The manual process that wasn't scaling
Before working with Conexie, the client's billing cycle looked like this: download interval data from multiple meter portals, reformat it in Excel, cross-reference against tenant agreements, calculate charges, and then manually enter line items into Xero one invoice at a time. Each billing run took the better part of a day, and the process was repeated every month for dozens of tenants. As the portfolio grew, so did the risk of a transcription error making it into a sent invoice.
Mapping the data sources
The first step was establishing reliable data feeds from all meter sources — NEM12 files, pulse-counter data loggers, and a third-party monitoring platform the client had been using for site-level visibility. Conexie ingested all three into a unified data model, normalising intervals and resolving gaps through anomaly detection rules that flagged missing or implausible readings before they reached the billing stage.
Building the automated billing pipeline
Once the data layer was stable, Conexie built a billing pipeline that applied the client's tariff structures — including time-of-use components, network pass-through charges, and solar credits — to each tenant's metered consumption. The pipeline ran automatically at the end of each billing period and produced draft invoices in Xero, pre-populated with the correct line items, amounts, and Xero contact mappings. No manual data entry was required.
Anomaly detection as a quality gate
Rather than simply automating the existing process, Conexie introduced a validation layer: every draft invoice was checked against historical usage ranges for that tenant and flagged if any line item was outside expected bounds. This meant that the hours previously spent checking figures manually were replaced by a brief review of only the flagged exceptions — typically fewer than one per billing cycle. The client went from auditing everything to auditing only the outliers.
The outcome
The client now runs their full billing cycle — across all tenants, all sites — in a fraction of the time it used to take. Error rates dropped to near zero because human transcription was removed from the process entirely. The team redirected the recovered hours toward client relationship work and portfolio expansion. The Conexie platform also gave them an audit trail for every invoice: every dollar amount traceable back to a timestamped interval reading from a specific meter.