From Cash Records to Management Control

How AI-supported decision-making helped a growing manufacturer transform its finance model in six weeks

A rapidly growing export manufacturer in Vietnam had reached the point where its financial control model needed to change.

The company employed several hundred people, generated approximately US$8 million in annual turnover and was planning further growth under the CEO’s expansion strategy. It had developed through entrepreneurial drive, customer responsiveness and the determination of experienced people who understood how to keep the business moving.

The finance model had evolved alongside the company. For many years, cash-based accounting had provided a practical way to record money received and payments made. The finance team understood the company’s established routines and was competent in maintaining the bookkeeping processes that supported them.

As the business became larger and more complex, however, those routines were no longer providing management with the information needed to understand performance or support future growth.

Management needed a reliable monthly view of revenue, profitability, working capital, liabilities, customer balances, supplier balances, inventory movements and operational commitments. Knowing what had moved through the bank was no longer enough. The company also needed to understand what had been earned, what had been incurred, what belonged to the current reporting period and what remained outstanding.

The company was not simply changing an accounting method. It was changing the way management saw the business.

How the existing finance model worked

The company operated in a fast-moving export manufacturing environment. Orders were received from overseas customers, materials were sourced from multiple suppliers, products moved through manufacturing and quality processes, and finished goods were shipped internationally.

Every customer order created a flow of operational and financial information across customer support, merchandising, procurement, production, logistics and finance. For the accounts to provide a reliable view of the business, information from those activities had to reach finance accurately, consistently and at the right time.

The company used Xero as its accounting platform. Xero was practical, accessible and well suited to many small and medium-sized businesses. However, the system had been configured around the company’s long-established cash-accounting routines rather than the more complex financial-control requirements of a growing export manufacturer.

Customer deposits and advance payments to suppliers were material. Under the existing process, customer deposits were posted directly to revenue and supplier advances directly to cost. Sales invoices were subsequently posted net of deposits already received.

This meant that the timing of cash movements could materially distort monthly profitability. A large customer deposit might increase reported revenue before the related goods had been produced or delivered, while an advance payment to a supplier could be treated as a current cost even though the related materials had not yet been received or consumed.

The accounting records therefore did not consistently show what had been earned or incurred during each month.

Some important accounting information was also maintained outside Xero. Customer support used spreadsheets to track customer balances and prepare sales-invoice information before passing it to finance for posting. Supplier invoices were authorised by buyers, and there was no complete three-way matching process connecting the purchase order, receipt of goods and supplier invoice.

Finance recorded the transactions, but it did not fully control the upstream processes that created many of the accounting entries.

In practical terms, the company had an effective record of cash movements, supported by invoices and operational spreadsheets, but it did not yet have one integrated and reliable financial view of what customers owed, what remained payable to suppliers, what represented current income or expenditure, and what should remain on the balance sheet for future periods.

The issue was therefore larger than the configuration of the accounting software. The operating model surrounding finance also had to change.

The management problem behind the accounting problem

The visible requirement was to move from cash-based accounting to accrual-based financial reporting.

The underlying management challenge was more complex.

Accrual accounting required the company to recognise revenue and expenditure in the periods to which they related rather than simply when cash was received or paid. This introduced the need for reliable month-end cut-off procedures, reconciliations, accruals, prepayments, deposits, advances, receivables, payables and disciplined opening balances.

Before those processes could operate reliably, the existing records had to be understood and validated.

Bank accounts, customer balances and supplier balances had not always been kept fully aligned. Control accounts had not been reconciled consistently. Historical sales and purchase balances required review before management could establish a dependable opening balance sheet.

The records themselves were only one part of the challenge. The existing finance team had worked within the company’s cash-accounting environment for many years. There was no qualified accountant inside the team with previous experience of leading a transformation of this scale.

At the same time, three separate legal entities had become mixed within the same accounting environment. The transition therefore had to address not only the timing and classification of transactions, but also the legal entity to which those transactions belonged.

The company could not simply switch on accrual accounting and expect the existing processes to adapt around it.

Responsibilities had to change. Information previously controlled through spreadsheets needed to move into the accounting system. Finance needed greater ownership of the processes creating accounting entries. Existing balances had to be reconciled. New accounting treatments had to be understood. Procedures had to be documented, communicated and applied consistently.

The transition also had to be introduced without overwhelming the existing team. Moving too quickly could create confusion, resistance, incorrect postings or the loss of experienced staff whose knowledge remained important to the business.

Moving too slowly would leave management relying on financial information that was no longer adequate for the company’s size and growth plans.

Six weeks to redesign the model

The company wanted the new accrual-accounting model to go live on 1 January, the beginning of its new financial year.

There were approximately six weeks to prepare.

That deadline created a practical management question:

What had to change before 1 January, what could safely be deferred until after go-live, and in what sequence did the work need to occur?

The order mattered.

Historical balances had to be reviewed before reliable opening balances could be established. Accounting treatments had to be agreed before procedures could be written. The chart of accounts had to support the new model. Transactions had to be tested before they were introduced into the live environment. Staff needed to understand the new processes before they became responsible for applying them.

A decision made in the wrong sequence could create extensive rework later and place the 1 January deadline at risk.

