Restaurant accounts without an accountant: closing the week in thirty minutes.

What you collected, what you paid suppliers, what you paid staff, and what cash is left. Every other line in a restaurant's books is a breakdown of one of those four. If you can answer them on Sunday night for the week that just ended, you are keeping accounts, whatever you keep them in.

Weekly, because a restaurant runs on two clocks. GST is monthly: GSTR-1 by the 11th (ClearTax) and GSTR-3B by the 20th (RegisterKaro), both read 2026-09-09. Cash and supplier dues move every day. No Indian authority prescribes a weekly close, so nobody will make you do it, and that is why it slips. A wrong price on a thali or a delivery billed twice costs you four weeks before a monthly close notices.

Plenty of Hinglish shop-accounting writing covers the first half of this, and a dukan ledger tracks what you owe and what is owed to you. It does not carry the two things a restaurant needs: the split between cash, UPI and card, and what each dish sold.

Ten minutes on the last night of the week is worth an hour of remembering on Tuesday. Count the drawer and write the figure down. Keep a slip for every rupee that left it. Stack the supplier bills nobody has paid yet. Those three piles are the weekly review, and none of them survives being rebuilt from memory.

The timing argument is worth borrowing from RestaurantOwner.com's "Critical Numbers" (read 2026-09-09, a US page in dollars, so take the practice and not the figures): a week that ends Sunday and read on Monday catches stock at its lowest, and next week's roster can still change while the numbers are fresh.

On the register the counting lives in Orders, then Cash box. Start the cash shift with an opening float, and record every Paid out and Cash in with an amount and a note saying what it was for, listed as "This shift's slips". Count and close asks for the count first: you type what is in the drawer before the register shows anything, and only then does it display Expected, Counted and Variance, reading "balanced", "short by ₹100" or "over by ₹40". Every row is stored against whoever was signed in, and the till, though the slip list itself shows the time, the note and the amount.

Counting before you see the expected figure matters: a counter who knows the till expects ₹18,450 arrives at ₹18,450. The cash handling post covers floats, paid-outs and variance; the day-end report is the nightly version of this page.

A closed period stops moving. Sales collected, online payouts, expenses paid, staff paid, profit this period, and the cash carried forward that opens the next one: once you press Close this period, those numbers are frozen and next week begins from a figure the books derive rather than from a running total nobody closed. Your drawer count is the check against it, not its source.

Late paperwork is normal and does not force you to reopen anything by hand. A bill for last Thursday that arrives on Tuesday makes the period say "Changed since you closed" and offer Recalculate, which refreezes the week and carries the new closing figure into the period after it. A brand-new store shows a "Start your first period" card above the reports, because there is nothing to close yet. The close email is on by default: everyone who signs in to the dashboard gets it once per period, so the person who pressed the button is not the only one who saw the numbers, and a Recalculate does not send it again. The button-by-button version is the weekly accounts guide.

The expenses number is only as good as the habit behind it. A bill entered when it arrives and marked paid when it is paid lets the Expenses screen answer both questions that matter: what went out this week, and who is still owed. A bill entered on Sunday from a pile answers neither, because the pile is never complete.

Expenses are kept by vendor, each one Paid or Unpaid, with "What we owe suppliers" totalling the second kind. "Regular costs" spots the bills your own history repeats, rent, gas, the internet line, and offers to fill one in, flagged when it is overdue. Nothing is recorded until you check the amount and save it.

A paid-out slip is not an expense entry. The ₹450 of vegetables taken from the drawer at 11am is a Paid out in the register's cash box, and its job is to explain why tonight's drawer is ₹450 light. The supplier bill in the dashboard's Expenses is what the week's accounts read. One balances the drawer, the other says what the kitchen cost.

The Profit and loss tab runs gross sales, discounts, written off and refunds down to net sales, then cost of goods, gross profit, labour, prime cost, occupancy, controllable costs and net profit, each with its share of sales beside it. That share is where the labour ratio lives, and cost of goods is the honest limit: it is what you filed as purchases. If the last line surprises you, ask which line above it moved.

Two published Indian benchmarks are worth knowing, both from POS vendors rather than from a survey, so read them as a direction. Petpooja's food cost calculator page (11 May 2026, read 2026-09-09) puts full-service food cost at 28% to 32%, and says above 35% means waste, over-purchase, theft or wrong pricing. DineOpen's margins guide (12 March 2026, read 2026-09-09) gives 28% to 35% for food and 20% to 30% for labour including PF, ESI and overtime, a labour band other vendors put several points higher. On net margin the two disagree openly: Petpooja says 3% to 5% is typical, DineOpen gives 10% to 35% by format. Neither states a method, so the comparison to trust is your own last four weeks.

The Labour tab sits beside it with the hours and the cost by person, read field by field in the staff and labour post. The rupees-per-hour line is the underused one: a percentage moves when sales move, while sales per labour hour tells you whether Tuesday evening needs four people.

Cost of goods here is your purchases, not a measured consumption. The page says so on the line itself: cost of goods is your purchases in this period, and a stocktake-based figure is not available yet. There is no recipe module and no stocktake screen, so a heavy buying week lands in that week whether or not the kitchen used it. Petpooja and Restroworks (both read 2026-09-09) ship inventory and recipe modules if you need consumption rather than purchases. The "Cost to make (₹, optional)" beside a dish's price in Menu and stock does not touch this line: it feeds the Menu report, which is menu engineering.

What an accountant asks for is monthly and short: sales with the tax split, the bills you issued and their numbers, expenses with GST where it applies, and staff pay. The weekly close is not for him. It is what makes those questions answerable in one sitting instead of reconstructed from a shoebox.

• The returns, and how long you keep them. The monthly filings, the bill-number range Table 13 asks for and the 72 months you have to keep the records are the GST post's whole subject, sourced there.
• Pay staff on the clock the law sets. Section 17 of the Code on Wages requires monthly wages before the expiry of the 7th day of the following month, and a full and final within two working days of a resignation or dismissal (section 17, read 2026-09-09). PF and ESI sit on top of the salary, so check those thresholds with your CA against EPFO and ESIC, not against a vendor's blog.

There is no export button today. The numbers sit on the Sales and Expenses tabs for whichever period you pick, and a CA usually wants totals rather than a file. If yours wants every bill, the Orders screen keeps them: Open, Today and Earlier, and Earlier takes a date, one business day at a time.

Half an hour on a Monday is the whole routine. EasyKOT costs ₹299 a month or ₹2,990 a year, with 18% GST as its own line, and the first three months are free with no card. Start with one week: count the drawer on Sunday, close the period on Monday, and read what the four numbers say about a week you thought you knew.