Data Scientist — Incrementality & Media Mix Modeling

Vysta PMG manages $200M\+ in annual Google and YouTube ad spend across 150\+ ecommerce brands. We're hiring a Data Scientist to own the measurement stack that proves what our spend is actually driving. Most ecommerce brands don't trust the ROAS their ad platforms report — and they're right not to. Last\-click attribution overstates what paid is contributing, platform models double\-count across channels, and the number on the dashboard rarely matches the number on the P\&L. We're building the proof layer underneath all of it: geo\-based incrementality testing, conversion lift studies, and Media Mix Models that reconcile Google with Meta, TikTok, and the rest of the funnel. What You'll Own \- Design and run geo holdout incrementality tests across top\-spending client accounts \- Stand up our MMM pipeline — white\-labeled on Measured initially, then proprietary on Google Meridian for S\-tier accounts spending $500K\+/month \- Translate model output into client\-facing proof of incremental impact, working alongside senior media buyers and account leads \- Partner with our engineering team on the data infrastructure that feeds the models (BigQuery, dbt) \- Build the test design playbook that scales across 150\+ accounts without requiring you in every conversation What We're Looking For \- 4\+ years running incrementality and/or MMM in production at an agency, brand, or measurement platform \- Strong Bayesian fundamentals — hands\-on experience with Meridian, Robyn, PyMC, or a homegrown framework \- Strong Python, SQL, and BigQuery \- Working knowledge of the Google Ads and Meta ecosystems \- Ability to explain causal inference to a media buyer or client in plain English Why This Role Is Different Cross\-account data from 150\+ ecommerce brands is something no in\-house team and no point\-solution vendor can replicate. The frameworks you build get applied to $200M\+ in active spend, and the work ships directly into client conversations — this is not a research role parked in a notebook. -