsupported like-for-like growth assumption or a mis-specified working capital bridge are the signal.

What the platform screens for

  • Verifiable deal-side background. Roughly 2 years in investment banking (Consumer/Retail or generalist M&A, ideally Paris, London or elsewhere in Europe), then 1–2 years as an associate at a mid-market or large-cap PE fund. French deal experience or a Paris office is a plus, not a gate.
  • Sector specificity. Screeners follow up on unit economics, cohort behaviour in subscription-based leisure, owned versus franchised estate economics, and how a roll-up's acquisition pipeline shows up in the model.
  • From-scratch modelling. You will be asked how you structure a workbook when nothing is given to you — tab architecture, circularity handling, how you carry sources.
  • Review judgement. How you rank errors by materiality, and how you write a criticism that a peer would accept.

Logistics

Fully remote and asynchronous. Kickoff is the week of September 22 with all deliverables due October 5, 2026 — a tight window, so the availability questions are real and the platform checks them. Total effort is around 20–25 hours split across the build and the two reviews. A supplemental project NDA is required beyond any standard platform agreement. Generative AI may not be used to build or review the models; expect the screen to ask about this directly. Working language is English. Observed pay for this listing is $130–140/hr, as reported by the platform and not guaranteed for any individual engagement.