The timing created additional constraints. The transition crossed the year-end holiday period and had to take account of the operational build-up ahead of the February Tet factory shutdown.

Most importantly, the business had to continue operating while the finance model changed.

Customers still needed to be invoiced. Suppliers still needed to be paid. Bank balances still had to be monitored. Production and shipments could not stop while historical records were reviewed and new accounting processes were designed.

Management still needed information throughout the transition.

The project was therefore not simply an accounting conversion. It was a live management transformation involving systems, processes, information, responsibilities, people, historical data and operational continuity.

How AI supported management decision-making

AI was used throughout the project as a management decision-support capability, not as a replacement for accounting judgement or management accountability.

Its most important contribution came during the design of the transition.

The project contained many dependencies, some obvious and others less visible. Decisions about opening balances affected reconciliations. Decisions about customer deposits and supplier advances affected both the balance sheet and monthly profitability. Changes to accounting treatments affected the chart of accounts, procedures, staff responsibilities and reporting.

AI provided a continuing sounding board through which management could explore those relationships.

Assumptions were challenged. Alternative approaches were considered. Different implementation sequences were stress-tested. Potential consequences were examined before decisions were made.

This helped management distinguish between work that was essential for a controlled 1 January go-live and work that could reasonably move into the post-implementation phase without weakening the new financial model.

That distinction reduced pressure on the project. Rather than attempting to solve every historical issue and redesign every process within six weeks, management could concentrate on establishing a reliable opening position and the controls necessary for the new accounting model to operate successfully.

AI also helped identify where apparently convenient shortcuts could create future rework, accounting inconsistencies or control weaknesses.

The resulting implementation plan was not produced by AI independently. It emerged through repeated discussion in which management experience, accounting knowledge, operational understanding and AI-supported analysis were brought together.

Supporting the implementation

Once the transition approach had been established, AI was used to support much of the detailed analytical and implementation work.

Historical transactions covering approximately three years were analysed to assist the reconciliation of bank accounts, customer balances and supplier balances. This helped identify inconsistencies, classification issues and areas requiring further investigation before reliable opening balances could be established.

AI was also used to research and challenge proposed accounting treatments and identify matters requiring professional consideration across the three legal jurisdictions involved.

A separate Xero test environment was created so that the proposed accounting structure could be evaluated before changes were introduced into the live system.

Test transactions included both normal day-to-day activity and less common edge cases. This allowed management to examine how the new chart of accounts, transaction treatments and reporting structure would behave under different circumstances.

The testing also exposed system constraints. Some proposed transactions were rejected because of Xero limitations, including restrictions that were not clearly documented or differed between system versions.

Identifying those constraints in the test environment avoided significant rework and reduced the risk of disrupting the operating business during implementation.

AI also supported the development of procedures, work instructions, training material and internal communications in both English and Vietnamese. Because the documents were developed within the same project context, they could reflect earlier discussions, agreed decisions and implementation priorities consistently.

This reduced the documentation burden on the project team while helping the finance staff understand not only what was changing, but how the new processes were intended to operate.

The result

The new financial model went live on 1 January as planned.

The finance team continued its normal day-to-day responsibilities throughout the transition, and the implementation did not create the operational disruption that management had sought to avoid.

January became the company’s first month-end under the new accrual-accounting model.

The month-end process was completed in accordance with the new accounting procedures, and the January profit-and-loss statement was delivered by 6 February.

The finance team operated normally, with little additional overtime required.

Following the close, AI was used to review the month’s general-ledger entries against the expanded chart of accounts. Despite the number of new accounts and the significant change in accounting treatment, only two mispostings were identified.

The result was more than a successful accounting conversion.

Management now had a more reliable monthly view of financial performance, customer balances, supplier obligations and working capital. Information previously dispersed across cash records, invoices and operational spreadsheets had begun moving into a more integrated financial-control environment.

Finance was also beginning to assume greater ownership of the processes and information underlying the company’s financial reporting.

Management lesson

AI added practical value throughout the project.

It helped analyse historical transactions, support reconciliations, identify classification risks, test accounting scenarios, expose system constraints and produce consistent procedures and communications.

Those contributions saved time and expanded the amount of work that could be completed within a demanding six-week implementation period.

The greater value, however, came from decision support.

Management was not faced with a shortage of tasks. It was faced with a complex set of interconnected decisions: what needed to change, what could remain temporarily unchanged, what had to happen first, which dependencies were critical, and where an apparently simple shortcut might create future risk or rework.

AI provided a knowledgeable and persistent challenge to those decisions. It helped test assumptions, examine alternatives and make less visible dependencies easier to understand. AI did a first pass on cross boarder tax implications of the change  and the need for specialist tax advice.

Management retained responsibility for the decisions. Accounting judgement remained essential. The experience and knowledge of the existing finance team remained important.

AI strengthened the process through which those capabilities were brought together.

The company began the project with a finance model that recorded cash movements but no longer provided management with a sufficiently reliable view of business performance.

Six weeks later, it entered the new financial year with accrual-based accounting, a more disciplined financial-control model and a finance team capable of operating it.

The transformation succeeded not because AI replaced management expertise, but because management expertise was amplified by AI at the points where understanding complexity, testing alternatives and sequencing decisions mattered most.