New Harvest Finance & Analytics

Starting from scratch: a chart of accounts designed for meaningful reporting (not just compliance), tracking categories aligned to the business's cost structure, bank feeds configured, and a clean cutover from manual records. This is the foundation — the structure set up here shapes every report and analysis that follows.

Decisions made at this stage — which categories to track, how to split revenue lines, what cost centres matter — compound in value over time. Done right, the accounting system becomes a genuine source of business insight rather than a compliance obligation.

A Python application that connects to the Xero API via OAuth 2.0, pulls financial data (invoices, bills, journal entries, contacts), transforms and normalises it, and makes it available to Power BI — without needing a commercial connector. Runs on a schedule so Power BI always has current data without any manual export or intervention.

Commercial Xero→Power BI connectors exist but cost money, impose constraints on data shape, and are a black box. Building the bridge myself meant I could pull exactly the fields needed, flatten nested objects for easy querying, and control the refresh schedule. When the Bridge App runs, it triggers the Power BI dataset refresh — closing the automation loop.

A web application for logging hours by project and team member — capturing the operational data that accounting software alone can't see. Tracks time entries, project codes, and billing rates so the analytics layer can calculate effective hourly rates, project margins, and utilisation without manual exports.

The key insight here is that financial data and operational data need to be aligned to the same project and client identifiers. The ops app is designed from the start to share those keys with Xero — making the join in Power BI trivial rather than a data-wrangling exercise.

Dashboards that combine Xero financial data (via the Bridge App) and Ops App hours data in a single model. Key metrics: effective hourly rate by client, revenue vs cost by project, utilisation rates, and cash flow — all updated automatically as both data sources refresh.

The combined model is what makes the analysis interesting. Seeing revenue per project alongside hours logged per project — in the same view — surfaces things that neither dataset would show alone: which projects are actually profitable, which clients consume disproportionate time, and where the business is growing efficiently.