It's 8:40am on Monday morning. You've opened an enquiry that arrived on Saturday, checked a paper diary for availability and searched through an old email to find the price you quoted for a similar group last year. Before you can reply, the phone rings.
By lunchtime, your prospective customer has booked somewhere else.
It's tempting to see this as a marketing problem. Perhaps you need better website content, more social posts or an AI assistant to write quicker replies. But your marketing worked: the customer found you and wanted to buy. The booking was lost in the journey afterwards.
Welsh tourism businesses currently have a good reason to explore AI. The Welsh Government has announced £2.1m of wider support for Welsh SMEs, including £500,000 aimed at tourism and events. MWT Cymru is also delivering free practical workshops covering content, online visibility and customer communications.
We think you should use that help. We also think you should resist buying or building anything until you know where your booking journey is actually failing.
Set a timer for 15 minutes and answer the following questions. Estimates are fine. An honest guess is more useful than a perfect figure you never finish calculating.
Don't measure the automatic acknowledgement. Measure the time between your customer sending an enquiry and receiving an answer that confirms availability, gives an accurate price and explains what happens next.
Check your last ten enquiries. If the answer is usually several hours, or stretches into the next working day, write that down. AI might help you read and draft a response, but it can't safely confirm a booking unless it has dependable access to your real availability, prices and rules.
Follow one recent booking from beginning to end. Did you copy the customer's name, dates or party size from an email into a diary, spreadsheet, invoice, payment request or delegate list?
Count every transfer. Each one takes time and creates another opportunity for a name, date or dietary requirement to be entered incorrectly. If your information already exists digitally, your systems should usually move it for you.
Can you and everyone who needs it see the same current availability? Could two people promise the same room, instructor, vehicle or date? Does somebody have to check with the one person who knows before confirming anything?
If your availability depends on a paper diary, an isolated spreadsheet or someone's memory, faster AI-written replies won't solve the underlying delay. You need one dependable source that your booking process can trust.
Look at deposits and outstanding balances from your most recent busy month. How many customers had to be reminded manually? How many reminders were sent later than you intended? How much money remained outstanding after its due date?
A clear payment rule rarely needs AI. If a balance is due 28 days before arrival, ordinary automation can request it at the right time, record the payment and alert you when something unusual happens.
Confirmations, joining instructions, accessibility questions, equipment lists, arrival times, review requests and invitations to return can all become vulnerable during a busy week.
Write down anything that happens only because you or a trusted member of your team remembers to do it. Those are not personal failings. They're signs that your process is asking people to behave like software.
You don't need AI for every item you've found. In fact, we'd advise against it.
If an established booking product solves your problem properly, use it. You don't need Pedwar to rebuild software that already fits. Bespoke development becomes worthwhile when you're repeatedly working around the product, maintaining parallel spreadsheets or changing your operation to satisfy the tool.
Imagine you run guided activities. An email arrives asking about a Saturday session for eight people, including one guest with an access requirement.
AI could identify the requested date, group size and access note. Your booking system should check capacity and apply the correct price. Automation could prepare the payment request and confirmation. The access requirement should be placed clearly in front of you, with the rest of the enquiry, so you can respond personally.
That's one journey using four different kinds of help. Asking AI to do all of it would make the process less dependable, not more.
You may finish this audit with a page full of problems. Don't turn it into a transformation programme. Choose the leak that costs you the most in lost bookings, repeated work or delayed payment. Map that workflow as it operates today, improve one handover and measure the result for a month.
This is how we approach the work at Pedwar. We follow your journey from the first enquiry through availability, payment, communications and follow-up, then identify where information, time or revenue is being lost. The right answer might be a configuration change, an integration, focused automation or a bespoke system. The technology follows your problem, not the other way around.
We've applied that thinking to the booking platform we built for Jones International, where passenger details, rooms, payments, reminders and customer correspondence must work together across thousands of bookings.
If you've completed the audit and found a leak you can't sensibly fix, our Tourism & Events Automation Review gives you a fixed-scope assessment, written priorities and a practical first step. If your existing setup is already doing its job, we'll tell you that too.