---
title: "Check-in and boarding: from paper lists to QR codes · Supaferry"
description: "What check-in is for, why the ticket can be the boarding pass, what a QR code should carry, and how two officers board 306 passengers in 90 minutes, with or without network coverage."
url: https://supaferry.com/resources/journal/ferry-check-in-and-boarding
language: en
published: 2026-04-21
---
[Journal](https://supaferry.com/resources/journal)

# Check-in and boarding: from paper lists to QR codes

2026-04-21 · 20 min read

What check-in is for, why the ticket can be the boarding pass, what a QR code should carry, and how two officers board 306 passengers in 90 minutes, with or without network coverage.

![Illustration of a large QR code next to a boarding progress bar reading BOARDED 214 / 306 and a short list of scanned ticket codes, one of them marked NOT PAID in orange.](https://supaferry.com/blog/check-in-and-boarding.webp)

*A boarding desk halfway through a call: 214 of the 306 expected passengers are on board and every scan is listed with its result. The orange line is a hold that was never paid, refused at the gangway with the reason shown to the officer.*

## Key takeaways

- Check-in exists to make the manifest match the ship, to count persons before departure and to close the gap between tickets sold and people carried.
- When the ticket already names the voyage and the call and carries a scannable code, the ticket is the boarding pass and the gangway scan is the check-in.
- A QR code should carry only the ticket code, printed again in letters underneath; the scanner checks existence, payment, call and prior use, and refuses with a reason.
- On a multi-stop voyage the desk boards by call, and the count the master signs is passengers boarded plus crew, reconciled with an independent tally before the gangway lifts.
- An offline boarding desk needs a cached list, queued scans, a sync that keeps the real scan time and a rule for conflicts; test it in flight mode before signing a contract.

Boarding is the moment a booking becomes a person on a ship. Everything before it, the search, the fare, the payment, the e-mail with the ticket, exists so that this moment takes a few seconds per passenger and leaves behind an accurate list. On most small ferry lines and inter-island services it still takes much longer than that, and the list it leaves behind is a stack of torn stubs, a tally counter and a number written by hand on the departure report.

This article describes what check-in and boarding are for, the two models operators use, what a boarding pass is and what a QR code should contain, how to run boarding at a wharf with two officers and three hundred passengers, and what to do when the gangway is on a ship anchored off a reef with no network. The procedures are the same whether the line crosses a strait in the Canary Islands, a sea loch in Scotland's Western Isles or a lagoon in the Pacific.

## Why check-in exists

Operators sometimes treat check-in as a formality inherited from airlines. It has three purposes of its own, and each one is worth more than the time it costs.

The first is **manifest accuracy**. The list of persons on board has to match reality, and reality is decided at the gangway, not at the point of sale. People buy and do not travel. People travel on a ticket bought for someone else. People turn up for the wrong sailing. A list built from sales alone will be wrong by a few percent on every departure, and nobody will know in which direction.

The second is the **safety count**. SOLAS regulation III/27 requires that all persons on board a passenger ship are counted before departure and that the number is recorded and communicated ashore. In the European Union, Directive 98/41/EC, amended in 2017, goes further: name, sex, age category and, where the passenger wants it, information on care or assistance, registered before departure and transmitted to the national single window for voyages over twenty miles. A number without names is a count. A list with names is a manifest. A search and rescue coordinator works from the second.

The third is **revenue control**. Boarding is the only point at which the operator can verify that every person on board has paid. A ticket sold is revenue. A person on board without a ticket is revenue lost. A ticket sold twice for the same seat is a complaint and, on a full sailing, a safety breach. Check-in closes the gap between tickets sold and persons carried, which is the figure a finance department, an auditor or a donor reconciles at the end of the month.

In the paper era all three purposes were served by three different objects: a passenger list typed at the agency, a stub torn by the purser at the gangway and a clicker in the mate's hand. The three numbers rarely agreed, and the reconciliation happened after departure, when nothing could be corrected.

## The counter model and the self-service model

There are two ways to run check-in, and most lines end up with a mixture.

In the **counter model** the passenger presents at a desk, an agent checks the ticket and an identity document, prints or stamps a boarding pass and adds the name to the list. The model has real advantages. A human verifies identity, handles changes, takes cash, manages groups and vehicles. It also has real costs. It creates a queue in front of the queue at the gangway. It needs staff for two hours before every sailing. And it sends the passenger towards the ship holding a second document that can be lost between the desk and the gangway.

