A charging platform holds sessions and energy. Your business holds customers, invoices, assets and support cases. The join between them is usually built last, done manually first, and becomes the operational bottleneck exactly when the network grows.
That timing is not coincidence. Manual reconciliation works at small volume, which is precisely when nobody prioritises replacing it.
What actually needs to flow
- Billable sessions into finance, at the granularity invoicing requires.
- Customer and account identity, so charging users map to entities you already know.
- Asset records, so chargers appear in the same inventory as everything else you maintain.
- Faults into the support system, so charging incidents follow existing process.
- Energy consumption into reporting, where sustainability or cost allocation requires it.
Identity mapping is the hard part
A driver in the charging platform, a customer in the CRM and an account in the finance system are three records that must refer to the same entity. Where they are created independently, they diverge, and reconciliation becomes fuzzy matching on names.
Deciding which system is authoritative for identity, and having the others reference it, is the decision that prevents this. It is easy at the start and painful afterwards.
Silent failure is the dominant risk
An integration that stops without alerting anyone is discovered at month end, when the gap is already large and the underlying records may have rotated out of retention.
Every integration needs visible failure, a retry path and a reconciliation check that counts records at both ends. Without the last one, an integration can appear healthy while dropping a proportion of what passes through it.
Version changes on either side
The charging platform releases, the ERP is upgraded, an API version is deprecated. Integrations break at these moments, and the party who changed usually does not know the integration exists.
Naming an owner for each integration, who is notified of releases on both sides, is the control. Unowned integrations fail quietly and stay failed.
Do not rebuild the charging platform in the ERP
A common failure is pushing so much charging detail into the finance system that it becomes a second, worse charging platform. Finance needs billable outcomes, not meter samples.
Deciding the granularity deliberately, and keeping detail in the system designed for it, keeps both systems doing what they are good at.
Build it before you need it
The right time is while volume is low enough that errors are correctable by hand. Building it under pressure, with a backlog of unreconciled months, is considerably harder and the pressure encourages shortcuts that persist.