Insights · 5 min read
Custom accounting software starts with the reports, not the screens. What to build first, what it costs, and when buying beats building.
By GGP Editorial
Custom accounting software usually starts the same way: a finance team that has outgrown spreadsheets and generic tools that force their workflow into someone else's boxes. The question is rarely whether to build. It is what to build first and how much to spend.\n\n## Start with the reports, not the screens\n\nMost accounting products get designed around data entry. That is the wrong end. The finance team cares about three outputs: the trial balance, the P&L, and whatever custom reports their auditor asks for. If those come out right, the entry screens matter less.\n\nWork backwards. Write down the monthly close the team runs today, the reports they export, the numbers the owner actually looks at. That list becomes the scope. A ledger that produces clean, auditable reports beats a beautiful dashboard nobody uses.\n\nAsk the owner which three numbers they check on Monday morning. Build the screens that answer those three questions and stop there. Everything else is a nice-to-have that can wait.\n\n## General ledger first, automation second\n\nThe core is the general ledger with a proper double-entry structure. Every module sits on top of it: invoicing, payables, payroll, fixed assets, bank feeds. Teams that skip the ledger and bolt features onto a list of transactions regret it later.\n\nDouble-entry sounds obvious to an accountant and exotic to a developer who has never built one. Every entry needs a debit and a credit that balance, and every report is just a view over those entries. Get this right and the trial balance always ties out. Get it wrong and you spend the project chasing two-cent imbalances.\n\nHere is a realistic build order:\n\n| Phase | What it covers | Typical effort |\n| --- | --- | --- |\n| 1 | Chart of accounts, journal, reports | 6-8 weeks |\n| 2 | Invoicing, payables, bank reconciliation | 6-8 weeks |\n| 3 | Payroll, fixed assets, multi-entity | 8-12 weeks |\n\nCost scales with the number of entities and the reporting standards. A single-entity system for one company is a different project from a multi-entity setup that consolidates across currencies. The second one costs two to three times as much, and most of that is accounting logic, not screens.\n\n## Compliance is part of the scope\n\nAccounting software has to match the rules your accountant and tax authority expect. Different countries have different chart-of-account norms, tax codes, and filing formats. A system built for one jurisdiction will not drop into another without work.\n\nSouth Africa, for instance, has VAT at a standard rate plus a wide list of zero-rated and exempt items, and its filing format is not the same as Brazil's. Getting those mappings right before launch is cheaper than correcting six months of posted transactions.\n\nThis is where a partner who has seen multiple markets helps. We have delivered financial management systems and payment systems for clients across Brazil, South Africa, Singapore, and the US. Each came with its own local rules, and the differences are exactly the kind of thing that gets missed in a spec written from a template.\n\n## The parts people forget\n\nA few things get cut from the first estimate and then turn out to be unavoidable. Multi-currency: even a single-country business invoices overseas clients sometimes, and you want the gain or loss on conversion recorded, not ignored. Audit trail: who changed what and when, because an auditor will ask. Permissions: the person who posts invoices should not be the person who approves payments. Each of these is a week or two of work, and skipping them means rebuilding later under pressure.\n\n## Buy, or build the parts you cannot buy\n\nHonest advice: do not rebuild what QuickBooks or Xero already do well for your size. Build when your workflow genuinely does not fit, or when the accounting sits inside a larger product you already sell. Property firms often need rent collection tied to accounting, which is why we have built a property rental system that posts directly to the ledger.\n\nStartups get this backwards sometimes and build a general ledger from scratch when they only needed a report. If you already run a standard accounting package, ask whether an API or an export will do before you commission a build.\n\nIf you do build, keep the first version narrow. One entity, one currency, the reports your finance team actually reads. You can add payroll and multi-entity later once the close is clean. A monthly close that used to take a week and now takes a day is the only metric that matters.\n\nWe are a team of 40+ engineers that has shipped more than 300 projects, and we take work from any industry as long as it is legal. If you are thinking about accounting software, or a system that has accounting inside it, tell us what your month-end looks like.\n\n## Related reading
Tell us what you are building and where you are today. We typically reply within 24 hours.