In the **self-service model** the passenger receives a ticket with a machine-readable code at the moment of purchase, by e-mail, by SMS or printed at the agency, and presents the code at the gangway. The scan is the check-in. Two officers can board three hundred people because nobody is retyping anything. The passenger needs nothing other than the code, on a screen or on paper. The model has one dependency: the code has to reach the passenger and the device has to read it.

The sensible arrangement on a small line is a **hybrid**: the gangway scan is the rule, and a counter remains open for the exceptions, which are changes of date, payment at the counter, group leaders holding thirty tickets, and cargo. A rule of thumb: if more than one passenger in ten needs the counter, the problem is upstream of boarding, usually in how tickets are delivered.

## A boarding pass is not a ticket

The two words are used interchangeably at small ports, which hides a decision the operator should make deliberately.

A **ticket** is proof of the contract of carriage: who travels, on which voyage, from which call to which, in which class, and how much was paid. On some lines a ticket is open, valid for any sailing within a period, and is only attached to a specific departure later.

A **boarding pass** is issued at check-in. It confirms that the passenger has presented for a specific sailing, carries the seat or cabin if one is assigned, the gate or berth and the boarding time. Airlines keep the two documents separate because the pass is also the token for security screening and baggage, and because the ticket is often sold months before the aircraft is assigned.

On a ferry whose ticket already names the voyage and the call, and already carries a scannable code, the two documents collapse into one. The ticket is the boarding pass, and the gangway scan is the check-in. Keep two documents only where there is a reason: open tickets that must be assigned to a sailing, cabins allocated at check-in, vehicles that must be weighed at the port before a lane is assigned. Everywhere else the second document is a queue with no purpose.

> A boarding pass is a ticket that has been matched to a sailing. When the ticket already names the sailing, the second document is a queue with no purpose.

## What the QR code should carry

A QR code on a ticket should carry the **ticket code** and nothing else. The ticket code is the booking reference plus a passenger index: 7K3M9Q-2 is the second passenger of booking 7K3M9Q. Six characters from a reduced alphabet, with no zero, no letter O, no one, no I and no L, so that nothing is ambiguous when it is read over a phone, and no vowels, so that the reference never spells a word. One code per passenger, because the gangway boards people, not bookings.

Three things do not belong in the code. Not a URL, because then the scanner has to be online to resolve it and the officer at an anchorage is left with a dead link. Not personal data, because a printed sheet with the name, the date of birth and the passport number inside the QR is a data breach every time it is left on a bench. And not the fare, because a code that can be altered to change the price will be altered.

The same code is **printed in letters underneath** the square, in a large monospaced type. This is the fallback for every failure mode of the square: the officer types it when the scanner's camera is cracked, the passenger reads it to the agency over the phone, the agent at the counter looks it up when the printout was folded through the code. Everything else the human reader needs, name, voyage, calls, class, fare, date, is printed around the code where a person can read it, and the ticket carries one page per passenger so that a family of five has five pages or five screens and each one is scanned separately. Supaferry's PDF tickets are built this way, one A4 page per passenger with the code under the QR, and the same code is what the passenger types to retrieve the booking on the website.

What the scanner checks takes under a second and has four parts, in this order: the code exists on this voyage; the booking is confirmed and paid; the passenger embarks at this call; the code has not been used already. Each check that fails produces its own message, because the officer's next action is different for each. A code that is not on the voyage is a passenger at the wrong ship or a forged ticket. A booking that is not paid is a conversation about money. A passenger who embarks at the next call needs to wait. A code already used is the subject of the next section.

A line that wants to detect forged codes offline can have the server sign the ticket code with a short keyed hash that is included in the QR and verified by the desk from a key it holds. It is a cheap addition, and it matters more on lines where tickets are resold on the wharf than on lines where everyone knows everyone.

## Duplicate scans and refusals for non-payment

The system records every scan, including the refused ones, so the second scan of 7K3M9Q-2 shows **already boarded at 14:12, desk B**. There are three honest reasons for a second scan and one dishonest one. The passenger stepped ashore for a cigarette and came back. The officer pressed the button twice. Two desks both scanned the same group when it split across the two queues. Or a screenshot of the ticket was sent to a friend who is now standing at the gangway while the original holder is already on board.

