Skip to content

What can website managers learn from the Spirit of Tasmania website crash?

    The Spirit of Tasmania website crashed this week (September 2026) as 150,000 visits hit the site in just 30 minutes. As we start the run towards Black Friday, Cyber Monday, Boxing Day and the New Year’s Day sales, it’s a timely reminder to be ready for peak demand.

    So, what are the lessons?

    Normally, the Spirit website gets about 2,000 visits a day, so that’s statistically it experienced a massive increase (although if you read to the end, maybe not as large as you might think).

    But was it really a surprise?

    The rush came as additional capacity on the new Spirit of Tasmania ferries went on sale at 10am. There had already been considerable publicity. Travellers had been waiting for the extra sailings to become available, particularly those wanting to take cars and caravans across Bass Strait.

    In other words, Spirit of Tasmania effectively told a very large number of people to turn up at its website at 10am.

    And guess what, like fans trying to get AFL finals tickets, that’s exactly what they did.

    According to the ABC, customers encountered error messages and problems completing bookings. Some received a 502 Bad Gateway message.

    TT-Line said it had anticipated strong demand and increased its website capacity ahead of the release.

    It just didn’t increase it enough.

    Surely you can just add more servers?

    This is the obvious question, and the answer is partly yes.

    Cloud hosting has made it relatively easy to add substantial computing capacity when you know a traffic spike is coming.

    But there’s an important difference between a busy website and a busy booking system in terms of complexity, which affects load.

    If 150,000 people suddenly want to read an article on a website, that’s principally a traffic problem.

    If they want to make 150,000 bookings, it’s a different proposition.

    A customer searching for a Spirit of Tasmania fare could be accessing several systems. What sailings are available? Is there room for the car? Is a cabin available? What does it cost? Has somebody else just taken the last available space? Can the payment be processed?

    In this era of super smart integrations, somewhere behind the website is a booking and inventory system that has to keep track of all this.

    Adding more web servers doesn’t necessarily make that system any faster, so simply throwing more server capacity at the problem isn’t necessarily the answer.

    But that doesn’t mean there wasn’t one.

    This is how we would approach it

    If Boylen was involved in a major online release such as this, we would start by assuming things could get ugly.

    Then we’d test for it.

    Load testing allows you to simulate large numbers of people using a website before the real customers arrive… and we’d test well beyond the expected traffic.

    As we keep increasing the load , our techs can see what happens. Does the website slow down? Does the database struggle? Does the booking engine stop responding? Does a third-party service become the problem?

    Trust the data – it will show you the weak point.

    Then fix it, or find a way to protect it.

    Waiting Room Strategy

    For a release such as this, that could include something we’ve all encountered when buying tickets online: a virtual waiting room.

    Instead of allowing everyone into the booking system at 10am, customers are progressively admitted.

    The rest might see:

    “You are in the queue. There are 8,426 customers ahead of you. Your estimated waiting time is nine minutes.”

    People don’t like it – but they do understand it.

    That’s a much better customer experience than staring at an error message, refreshing the browser and wondering whether the booking they wanted is disappearing while they’re locked out.

    It also protects the booking system from being overwhelmed.

    The numbers aren’t quite what they first appear

    The 150,000 figure makes a great headline, particularly when it’s compared with 2,000 visits on a normal day.

    Look at it another way, though.

    There are 1,440 minutes in a day. At 2,000 visits a day, the Spirit of Tasmania website normally averages about 1.4 visits a minute.

    On Wednesday morning it received about 5,000 a minute.

    That’s a phenomenal increase on its normal traffic.

    But 5,000 website visits a minute averages out at about 83 a second and, sorry Spirit team, that’s not a frightening number for modern web infrastructure.

    The bigger issue is that these weren’t people casually browsing the website. They were hitting a booking system, checking the same sailings, looking for limited inventory and trying to buy it.

    We think you can see now that this is the tricky bit.

    It’s also why testing the website alone isn’t enough. You need to know what happens to the entire booking process under extreme load.

    Test for the rush, not the average day

    Most businesses don’t need to build a website capable of coping with 150,000 people arriving in half an hour.

    But plenty of businesses create their own, smaller version of exactly the same problem.

    They send 100,000 emails announcing a sale. Registrations for a popular event open at midday. Tickets are released at a nominated time. A new product is launched after weeks of promotion.

    Then everybody clicks at once. BOOM!

    That’s when discovering the limits of your website gets expensive.

    The Spirit of Tasmania situation is unusual because the company knew exactly when the rush would start. It knew people were waiting. It expected heavy demand and says it had already increased capacity.

    Yet customers still couldn’t book.

    Sometimes extraordinary demand really does come out of nowhere.

    This one didn’t.

    Phishing awareness notice: we're aware of phishing emails impersonating Boylen staff.

    Close Welcome Bar