Working prototype · public data only · Mattel supply-chain decision intelligence
The supply decisions you commit this quarter set next quarter's service, penalties, and margin. Each tool reads public filings or published rules, scores one operator number 1 to 99, and names the lever, with the lead time to still act on it.
Two families. The first three read the signals that move margin before the P&L sees them. The last two sit on the daily scorecard: the retailer penalty your fill rate controls, and the peak order you commit months early against a forecast you cannot trust yet. All run on public data or public rules, never a number the model claims to know.
Margin signals · before the print
Inventory running ahead of sales
Reads how far inventory has pulled ahead of Mattel's own sales history and scores the margin-event risk, calibrated to Mattel's baseline: the signal that fired before the 2023 gross-margin trough.
Today · CLEAR Open the radar → 02 · SOURCING → TARIFFChina-origin tariff exposure
Sizes the tariff bill on Mattel's China-sourced import mix against the disclosed 50%→<15% rebalance, and prices what the next move off China is worth in basis points of operating income.
Today · WATCH · ~$106M Open the model → 03 · FREIGHT → MARGINOcean-freight margin exposure
Moves a public container-freight index against Mattel's Q3 holiday build and computes the landed-cost-to-gross-margin bridge. The freight bill is committed months before the margin prints.
2021-scale shock · ACT Open the bridge →Operator decisions · on the daily scorecard
Not a margin retrospective. These two sit on the metrics a supply-chain leader is measured on directly: the retailer fill-rate penalty, and the peak-season order committed against a long lead time. Both lead with the recommended move on load.
Retailer on-time, in-full deductions
Reads a book of inbound PO lines against Walmart's published 90 / 98 / 95% goals and the 3% penalty, scores the deduction exposure, and names whether on-time or in-full is the lever to pull first. The fine your fill rate controls, line by line.
Peak week · AT RISK Open the console → 05 · FORECAST → COMMITOrder quantity vs a long lead time
Runs one seasonal SKU through a Q4 demand band and the Asia lead time, lands the math's recommended order on load, and shows the cost of being wrong: stock out and lose margin, or over-commit and mark down. The planner's call, not the CFO's.
Hero SKU · BALANCED Open the simulator →How to read them
Each tool is calibrated to Mattel, not a peer average. A single off-the-shelf trip-wire would already be in the literature; the work is tuning each signal to Mattel's own operating history, so the score means something for this business.
Each leads with the verdict. Open a tool and the recommended read is on screen. The scenarios are one click, the controls optional. Built on the operator-legible pattern, not a dashboard to configure.
Public data, every number traced. SEC 10-K and earnings disclosures, the HTS/Section-301 tariff schedule, and public freight indices (Drewry, Freightos). Where a private input is needed, it is shown as a clearly labelled, calibrated estimate, never a number the model claims to know.
The lead time is the asset. Inventory, sourcing, and freight all commit before the margin event. Reading them early turns a markdown or a tariff hit into a decision you can still make.
Prototype suite prepared for Mattel · data: SEC EDGAR (Mattel 10-K, earnings 8-K/calls), USITC HTS + Section-301 schedule, Drewry World Container Index / Freightos Baltic Index. Synthetic inputs are visibly labelled and calibrated to cited anchors. Jeff Pinto, data science / ML / operations analytics · github.com/bigbrownjeff.