The procedure is the same in every case. The second holder waits to one side. The officer finds the name in the manifest and asks for an identity document. If the first scan was this same person, the record is marked as re-boarded and the count does not change. If it was not, the second holder is refused, and the incident is logged with the time and the officer's name, because the operator will want to know whether the same booking reference shows up again next week. What the officer never does is let both through to keep the queue moving. Two people on one seat is exactly the error the whole process exists to prevent.

**Not paid** is the other common refusal. An agency created a hold for a family, the family went to pay at the finance office and did not reach it before the hold expired, and the ticket in their hand is for a booking that was released. The scanner shows the reason, the number of passengers in the booking and the amount due. The desk has two options. If it is authorised to take payment and issue a receipt, it records the payment and boards the family. If it is not, the family goes to the counter or stays ashore. Either way the override is not silent: it needs a reason and the officer's name, and it appears on the voyage report that the finance manager reads the next morning. The list of refusal reasons should be short and fixed. Supaferry's boarding desk uses five: unknown code, not paid, boards at another call, already boarded, cancelled. Any system that shows the officer a generic error instead is adding a minute to every exception.

## Boarding by call on a multi-stop voyage

On a crossing between two ports there is one boarding list. On a voyage with five calls there are five, and the boarding desk has to be tied to a **call**, not to the voyage. At the second island, the list is the passengers whose origin is the second island. A ticket from the third island to the capital scanned at the second shows **boards at the next call**, and the passenger, who may have taken the island boat out to the ship early, waits aboard the tender or goes back ashore.

Through passengers, those who stay aboard between calls, are not re-scanned. If they go ashore while the ship works cargo, a line can issue shore passes or record a disembark and re-embark scan; the simpler rule, used on most Pacific and Aeolian services, is that anyone going ashore takes their ticket and is scanned again on return, so that the count after the call is still right.

The number on board after each call is on board before, plus embarked, minus disembarked. The embarked figure comes from the scans. The disembarked figure can come from a disembark scan, which is worth doing at islands where people leave with a lot of cargo and the tender makes several trips, or from the officer checking off the list of passengers due to leave here. The multi-stop voyage model that makes this possible is the subject of an earlier article; the point for boarding is that the desk only ever shows the people who should be walking up the gangway at this island, and refuses everyone else with the reason.

## The expected list and manual boarding

Behind the scanner there is always a list: everyone with a confirmed booking who boards at this call, sorted by surname, searchable by name, by reference, by phone number and by agency. The list has three uses. It finds the passenger whose phone has died. It shows who has not arrived yet, so that at fifteen minutes before departure the officer can call the missing names across the wharf. And it produces the no-show list, by name, so that the agency can refund or rebook the next morning.

**Manual boarding** is the procedure for a passenger without a readable code: phone dead, printout left at home, code unreadable because the page went through the wash. The officer finds the name, verifies identity against the name and the date of birth or the document number recorded at booking, and marks the passenger as boarded with the reason and their own name. A manual boarding counts exactly like a scan. It is flagged separately in the report, and the share of manual boardings is one of the few numbers an operator should watch every week. Under five percent is normal. Above it, something upstream is broken: confirmation e-mails are not arriving, the agency has stopped printing, the SMS gateway has run out of credit.

Before the desk opens, the expected list is also printed, twice. If every device fails, boarding continues with a pen, and the ticks are entered into the system afterwards, with the real boarding time. The paper list is the last line of defence, not the first. The difference from the paper era is that it is printed from the same data that the scanner uses, so there is only one list.

## The count the master signs

At the moment the gangway lifts, the number that matters is **persons on board**: passengers boarded, by scan or manually, plus crew, plus anyone else, a pilot, an inspector from the authority, a supernumerary. Most operators keep the tally counter as well, in the hand of the mate at the top of the gangway, and that is correct: two independent counts, reconciled before departure. If the desk says 304 and the clicker says 305, the one is found before the lines are let go, not after. It is usually an infant on a lap, counted by the clicker and listed by the system as not occupying a seat, or a crew member who walked up with the passengers.

The departure report that comes out of the desk has a fixed shape: expected 306, boarded 304, no-shows two, by name, refusals with their reasons, manual boardings with their reasons, overrides with the officer's name. The master signs it. One copy stays ashore with the agent or the harbour master, one goes with the ship, and one is transmitted to the authority where the law requires it: to the single window before departure in the European Union, to the maritime administration in the Philippines under the 2026 electronic ticketing rules, by radio at the departure report on most Pacific services. In Supaferry the report is the manifest page for that call, frozen at departure and exported as CSV or PDF, so the figure the master signs and the figure the authority receives are the same row of the same table.

