SQL vs NoSQL for building a custom multi-tenant ERP for retail chain (new build inspired by Zoho, current on MS SQL Server, debating pivot)

Hey folks, We're planning a ground-up custom multi-tenant ERP build (Flutter frontend, inspired by Zoho's UX and modular patterns) to replace our current setup for a retail chain in India. Existing op

Hey folks, We're planning a ground-up custom multi-tenant ERP build (Flutter frontend, inspired by Zoho's UX and modular patterns) to replace our current setup for a retail chain in India. Existing ops: 340+ franchise outlets (FOFO) + 10+ company-owned (COCO), scaling hard to 140+ COCO, exploding userbase, and branching into new verticals beyond pharmacy (clinics, diagnostics, wellness, etc.). The must-haves that keep us up at night: • Ironclad inventory control (zero tolerance for ghost stock, unbilled inwards, POS-inventory mismatches) • Head-office led procurement (auto-POs, MOQ logic, supplier consolidation) • Centralized product master (HO-locked SKUs, batches, expiries, formulations) • Locked-in daily reconciliations (shift handover, store closing) • Bulletproof multi-tenancy isolation (FOFO/COCO hybrid + investor read-only views) • Deep relational data chains (items → batches → suppliers → purchases → stock → billing) Current system: On MS SQL Server, holding steady for now, but with this rebuild, we're debating sticking relational or flipping to NoSQL (MongoDB, Firestore, etc.) for smoother horizontal scaling and real-time features as we push past 500 outlets. Quick scan of Indian retail/pharma ERPs (Marg, Logic, Gofrugal, etc.) shows they mostly double down on relational DBs (SQL Server or Postgres)—makes sense for the transactional grind. What we've mulled over: **MS SQL Server:** ACID transactions for zero-fail POs/reconciliations, killer joins/aggregates for analytics (ABC analysis, supplier performance, profitability), row-level security for tenancy, enterprise-grade reliability. **NoSQL:** Horizontal scaling on tap, real-time sync (live stock views), schema flex for new verticals—but denormalization headaches, consistency risks in high-stakes ops, and potential cloud bill shocks. No BS: For this workload and growth trajectory, does staying relational (maybe evolving MS SQL) make more sense, or is NoSQL the unlock we're overlooking? Who's built/scaled a similar multi-outlet retail ERP in India from the ground up? What DB powers yours, and why? Any war stories on Zoho-inspired builds or relational-to-NoSQL pivots? Appreciate the raw insights—let's cut through the hype. **TL;DR:** Ground-up ERP rebuild for 500+ outlet retail chain in India—stick with MS SQL Server for ACID/relational power, or pivot to NoSQL for scale/real-time? Need brutal takes on pros/cons for transactional inventory/procurement workflows.