The Forecasting Company

What do they actually do

The Forecasting Company builds and hosts time‑series forecasting models and tools. Customers can use Retrocast (a browser-based app) or an API with a Python SDK to upload time‑series data, add contextual variables, run backtests, choose forecast horizons and frequencies, and compare models. The platform supports their in‑house model (t_0) as well as selected open‑source models, and is currently offered via an “apply for access” flow (site, docs).

The company positions its product for operational forecasting across industries (e.g., logistics, retail, manufacturing, energy, pharma) and emphasizes quick setup and model comparisons rather than bespoke, months‑long modeling projects (site). Their materials highlight handling promotions/peaks and enabling planners to incorporate external signals, with both a web interface and API integration routes (About, docs).

Who are their target customer(s)

• Demand planner (retail / e‑commerce): They need reliable forecasts around promotions and peak days; poor accuracy causes stockouts or excess inventory on spikes like Black Friday. Existing tools often miss these spikes, making it hard to automate ordering decisions.
• Supply chain / logistics planner (manufacturing & distribution): Shipment timing and customer demand are noisy, so teams carry large buffers or miss delivery SLAs. They want better ETA and demand forecasts to plan safety stock and transport with less uncertainty.
• Maintenance / operations planner (field service & factories): Uncertain spare‑parts and consumables demand leads to downtime or emergency shipping. They need predictable parts needs so maintenance schedules don’t stall production.
• Finance / FP&A manager: Short‑ and medium‑term revenue and cash forecasts are noisy and require heavy manual adjustment. They want forecasts they can trust to plan budgets, hiring, and working capital.
• Head of data science or analytics (enterprise): Teams are overloaded building and maintaining bespoke forecasting models; it’s expensive and slow to deploy. They want off‑the‑shelf accuracy with less custom engineering and maintenance.

How would they acquire their first 10, 50, and 100 customers

• First 10: Founder‑led, high‑touch pilots via YC and personal networks: run 6–12 week paid pilots focused on upcoming promotions or peak events, handle data integration, and deliver a concise accuracy/inventory‑risk report that replaces current spreadsheets (YC, site, docs).
• First 50: Codify 2–3 vertical templates (retail promos, manufacturing lead times, spare‑parts) and one‑page ROI case studies; run targeted outreach and co‑sell with ERP/WMS vendors and SIs. Standardize connectors and deliverables to keep pilots low‑cost and shorten cycles (site, GitHub, docs).
• First 100: Productize onboarding with self‑serve signup, prebuilt connectors and templates, clear pricing tiers, and a channel program for consultancies/resellers. Use the API and developer docs to enable partner‑led onboarding at scale while reserving a small team for strategic enterprise deals (site, docs, YC).

What is the rough total addressable market

Top-down context:

The closest near‑term market is specialist demand‑planning/forecasting software, estimated around $4.8B in 2024 with other reports placing broader forecasting software near $7.2B, and an “AI demand forecasting” segment under $1B (Grand View, Growth Market Reports, FMI). Expanding to adjacent areas like predictive maintenance and supply‑chain solutions moves the opportunity into the tens of billions, while the overall enterprise software market (~$264B in 2024) is an upper‑bound context (Predictive Maintenance, Supply Chain Solutions, Enterprise Software).

Bottom-up calculation:

As a bottom‑up proxy, if 40k–80k mid‑market and enterprise buyers adopt specialized forecasting at an average $50k–$100k ACV, that implies roughly $2–8B in practical spend, consistent with top‑down ranges. Upside grows via additional use cases in supply chain, maintenance, and FP&A sold through integrations or channel partners.

Assumptions:

• Market definitions overlap (demand planning vs. forecasting vs. supply‑chain analytics), so ranges avoid double‑counting.
• A large share of budgets sits inside ERP/SCM/EAM/FP&A suites; access may require integrations or co‑sell rather than rip‑and‑replace.
• Adoption hinges on trust and handling spikes/promotions; buyers will require proof on those edge cases (YC, site).

Who are some of their notable competitors

• Nixtla (TimeGPT): Offers TimeGPT, a foundation model for time series with API/SDKs for forecasting and anomaly detection—very similar “model‑as‑a‑service” positioning (docs).
• AWS Forecast: A fully managed forecasting service on AWS used by developers and data teams to build and deploy demand forecasts within the AWS ecosystem.
• Blue Yonder: Enterprise supply‑chain planning and demand planning suite widely used in retail and CPG; strong incumbent in the buyer budgets TFC targets.
• o9 Solutions: Integrated business planning and demand/supply planning platform; competes for end‑to‑end planning workflows and budgets.
• Kinaxis: RapidResponse platform for supply‑chain planning (including demand and S&OP); entrenched with large enterprises and complex operations.