> The count the master signs is not the number of tickets sold. It is the number of people who walked up the gangway, and only the gangway can produce it.

## Boarding without network coverage

A boarding desk at a wharf with fibre is easy. The desk that matters is the one on a ship anchored off a reef, or at a jetty whose only signal is one bar of 2G that disappears when the tender's outboard starts. An **offline boarding desk** has four requirements, and an operator should check each of them before buying anything.

1. **A cached list.** Before the ship leaves coverage, the desk downloads the whole voyage: every call, every booking, every passenger, every payment status, and shows the time of the download on screen. Three hundred passengers is a few hundred kilobytes. The download happens at the last port with signal, and again any time coverage reappears.
2. **Queued scans.** Each scan is validated against the cached list with the same four checks as online, and stored locally with the device time, the call and the desk identifier. The officer sees the same green or red as at a wharf with signal. The queue is written to storage, not held in memory, so it survives the app being closed and the battery dying.
3. **Sync that keeps the real scan time.** When coverage returns, the queue is uploaded and every scan is recorded at the time it happened on the device, not at the time of the upload. The manifest has to say that the passenger boarded at 14:12 at the second island, not at 09:40 two days later at the capital. Device clocks drift, so the desk records its offset against the server when it downloads the list and corrects the queued times by the same offset.
4. **Conflicts resolved by rule, not by hand.** Two desks offline scan the same code: the earlier scan counts as the boarding and the later one is flagged as a duplicate for review. A booking cancelled at the office after the list was cached: the ship cannot know, so the scan is accepted on board and flagged at sync for the office to act on. The rules are written down and the same every time.

Several desks on one call follow from the same design. Each desk has an identifier, A, B, tender one, works from the same cached list, and shows its own count and the voyage total as of the last sync. When the desks are on the same local network, and a phone hotspot without internet is enough, they can exchange queues with each other so that the total is live even with no server in reach. Before departure, the officer in charge opens the count page, sees the totals of every desk, reconciles them with the clicker and signs. Supaferry's boarding desk is built on this model, a web page installed on a phone or tablet that caches the voyage, queues scans and syncs with the scan time, and it is the example used here; the test below applies to any product.

> **Note:** Test offline mode before you need it. Put the tablet in flight mode in the office, scan twenty test tickets, close the app, reopen it, switch the network back on and check that the twenty scans appear on the manifest with the time at which they were scanned and not the time at which they were uploaded. Then scan one of them again on a second tablet, also in flight mode, and check that the sync flags it as a duplicate instead of counting it twice. If a vendor cannot do this in front of you, the desk is not offline-capable, whatever the brochure says.

## Accessibility at the gangway

A boarding desk has three audiences: the officer looking at the screen, the passenger standing in front of it and the person who has to get physically from the wharf to the deck.

- The officer's screen shows the result in two states only: green, with a large tick and the passenger's name, or red, with the reason in one line, both readable from a metre away in full sun. A sound goes with each, one tone for boarded and two for refused, so that the officer can watch the person rather than the screen.
- The passenger hears the result. The officer says the name aloud, Maria Santos, boarded, and the desk has a read-aloud button for passengers who cannot see the screen. A passenger who asked for large print at booking gets a ticket printed in large type, with the same code.
- The physical gangway gets a **priority boarding window** before general boarding, for wheelchair users, elderly passengers, pregnant women and parents with infants, with a member of staff on the gangway or in the tender. Companions board together. The assistance need entered at booking is on the manifest, so the desk knows that a passenger needs the ramp before the passenger arrives, and the ship knows how many people will need help to leave it at the other end.
- The screen is in the officer's language and the ticket in the passenger's. On a line that serves three languages the ticket is printed in the one chosen at purchase, and the code is the same in all of them.

