Open-source accounting landscape: what a source survey can establish
The useful finding is not a winner: similar search labels cover different permissions, operating models and accounting responsibilities.
Key takeaways
A defensible software survey separates documented scope from tested behavior and leaves unsupported comparisons unresolved.
- The accounting category mixes business ledgers, ERP suites, invoice tools and personal-finance applications.
- Current licence files can contradict older descriptions of a project.
- Deployment language is evidence about dependencies, not a benchmark of cost or speed.
- Country features require their own current acceptance evidence.
- No common installed-product benchmark, accountant certification or customer-adoption survey was performed here.
Method and source boundary
We checked primary project sources on 16 September 2026 and classified their stated purpose, licence and deployment approach.
The selection began with projects that appear in discussions of self-hosted small-business accounting. We read official repositories and selected project documentation, then divided the field into five practical families. This is a purposive sample, not an exhaustive directory or a representative statistical sample of installations. It includes source-available and proprietary alternatives as adjacent decision categories so readers do not confuse practical relevance with open-source classification.
We did not create competitor accounts, install every product, measure throughput or interview users. Product descriptions are therefore attributed observations rather than independent capability certifications. The source date matters: a repository branch can change after publication, and an official feature page may apply to a particular edition. Before a purchasing or deployment decision, verify the exact release and agreement again. The selection guide turns these observations into a practical evaluation worksheet.
How were licences classified?
The checked licence file takes priority over a general marketing description or a historical category listing.
We use the Open Source Definition and OSI licence register, checked on 16 September 2026. A product can be useful and self-hostable while using terms outside that definition. Conversely, an open-source application may have paid hosting, implementation or support. These properties belong in separate columns. A binary free-versus-paid classification loses information that matters to a business operating its own installation.
The licence of a repository's current branch is not necessarily the licence of all previous packages. Dependencies, fonts, extensions and trademarks can have separate terms. The survey identifies the checked project position but does not provide a legal opinion about a modified hosted service or commercial distribution. Readers should preserve the licence file from the actual release being evaluated and resolve unusual uses before building a service around it.
What did the business-accounting family show?
Several projects focus on business records, but their hosting models and licence terms differ enough to prevent a single category assumption.
The Bigcapital repository describes accounting and inventory under AGPL licensing with Docker self-hosting. FrontAccounting's official mirror represents a GPL-licensed PHP accounting project. LedgerSMB describes double-entry accounting and ERP for the web. These sources were checked on 16 September 2026. Their presence establishes relevant candidates, not comparable support capacity or installed performance.
The observation suggests an evaluation principle: compare the path from a business document to the financial statement. Look for the supported source types, period controls, detailed schedules and corrections. A project calling itself accounting software does not tell you whether your partial-payment or foreign-currency case works. A project with fewer public stars may still have a suitable history outside GitHub. Repository visibility should not be substituted for a documented workflow test.
What changed the treatment of Akaunting?
Its current BSL terms require a different classification from an unrestricted open-source label.
The Akaunting licence file, checked on 16 September 2026, describes production-use conditions and a future GPLv3 change. The official Double-Entry app page presents a paid app. Together, those sources show why a survey must identify both licensing and the installed feature bundle. We do not generalize these terms to every historic version or commercial agreement.
This does not establish that users should avoid Akaunting. It establishes that source visibility, present permission and paid functionality need to be evaluated separately. A business can reasonably choose a commercial arrangement when it meets its needs. An article should not describe that choice as open source without specifying the applicable definition, nor should it characterize a paid app as an accounting defect.
What did the ERP family show?
ERP projects organize accounting alongside operational records, which changes both the value and the implementation burden.
The ERPNext, Dolibarr and Odoo repositories, checked on 16 September 2026, describe wider business suites. Their platform and edition choices differ. The survey does not infer that every module is available under identical terms in every installation. A hosted demonstration, a community repository and a partner's localized package may represent different products for practical evaluation.
Our interpretation is that operational linkage can matter more than raw feature count. A sale that affects delivery, stock, customer debt and revenue creates several records that must agree. A connected suite can reduce manual handoffs if configured well. It also requires the organization to make decisions about roles, document flow and exceptions. These are evaluation implications drawn from the product families, not a measured claim that one suite reduces errors by a particular amount.
What would establish correction quality?
Evidence must include the original record, supported correction, reporting dates and the effect of period restrictions.
ERPNext's immutable-ledger documentation, checked on 16 September 2026, provides a concrete source for its cancellation and reversal model. Dolibarr's double-entry overview provides a source for its accounting preparation process. These are different documented mechanisms, not grounds for declaring one product incorrect without examining the actual version and settings.
A useful independent study would install specified versions, post the same accepted example and compare the visible history after correction. It would test ordinary and privileged roles and repeat the operation after closing a period. It would retain expected results and report all manual steps. This survey has not done that experiment, so it avoids turning architectural preferences into claims of measured competitor failure.
What did the invoice and personal-finance families show?
Adjacent tools may solve a narrow task well without replacing the complete business accounting process.
The Invoice Ninja repository, checked on 16 September 2026, describes invoice, quote, project and time-tracking scope and a source-available model. The Firefly III repository, checked on the same date, describes personal-finance management. Neither category name should be expanded into an unverified claim about business ledgers, staff permissions or local statutory reporting.
