- Role
- Owned the billing domain end-to-end
- Company
- SkyLab Group
- Period
- 2025
- Stack
- Cloud usage metering, Multi-currency, API-first integration
The problem
The insight
Billing only works once every provider's usage speaks one unit.
PAC is a cloud platform that sells compute, storage and network services to tenants across multiple regions. Underneath, four providers (AWS, Azure, Huawei Cloud and Alibaba Cloud) each price their services their own way. Left alone, that shows up as inconsistent invoices and cost that cannot be attributed to the right tenant.
So the tension was not collecting usage. It was that four price books cannot be compared, or billed from, until they are translated into one model. Try it below with made-up rates: guess which book is cheapest, then normalise.
The decision
Bill each provider's price book, or one model?
Translate the four price books into one canonical model, then meter per tenant per service.
Source: Thao's CV (SkyLab Group, PAC billing).
Ownership
What I owned
I owned the billing domain end to end at SkyLab Group: usage metering, multi-currency pricing, invoices, and revenue dashboards. I specified the API-heavy integrations that turn multi-service usage into tenant invoices, and I designed and shipped the cost monitoring dashboard and the payment workflows.
- Metering: usage per tenant, per service, across compute, storage and network.
- Pricing: one price logic across four providers, billed in the tenant's currency.
- Invoices and credit notes: issued from metered usage, with corrections made as credit notes rather than edits.
- Dashboards and reconciliation: what was used, invoiced and paid, in one view.
Four price books into one invoice
- AWS
- Azure
- Huawei Cloud
- Alibaba Cloud
- 01One pricing modelStandardised, multi-currency
- 02Usage meteringPer tenant, per service
- 03Invoices and credit notesIssued to tenants
- 04Payment reconciliationUsage vs invoice vs payment
Reconciliation
Closing the loop
An invoice is only half of billing. The other half is knowing the three numbers agree: what was metered, what was invoiced, what was paid. When they do not, the fix depends on which one is off, and an issued invoice is corrected with a credit note, never edited. Play the operator on four fictional tenants.
Cost monitoring dashboard
Unified cost attribution across AWS, Azure, Huawei and Alibaba Cloud.
Payment reconciliation
Workflows that close the gap between metered usage and issued invoices.
Pricing standardisation
One pricing logic across four providers, removing inconsistencies in invoicing and cost attribution.
Product screens and outcome metrics for PAC are private, so this page shows the logic with illustrative data rather than the product. No figures here are measured results.
What shipped
- A cost monitoring dashboard across AWS, Azure, Huawei Cloud and Alibaba Cloud
- Payment reconciliation workflows: metered usage against invoice against payment
- One pricing logic across four providers, billed in the tenant's currency
- Invoices and credit notes issued from metered usage
Source: Thao's CV (SkyLab Group). Product screens and outcome metrics for PAC are private, so no figures are claimed.