How a Cox's Bazar guesthouse stopped losing restaurant charges off the room bill

Kamal runs a 14-room guesthouse a short walk from Cox's Bazar's main beach road. Twelve rooms go to tourists, two are held for long-stay guests, and the attached dining room seats around thirty — doing brisk breakfast and dinner trade even from people who aren't staying the night.

It ran on a hardbound diary for bookings, a spike of carbon-copy chits for restaurant orders, and a calculator at checkout. By his own account it worked. Which was true, in the sense that rooms got booked and meals got served and the season didn't bankrupt him. It did not mean the numbers were right.

Two lists, not one

Kamal keeps the booking diary at the desk. His brother-in-law covers evenings when Kamal is out, and his night staffer covers the rest. All three take phone bookings.

One afternoon the diary travelled with Kamal to the bank. His brother-in-law took two bookings that evening and, with no diary to write in, kept them in his own phone's notes app — meaning to transfer them later. He forgot one. During Eid week, that room sold a second time to a family who'd booked three weeks earlier and already paid a deposit over bKash.

The diary wasn't the problem. A diary only prevents double booking if every booking reaches it the moment it's made, by everyone taking bookings. The moment a second, informal list exists anywhere else — a phone, a memory, a scrap of paper "for later" — the collision is just waiting for a busy week.

Where the restaurant money actually went

Guests staying in a room often ate at the restaurant and asked for the meal to go on the room bill instead of paying cash at the table. The waiter would scribble "Room 4 — add to bill" on the chit and leave it on a spike by the register.

At checkout, whoever was closing the guest's bill had to physically find every chit for that room, across however many days the guest had stayed, and add them up by hand. On a quiet night this worked fine. On a full house on a Friday, chits got pinned under other chits, written in a hurry and hard to read, or simply missed because the person at checkout didn't think to ask.

When Kamal eventually sat down and reconstructed a month from chits, receipts and the diary, close to nine percent of the restaurant charges that should have reached a room bill never did. Not theft — chits that existed, were legitimate, and were never found at the right moment.

The diary could tell him a room was booked. Nothing could tell him, at checkout, whether every meal that room had eaten was actually on the bill.

The rollout, and what had to happen first

One device becomes the only place a booking exists

The structural fix for the double booking wasn't a better diary — it was removing the possibility of a second list. Kamal, his brother-in-law and the night staffer all agreed on one rule before anything else changed: a booking isn't real until it's entered on the one device at the desk, full stop. No phone booking is confirmed to a guest until that happens, even if it means asking the caller to hold for thirty seconds.

Restaurant orders tagged to a room post straight to that room's bill

When a server opens an order for a table and marks it as a room guest rather than a walk-in, the amount is added to that room's running bill immediately — not later, not from a chit, not from memory. The chit still goes to the kitchen for the food itself, but it stopped being the thing that decided whether the charge reached the bill.

New guests went in fresh; existing stays finished on paper

Rather than try to retroactively enter guests who were already mid-stay, Kamal started every booking made from a chosen Monday onward directly in the app, and let anyone already checked in finish their stay the old way. The overlap lasted four days and caused exactly one near-miss — a returning guest's extended stay got logged in both places for one night — which is the one thing he'd change about the rollout.

What changed

• One room board, one truth. Every booking goes through the same device the moment it's made. No second list has existed since, and no room has sold twice since.
• The restaurant bill finds the room automatically. A charge tagged to a room is already on that room's folio before the guest finishes eating — nothing to reconstruct at checkout.
• Checkout stopped being a stressful reconstruction. The guest's total is already sitting there when they come to the desk, rather than being assembled from chits while they wait.
• Storms stopped being a business risk. Cox's Bazar's mobile signal gets unreliable during monsoon weather and peak-season congestion. Because the app works offline, a dropped connection no longer means a guest can't be checked in or a bill can't be closed.
• The night staffer uses it in Bangla. He'd been quietly working around an English-only booking sheet for two years. The interface language switch took him from reluctant to the fastest person at the desk.

What we'd do differently

1. Cut over on a single day, not a four-day overlap. The overlap window was meant to be reassuring and instead created the one double-entry of the whole rollout. A clean cut, with the old diary physically put away, would have removed the temptation to fall back on it.
2. Train the waiters on the "tag to room" step before the season, not during it. It's a one-second habit once it's automatic, but the first week of Eid rush was the wrong time to be learning it for the first time.
3. Run a day-end report from day one. Kamal didn't look at the daily summary for the first two weeks because he didn't yet trust the numbers. Checking it daily from the start would have caught small entry mistakes while they were still small.
4. Set up the encrypted backup before the first busy weekend, not after. It was configured in week three, after a phone was briefly misplaced during a changeover. It should have been step one.

The transferable lesson

For any small hospitality business that combines accommodation with food service — guesthouses, homestays, resorts, even a clinic with an attached pharmacy counter — the real risk is rarely the diary or the chits themselves. It's having two separate records that are supposed to describe one guest's one stay. Every reconciliation gap in this kind of business tends to trace back to a charge that lived in a different book than the one used to settle the bill. Merge the record before you merge the software, and the software mostly just has to stay out of the way.