Demand forecast & reorder setup
Online retailers who reorder stock from last month’s sales and a spreadsheet.
- Clean your order history and segment the catalogue
- Weekly forecasts per product, backtested on your own past weeks
- A reorder policy with the service-level vs stock trade-off priced in your numbers
You get: Weekly reorder quantities you can import, a short written report, and the code.
Typically 2–3 weeks
The same work on public data →
Transaction anomaly alerts
Payment-heavy businesses, marketplaces, and agencies monitoring client accounts.
- Peer groups built on what each merchant or customer is, not on the behavior being tested
- Behavioral scoring within each peer group
- An alert queue sized to what your team can review each day
You get: A scoring pipeline, a daily alert list with the reason behind each alert, and an evaluation at your alert budget.
Typically 3–4 weeks
Data audit sprint
Teams that want to know whether their data can support a model before paying for one.
- A review of your data sources, their quality and their history
- Two or three concrete use cases, ranked by value and effort
- A clear go / no-go for each one, with the reason
You get: A written audit you can act on yourself, or hand to whoever builds the model.
Typically 1 week