When we take on an F&B engagement in Singapore, the first few weeks are often spent unpicking inherited mistakes. Some are operator habits. Some are software-side misconfigurations. Some are the predictable failures of an accounting setup that was designed for a generic SME and forced to fit a restaurant.
What follows is the catalogue of mistakes we've seen most often across F&B operators in Singapore — the twelve patterns that show up in nearly every inherited set of books. Each entry covers what causes it, what it actually costs, and how to fix it cleanly.
None of these are exotic. They're the boring, recurring structural mistakes that compound quietly over months — and that, taken together, are the difference between books an owner trusts and books that have stopped being useful for decisions.
1. Recording revenue net of platform commissions
What happens: GrabFood pays out S$10,000 after taking a 25% commission on S$13,333 of gross orders. The bookkeeper records S$10,000 as revenue. The gross is invisible. Commission cost is invisible. The actual margin per channel is impossible to measure.
What it costs: every operator decision about delivery platforms — whether to stay on Foodpanda, whether to push back on a commission rate, whether to absorb a promotional cost — is being made without the underlying numbers. The owner can't tell whether delivery is actually profitable as a channel.
How to fix it: record gross revenue and platform commission as separate lines. Each platform commission rate becomes visible as a percentage of channel revenue and a percentage of total revenue. The first time we do this for an operator, the reaction is usually surprise at how much commission has been absorbed silently.
2. Treating staff meals as a black hole on COGS
What happens: kitchen staff eat what they make. Some operators absorb it into food cost and never adjust; others let it inflate COGS quietly until the food cost percentage looks unhealthy.
What it costs: COGS overstated by 1-3 percentage points across the year. Margin reports that don't reflect actual unit economics. Decisions to raise prices or cut dishes made on the wrong base.
How to fix it: define a staff-meal policy in dollar terms and post a monthly adjustment from food cost to staff welfare or labour-related cost. Doesn't need to be exact; needs to be consistent. The chart of accounts should have a dedicated staff-meal contra account so the treatment is visible and defensible.
3. Petty cash quietly disappearing into the P&L
What happens: cash is taken from the till for small supplier purchases, taxi fares, urgent staff reimbursements, and so on. Receipts are forgotten. By month-end the cash balance per the books doesn't match the till count, and the difference is plugged.
What it costs: visibility into small operating expenses is lost. Anti-fraud control disappears. At audit (if you have one), persistent petty-cash variances are a flag.
How to fix it: tighten the discipline — every petty cash withdrawal logged at the time, with receipts attached within 48 hours. Modern alternative: replace petty cash with a corporate card (Aspire works well here) so every transaction is captured with receipt automatically.
4. Multiple POS systems with no unified chart-of-accounts mapping
What happens: each outlet has a different POS, often inherited from previous tenants. Outlet A's “Food revenue” doesn't map to Outlet B's “Sales — kitchen”. Imports into Xero are inconsistent. Outlet-level reporting is impossible without manual reconstruction.
What it costs: the owner can't reliably compare outlets. Underperforming locations are masked by aggregate revenue. Cost benchmarks across outlets are unusable.
How to fix it: define a canonical chart of accounts at the Xero level, then map each POS's daily export into that canonical structure. Where POS systems differ, standardise the export, not the POS itself (replacing POS is a bigger project). Long-term, consolidating onto one POS is usually the right answer, but it can be phased.
5. Delivery platform refunds posted as “other income reductions”
What happens: GrabFood refunds a customer for a bad order. The platform deducts the refund from the next payout. The bookkeeper sees a slightly smaller payout and codes the variance to a catch-all expense account.
What it costs: refunds aren't visible as a rate or a trend. A spike in refunds — usually a quality or operational issue — doesn't surface until the owner asks. By then the issue has run for months.
How to fix it: refunds get their own revenue contra account, by platform. Refund rate by platform becomes a monthly KPI. Spikes are caught early.
6. GST on supplier invoices claimed inconsistently
What happens: F&B operators deal with a mix of GST-registered suppliers and non-registered ones. Tax codes get applied inconsistently — sometimes standard-rated input claimed correctly, sometimes a non-GST supplier incorrectly coded as if GST applied. The GST F5 is filed with errors that nobody catches until IRAS does.
What it costs: at best, a corrected F5 (GST F7). At worst, interest, penalties, and an IRAS query that costs time across multiple periods.
