Grounded in your own order ledger, not in a plausible-sounding average

The fastest way to lose a shopkeeper’s trust is to tell them something confident and wrong about their own business. So the order Kadaikodi is building in is deliberate: the ledger before the model.

A merchant’s analytics view already computes, from live order data, a revenue overview with real period-over-period growth, order counts and acceptance rates, top-selling and least-selling offerings, category performance, peak hours and peak days, daily order trends, and the rating and review picture — recalculated for whichever window the owner picks. None of it is decorative. If a merchant has taken four orders, the page says four; it does not fill the gap with a plausible-looking curve.

The platform operator gets the same discipline one level up: revenue by category, order-status mix, top merchants, customer growth, and category distribution across the whole marketplace, so supply and demand can be balanced from evidence.

This matters for the AI story more than any model choice does. When demand forecasting and price proposals arrive, they are meant to be explainable against this data — “Tuesday 3pm is your third-highest hour over the last month, and this offering sells out at that hour” — rather than an opaque score. The provider record already carries an analytics snapshot slot in the schema, which is where those derived artifacts are designed to land, alongside the raw history rather than replacing it. And because every read is workspace-scoped and permission-checked at the gateway, a grounded answer can never quietly include a neighbouring merchant’s numbers.

To be exact about status: the analytics themselves are live and computed today. They are arithmetic over your own records, not machine learning — and Kadaikodi says so rather than relabelling them “AI insights”. The model layer that reasons over them is the roadmap, tracked as the commerce decision engine.

Pre-launch scenario — the analytics substrate exists in the app; the figures in any screenshot are illustrative, because the marketplace has not launched.