The same stock had conflicting stories.
Manufacturers worked across ERPs, spreadsheets, and operational records. Units of measure, locations, and transactions did not line up. When the numbers disagreed, people had to reconcile them manually before making a decision.
Customers needed agreement, not another replacement.
Discovery with 50+ customers surfaced shared concepts: items, units, locations, batches, and transactions. The systems differed in how they represented those concepts. The opportunity was a common model that could sit across them.
The decision
- Chose
- A canonical inventory model with configuration-driven mappings.
- Over
- Build a bespoke integration for each customer.
- Evidence
- Discovery with 50+ customers revealed common inventory entities across systems.
- Trade-off
- Invest in the shared model before optimizing individual connectors.
- Cost
- Upfront modeling work and tighter boundaries around custom requests.
- Tally
- Zoho
- Vyapar
- Custom systems
- Item master
- Units of measure
- Multi-location
- Batch / lot
- Transaction types
Reconciliation as a product capability.
Closphere mapped source systems into a canonical inventory model, then used that model for dashboards and alerts. Continuous reconciliation compared the result against the source ERP.
Silent-failure detection made stopped sync visible. A dashboard could look complete while its data had stopped changing, so connection health needed to be part of the product experience.
System Details
- Canonical model
- Item master, units of measure, multi-location stock, batch/lot, transaction types.
- Source systems
- Tally, Zoho, Vyapar, and custom systems.
- Mappings
- Configuration-driven translations into the canonical model.
- Sync
- Local sync and cloud aggregation.
- Reconciliation
- Continuous checks against the source ERP.
- Failure handling
- Surface mismatches and detect silent sync failures.
A foundation that customers could keep using.
Closphere reached 63 paying customers, 1K+ users, and 112 locations. Inventory record accuracy improved from 75% to 95%, stockouts fell 32%, and reconciliation effort fell 70%+.
Onboarding went from 6+ weeks to under 7 days. Referrals drove approximately 30% of new business. The business exited through a technology/IP sale.
The data model was the product.
Connectors and dashboards only became useful when the underlying meaning was consistent. As a founder, I also learned to treat team capacity as a product constraint before committing to customization.