## A worked example: 306 passengers, 90 minutes, two officers

> **Note:** The ship sails at 16:00 from a wharf with intermittent signal. The expected list has 306 passengers, of whom 22 have an assistance flag, 28 are a school group with one teacher, and 9 are a wedding party booked by one agency. Two officers, two tablets, one clicker. At 14:00 the office downloads the voyage to both tablets and prints the expected list twice. At 14:30 the desks open. Desk A takes priority boarding: the 22 flagged passengers and their companions at about thirty seconds each, finished by 14:42. Desk B starts general boarding with the school group, one page per pupil in the teacher's hand, ten seconds each, five minutes. From 14:42 both desks board the general queue at five passengers per minute each, twelve seconds per scan including the walk to the gangway; 49 minutes of two desks is a capacity of 490 against the 256 remaining, so the queue would clear by 15:10 with even arrivals, and in practice arrivals peak around 15:00 and the longest wait is fifteen minutes. Exceptions during the window: 14 manual boardings at ninety seconds each, 21 minutes of desk time absorbed by the spare capacity; three duplicate scans, one family scanned at both desks, one officer's double press, one screenshot refused, three minutes each; two holds not paid, one paid at the desk with a receipt and boarded, one refused. At 15:30 desk A closes to the general queue and becomes the exceptions desk. At 15:45 the expected list shows seven missing; the officer calls the names across the wharf and five appear. At 15:50 the count page reads 304 boarded, 290 by scan and 14 manually, two no-shows by name, one refused. The clicker reads 304. The master signs 304 passengers and 31 crew, 335 persons on board against a certificate limit of 350. At 16:00 the gangway lifts and the report goes out. Each officer spent about 42 minutes actually scanning inside a 90-minute window; the rest was the buffer that absorbed the arrival peak and the exceptions.

The numbers scale in the obvious way. One desk at five per minute boards 300 in an hour if there are no exceptions, which there always are. Two desks board the same 300 in 35 minutes of scanning and leave room for the family with the dead phone. Three desks on a 600-passenger sailing need a count page that adds them up, which is the real reason to insist on several desks working from one list.

Boarding done this way is not faster because the technology is clever. It is faster because the ticket carries the code, the code carries nothing else, the desk shows one list for this call, every refusal has a reason the officer can act on, and the count at the end is a sum rather than a reconstruction. Those are procedure decisions. The software only has to not get in their way.

## Questions people ask

**Do passengers need to print the ticket?**
No. The code on the phone screen scans as well as the printed one, and the code printed in letters under the QR can be typed by the officer if neither scans. Printing remains useful for passengers without a phone and for agencies selling at a counter, and the printed page should carry the same code as the screen.

**What happens when a passenger's phone has died?**
The officer finds the passenger in the expected list for this call by name, reference or phone number, checks an identity document against the details recorded at booking and marks the passenger as boarded manually with the reason. The manual boarding counts like a scan and is flagged in the departure report. If manual boardings exceed about five percent of a sailing, the problem is in ticket delivery, not at the gangway.

**Can two people board with the same QR code?**
No. The second scan of a code shows that it was already used, with the time and the desk. The second holder waits aside while the officer checks the name and an identity document against the manifest. If it is the same person re-boarding, the record is updated; if not, the second holder is refused and the incident is logged. Letting both through is the one thing the procedure forbids.

**How do we board at an anchorage with no signal?**
The desk downloads the voyage before the ship leaves coverage, validates each scan against that cached list and stores the scans locally with the device time. When coverage returns, the scans are uploaded and recorded at the time they happened, not the time of upload. Two desks can work from the same cached list and the count page adds them up before the master signs.

**How many officers does it take to board 300 passengers?**
Two, with two scanning devices and ninety minutes. One desk scans about five passengers a minute when codes are ready, so two desks clear 300 in roughly 35 minutes of scanning, and the rest of the window absorbs the arrival peak, priority boarding, manual boardings and refusals. One desk can do it in an hour only if nothing goes wrong, which never happens.

**What exactly does the master sign?**
The number of persons on board at departure: passengers boarded by scan or manually, plus crew and anyone else aboard. The desk's count is reconciled with an independent tally at the gangway before the lines are let go. The departure report also lists no-shows by name, refusals with reasons and manual boardings, and is kept ashore, on board and, where the law requires it, transmitted to the authority.

**Is it safe to print a QR code on paper?**
Yes, if the code carries only the ticket code. A ticket code is useless to anyone who does not also match the name on the manifest at the gangway. It becomes unsafe when the QR encodes personal data such as name, date of birth or document number, because every printed sheet left behind then exposes them. Keep the personal data in the printed text where a person reads it, and out of the square.
