Plan inventory on the data, not the gut.
Forecasting projects demand and return rates per SKU per marketplace, grounded in your settled history — not ad-account vanity numbers. Outputs land in ai_predictions with the inputs and the model version preserved.
- Per-SKU per-marketplace demand forecast, grounded in settled history (post-return, post-deduction revenue).
- Return-rate forecast per SKU per marketplace, so reorder math accounts for actual sell-through, not gross sales.
- Outputs persisted as ai_predictions — auditable, replayable, comparable across model versions.
What you get.
Settled, not gross
Forecasts use post-return, post-deduction revenue from your ReconciliationBatch — not the noisy gross numbers from the marketplace ad console.
Marketplace-aware
An SKU sells differently on Ajio than Meesho. Forecasts are per SKU per marketplace and respect each channel's seasonality.
Return-rate aware
Sell-through ≠ revenue. The forecast factors in your historical return rate per SKU per marketplace, so reorder quantities aren't built on RTO sand.
Forecasting horizons.
- Daily demand for the next 14 days, per SKU per marketplace.
- Return rate trend per SKU — early-warning when it's drifting up.
- Restock recommendations grounded in lead time + safety stock.
- Promotional uplift estimates — what changes if you raise spend by ₹X.
- Cross-marketplace cannibalisation — Meesho promo's effect on Ajio.
Practical answers.
Forecasts work with whatever history Robnu has. Below 30 days of data, the forecast surfaces a wider confidence band and is honest about what it doesn't know.
Yes. ai_predictions is queryable through the same scoped query helpers as the rest of your data. Export to CSV from the dashboard or query directly through your read-only role.
Try it inside your own dashboard.
Free during early access. No card. Forever free under 25 orders/day.