Guide · Parking

Where parking money disappears: a practical audit for building owners

Ask the owner of almost any commercial building in Dhaka what their parking made last month and you'll get a confident number. Ask what it should have made and the conversation changes.

That second number is the entire subject of this article. Without it, the first number is unauditable — not suspicious, just unverifiable. And an unverifiable revenue stream will lose money over time regardless of how honest everyone involved is.

The six leaks, roughly in order of size

1. The vehicle that was never recorded

A bike arrives, the queue is four deep, the operator waves it in and intends to write the slip afterwards. Sometimes the slip gets written. Sometimes it doesn't. No slip means no fee owed at the exit, and no trace that anything happened.

This is usually the largest leak and it requires no dishonesty whatsoever — only pressure and a slow recording method. It is also the one that responds best to a fix, because the cause is mechanical.

How to measure it: stand at the entrance for one hour at peak and count vehicles by hand. Compare against the slips or records issued in that same hour. Repeat on a different day. The gap is your floor, not your ceiling.

2. Fees calculated in someone's head

Duration times rate, under pressure, with a queue forming. Errors go in both directions, and the direction that favours the customer is the one that gets negotiated for. Over thousands of transactions a systematic downward bias is worth real money.

How to measure it: take twenty completed exits at random and recompute each from the recorded entry time and the rate card. Count how many match exactly.

3. Uncollectable disputes at the exit

A customer disputes the duration or the rate. There's a queue. The operator discounts to clear the lane. Individually reasonable; collectively expensive, and completely invisible because the discount is never recorded as a discount.

How to measure it: you mostly can't, directly. Proxy it by asking operators how many arguments they had yesterday. The number is never zero, and it correlates with how prominently rates are displayed.

4. Shift handovers with no expected total

If a shift ends with cash counted but no expected figure, a shortfall cannot be detected, let alone attributed. Every shift becomes a small act of faith. Over a year, faith is expensive.

How to measure it: ask whether you can produce, for last Tuesday's evening shift, the expected collection and the operator responsible. If not, this leak is unbounded.

5. The unrecorded free pass

Staff vehicles, tenant vehicles, a friend of someone, deliveries. All perhaps legitimate. But if free entries aren't recorded as free entries, the category quietly expands, and there's no way to know how much capacity it consumes at peak.

How to measure it: count vehicles present during a peak hour and compare against paid tickets open at that moment. The difference is your true free-pass volume.

6. Records that can be edited after the fact

The smallest leak in volume and the most corrosive in principle. If a completed transaction can be altered or deleted afterwards, then no report derived from it can be trusted, and every other control you build sits on sand.

How to measure it: ask whether yesterday's records can be changed today, and by whom. If the answer is anything other than "they can't", note it.

Fix them in this order

Sequence matters more than tooling, because each step makes the next one measurable.

1. Number every vehicle. Sequential, unavoidable, issued at entry. This converts leak 1 from invisible to visible and is a precondition for everything else.
2. Take the arithmetic away from people. Fee computed from the recorded entry time, not from memory. Kills leak 2 and most of leak 3.
3. Print the rates where the customer reads them. On the ticket, in the language they actually read. Moves the argument from the exit lane to the entry, where it costs nothing.
4. Attach every ticket to a signed-in operator and shift. This is what makes leak 4 measurable — a shortfall acquires a name and a time.
5. Record free passes as transactions with zero value. Not as nothing. You cannot manage a category you don't count.
6. Make records append-only. Corrections happen by adding a reversing entry, never by editing history.

Every one of these is a record-keeping change. None of them is a hardware purchase. That ordering is not an accident.

What not to buy first

Boom gates and number-plate cameras are the instinctive first purchase because they look like control. They aren't, for the money that's actually leaking. A barrier enforces a queue; it does not record a transaction, compute a fee, attribute a shift or produce an expected total. If vehicles are entering unrecorded and fees are being estimated by hand, a gate arm changes precisely nothing about either.

Buy the record first. If, after the record exists, you find that the specific problem is vehicles physically leaving without passing the exit desk, then a barrier is the right answer to that specific, now-measured problem.

The one-week version

If you do nothing else, do this. For one week, at one gate:

• Issue a numbered ticket to every single vehicle, no exceptions, including free passes.
• Write the entry time on it.
• At the end of each shift, total what the tickets say should have been collected and compare against the cash.

You will not enjoy the first three days, and you will learn more about your parking operation than in the previous three years. Whether you then automate it is a separate decision — but you'll be making that decision with a number instead of a feeling.