Kirana retailers run their reorders from memory, so they miss the week a SKU is overdue, the festival that's ten days out, or the new brand that's about to be asked for by name. This is a v1 concept for Kirana Club, the B2B kirana marketplace: three independent, hand-written rules that turn a shop's own order history into at most two cards a week, each with one plain-English reason to look. No AI-built cart, no brand swapping. The retailer still decides.
A large FMCG brand sends a field rep to do this kind of nudging in person. An independent kirana has nobody doing it for them, and a small platform team has no analyst or designer sizing it either. So v1 stays small and debuggable on purpose: three rules, each explainable in a sentence, capped at two cards a shop sees in a week so it never turns into another banner nobody reads. A card just gives a concrete reason to look, and the retailer taps through to the product page and decides for themselves. Anything that behaves like a black box, multi-item AI carts, brand swaps for margin, waits for a later version, once this one has earned some trust.
Fires when a shop is > 1.2× past its own historical median reorder gap for a SKU it already buys. Nothing about other shops, just this shop's own rhythm.
Bourbon Biscuit 5/JAR usually reordered every 14 days. It's now day 18.
Fires 10 to 14 days before a region-specific festival, only for a category the shop has already bought from. No cold-category push.
Diwali in 12 days, the shop has bought Caramel Eclairs Toffee before, and toffee demand rises about 18% that week.
Fires when a brand's order count is growing fast platform-wide and its margin beats this shop's own average basket margin, even if the shop has never ordered it.
Ankit Koffy is trending up, margin ₹88 versus this shop's own ₹40 average.
When more than one fires in the same week, priority is fixed instead of learned, so it stays explainable: Reorder Due first, then Festival Stock-Up if inside its window, then Rising Brand only if a slot is still open. Cap stays at two cards regardless of how many rules fired.
The actual feed screen, styled to match Kirana Club's own app: their violet header, terracotta buttons, the Buy / MRP / Margin table on every product, bilingual Hindi and English copy. It's clickable, try the new “मौके” tab, tap a card to see it move toward a product page, or open the “why this fired” tabs for the trigger, hypothesis and success metric behind each nudge type.
Open the demo full-screen ↗Every nudge is logged against a no-nudge control group of matching shops: tap-through and order-within-48h for Reorder Due, category order value against last year for Festival Stock-Up, and trial-then-repeat rate for Rising Brand Discovery. If a rule shows no lift after a few weeks, that's a real result and it gets cut, a fast honest no beats quietly running a card nobody acts on. Only real tap-through and a genuine order lift on Reorder Due and Festival, across a few weeks, would justify a v2: AI-assembled carts, in-card brand swaps for margin, and a brand-reliability guardrail.
Built as a design concept, not a shipped engine. The product names, rates and margins in the demo are taken from publicly visible Kirana Club app screenshots to make the concept concrete; the “KC” mark shown is a neutral placeholder, not their wordmark.