How to fix it: discipline at supplier setup. Every supplier in Xero should have their GST status confirmed and recorded. New suppliers should have GST status checked before the first invoice is paid. Tax codes should default per supplier, not be selected per transaction.
7. Service charge and tips treated as revenue
What happens: dine-in service charge is rolled into revenue. Tips on card are treated the same. Neither is adjusted for the portion that's distributed to staff.
What it costs: revenue is overstated, manpower cost is understated. The labour cost ratio looks healthier than it is. Profit per cover is misleading.
How to fix it: service charge and tips both get their own revenue lines. The portion distributed to staff flows through a separate payable account and is recognised as labour cost when paid out. The owner's real margin per cover becomes visible.
8. Inventory recorded only at year-end stock-take
What happens: stock is counted once a year (or never). Food and beverage purchases get expensed as paid. COGS for the year is reconstructed retrospectively from the year-end count.
What it costs: the operator has no monthly COGS visibility. Food cost as a percentage of revenue isn't known until annual accounts are produced — ten or eleven months too late to act on.
How to fix it: monthly stock-take discipline, at minimum. Better: an inventory management system connected to Xero so stock movement feeds COGS continuously. The stock-take then becomes a monthly verification, not a monthly construction.
9. Bank reconciliation lagging the month by weeks
What happens: the bookkeeper reconciles the bank on a monthly cycle — meaning June bank transactions get reconciled in mid-July at the earliest. Disputes, duplicate charges, missing deposits surface weeks after they could have been resolved.
What it costs: cash position is always lagged. Reconciliation is a cliff at month-end instead of a daily discipline. Real cash issues — a missed deposit, a chargeback, a fraudulent transaction — get caught late.
How to fix it: bank feeds wired into Xero, with daily (or every-other-day) reconciliation. Recurring rules coded for known suppliers, payroll, subscriptions. The month-end close then has zero bank-related work to do — it's already current.
10. Receipts living in WhatsApp groups
What happens: managers and staff photograph receipts and post them to a WhatsApp group. Somebody is supposed to download them and file. Usually it happens intermittently. By close, half the receipts are missing and the bookkeeper is chasing five different people for September's expenses.
What it costs: cost coding is delayed and inaccurate. Cash and credit card expenses can't be matched to receipts. GST input claims are missed because the supporting documentation isn't available.
How to fix it: a capture-at-source workflow. Receipts routed automatically from WhatsApp or email into the accounting system, classified by AI, attached to the right transaction, reviewed by a human. The discipline shifts from chasing at close to catching at the moment of capture.
11. CPF contributions missed or late
What happens: a new staff member is hired, payroll starts, but CPF setup lags by a month or two. Or CPF is processed but payment slips past the 14th-of-following-month deadline because nobody is owning the calendar.
What it costs: CPF interest charges, which compound monthly. In serious cases, MOM enforcement and reputational damage with the team.
How to fix it: CPF setup happens before the first payday for any new hire — not after. Payment is calendar-driven, not reminder-driven. A connected payroll system handles the timing automatically. CPF should be one of the most boring obligations in the business; if it isn't, something is broken.
12. ECI ignored because the owner thinks it doesn't apply
What happens: the owner has heard there's an ECI waiver for small companies. They assume they qualify. They don't check. Three months after FYE comes and goes without an ECI filing. IRAS issues a notice.
What it costs: late-filing penalties, IRAS scrutiny across the year's filings, and a stressful back-and-forth that should never have happened.
How to fix it: assess ECI applicability formally each year. The waiver criteria can change; confirm against current IRAS guidance. If in doubt, file — a nil ECI is easier than a late one.
The pattern beneath the mistakes
Most of the twelve mistakes above share a common cause: accounting that was designed for a desk business is being asked to handle F&B reality. Generic chart of accounts. Reactive bank reconciliation. Receipt capture as a manual chase. No channel- or outlet-level visibility. Tax codes set transaction-by-transaction instead of supplier-by-supplier.
The fix isn't to do the same accounting better. The fix is to design the accounting setup around the specific operating reality of F&B from the start. Chart of accounts cut by outlet and channel. Bank feeds and reconciliation rules. Capture-at-source for receipts. Inventory connected to COGS. Discipline at month-end replaced by discipline throughout the month.
That's the work. The mistakes are downstream of the design.
The good news: every one of these mistakes is fixable. The better news: the fixes compound. The first month after a properly designed F&B accounting setup is in place looks materially different from the month before — and gets better from there.
— The Acuitas team