The distinction matters when researching search results. A personal ledger can offer rigorous recordkeeping while requiring a workflow unsuitable for a team. An invoice tool can be a good customer-facing component while an accountant maintains the general ledger elsewhere. A fair survey asks which responsibility the software owns and what remains outside it. It does not punish a narrowly scoped tool for failing to be an ERP, and it does not pretend that a narrow tool provides a complete accounting system.
Why include proprietary and managed options?
A practical decision can include non-open-source products without changing the definition of open source.
The fifth family is included as a decision category rather than a vendor-by-vendor capability table. A managed service may suit an organization that cannot maintain a server. A proprietary local product may need consideration if it provides an evidenced mandatory workflow unavailable in the open-source shortlist. Such a decision still needs checks of data export, agreement terms, access controls and an exit procedure.
This survey does not publish current prices or compare country-specific plans for commercial vendors. Those require a dated review of the actual plan, modules and region. It also does not assume that paid support includes a particular response time or an accounting sign-off. The useful comparison is the complete responsibility assigned by the arrangement: who operates the system, who resolves errors, who owns the records and who can recover them.
Which claims remain unverified?
Adoption, comparative correctness, universal hosting cost and local compliance need evidence beyond the sources read here.
| Claim | Why this survey cannot establish it | Useful next evidence |
|---|---|---|
| Largest active user base | Stars and downloads do not identify active organizations | Defined, independently interpretable usage data |
| Fastest or cheapest deployment | No equal-workload benchmark was run | Common dataset, hardware, configuration and cost basis |
| Best accounting correctness | Documentation is not an acceptance test | Accountant-reviewed transaction and exception suite |
| No outbound dependencies | No installed network audit was performed | Traffic and configuration review of the exact edition |
| Country compliance | Generic tax fields cannot prove an entity's obligations are met | Current legal review and applicable authority acceptance |
How should hosting evidence be read?
A dependency list describes what must be available; it does not show what an individual provider account actually supplies.
The PHP hosting availability study demonstrates the difference between platform support, provider defaults and plan-level access. The same distinction applies to a container-based product. A deployment needs persistent storage, database administration, updates and a recoverable configuration. A language preference can simplify staffing decisions, but it cannot establish the full cost or suitability of an installation.
For repeatable research, preserve the exact release, runtime, database and dataset with the results. A test on one provider does not certify all its plans. An older minimum version in a document should be reconciled with current release notes rather than repeated as a blanket support claim. FrontAccounting's PHP 8 compatibility release announcement, checked on 16 September 2026, is one concrete reason not to rely on an old PHP-only description.
What would a stronger follow-up study do?
It would use a declared protocol, common synthetic cases and independent review before publishing comparative results.
- Define the business type and mandatory workflows before selecting products.
- Record licences, editions, versions and local modules.
- Agree expected accounting results with a qualified reviewer.
- Run normal, duplicate, correction, closed-period and recovery cases.
- Publish environment details, manual work and unresolved outcomes.
- Separate usability observations from accounting and operational acceptance.
Documented scope → reproducible experiment → reviewed results → bounded conclusion
- Comparative benchmark
- A test using a common protocol and comparable conditions across the products being compared.
- Capability claim
- A statement about supported behavior that needs evidence specific to the version, configuration and task.
Where does PHP Ledger stand in this survey?
It is an accounting-core development preview, and its limitations are part of the comparison rather than an exception to it.
The current project-owned core uses AGPL-3.0-or-later and runs on the documented PHP and MySQL profile. It provides a central posting service and source-linked reporting for its supported sources. The 0.4.0 starter adds core invoices, bills and payments, optional one-location Inventory/Purchasing and manual tax calculation. Electronic-invoice submission and country-specific tax rules remain outside its scope. Backend party, currency and correction foundations should not be described as those operational modules. A current test result can support a bounded mechanism without supporting a broader marketing claim.
This makes PHP Ledger unsuitable when a required task is missing or a stable, locally accepted system is necessary now. Its narrower scope may be useful for technical evaluation and contribution. The comparison hub gives direct reasons to select alternatives. The bookkeeping study examines another gap: a product listing cannot establish how businesses actually maintain complete books.
Questions and conclusion
Use the survey as a map of questions and sources, not as a universal ranking of accounting products.
Does a large repository prove maturity?
It is one public signal, but it does not identify active business users, response commitments or correctness for a particular workflow. Those require different evidence.
Does this research show that every project fails local e-invoicing?
No. Local capability must be assessed for the exact jurisdiction, module and entity. The survey does not establish a universal absence across projects.
Start with the five-family selection guide, then request the concrete demonstrations your business needs. All primary sources are linked in the relevant sections and were checked on 16 September 2026. The findings remain documentation-based, with installed behavior, market share and statutory acceptance explicitly unresolved.
Primary source register
The following source groups were checked on 16 September 2026; their documents support scope observations rather than installed benchmarks.
- OSI licence register ? classification reference.
- Akaunting, Bigcapital, FrontAccounting and LedgerSMB ? business-accounting sources.
- ERPNext, Dolibarr and Odoo ? ERP sources.
- Invoice Ninja and Firefly III ? adjacent product purposes.