The State of Tourism Digital Marketing 2026: Three Years On, Still Only One in Ten Can See Where Their Bookings Come From

Every conversation we have had with a tourism operator this year has started with AI.

AI search. AI agents booking trips. AI writing the descriptions. Whether ChatGPT is sending anyone, whether Google's AI answers are taking the traffic, whether the booking system's new AI assistant is worth turning on. It is a fair thing to be asking about, and we get asked it weekly.

So we went back to the same set of tourism businesses we first audited for our 2024 study and audited them again, to find out what had changed underneath all of that.

The answer is nothing that matters.

In our 2024 study, around one in ten tourism operators could trace a booking back to what caused it. In 2026, it is still around one in ten. The industry has not moved in three years on the single most important measurement it has.

We will come back to AI at the end, because the two things are connected, and the connection is about where you put your attention.

This is the full 2026 write-up, and the study has now run three years running. It follows our 2024 research into tourism website analytics and conversion tracking and the 2025 edition, covering 538 tourism businesses across 44 countries. Both are still published in full. If you run an accommodation, activity or tour business, the numbers below describe your competitors, and there is a decent chance they describe you.

How we did the research

This study has run for three years, and how we ran it has changed. We should say so up front, because the method decides what the numbers are allowed to claim.

In 2024 we did it entirely by hand. More than 250 hours, one website at a time, opening the developer tools on every site and checking the tags, the pixels, the tag manager, and whether anything at all fired when you clicked a phone number or moved through to a booking engine. No scraper doing a shallow pass, and no survey asking operators what they thought they had installed.

In 2025 we did it by hand again, trimmed down, going back over the same businesses to see what a year had changed.

In 2026 we automated the process we had already been running by hand. The checks are the same, the definitions are the same, and the questions asked of each site are identical to the two previous years. What changed is that a browser now performs the clicks instead of a person. That is the only difference, and it is what let us cover the whole list properly instead of a sample.

The automation was not trusted on its own. Every category of result was checked against a manual sample. Edge cases were opened by hand and worked through by a person. Anything we were not fully confident about was verified manually before it counted, and where a site fought us we went and looked at it ourselves. Sites that had gone offline, sites that blocked automated access, and sites where the booking step could not be found were held out, not counted as failures. A site we cannot read tells us nothing, and recording it as "no tracking" would have quietly inflated every number in this report.

A note on precision. There is an error margin in any study run this way, and we would rather say so than pretend otherwise. What matters here is not whether a figure is 9.8% or 11%. It is that the figure has sat at roughly one in ten for three consecutive years, measured by different methods, by different people. The scale of these numbers, and their refusal to move, is the finding. Treat the decimal places as indicative and the direction as solid.

A note on the businesses. Give or take a handful of changes, the list of operators we looked at in 2024, 2025 and 2026 is the same list. That is deliberate, and it is what makes a year-on-year comparison honest. We are not comparing three different samples and calling the difference change. We are watching the same businesses over three years.

It is a cross-section of tourism and hospitality rather than one niche. Accommodation of every size sits alongside activity providers and experiences: whale watching, cooking classes, scenic tours, helicopter flights, guided walks, wineries, adventure operators. The weighting is towards tourism.

The one thing every business in this study has in common is that they all take bookings through a third-party booking system rather than handling the transaction themselves on their own website. That is the shape of the industry, and it is exactly the shape that creates the measurement problem in Finding 1.

We find these operators by fingerprinting booking-system URLs, which is how you locate real trading tourism businesses at scale when no list of them exists. That method decides who ends up in the study. So when you see a platform named below, it is there because its customers were findable, not because it is bigger or worse than one that is absent. Nothing here is a market-share figure, and nothing here ranks one platform against another.

What we found, in short

Cohort of 545 businesses, of which 526 were still trading and 489 could be read in full. Audited August 2026 against the baseline we recorded over late 2023 and published as our 2024 study.

Finding 1: one in ten can tell you where a booking came from

This is the most important number in the study, and it is the one that has not moved.

Most tourism operators do not take the booking on their own website. You click Book Now and you land on a different company's system. A browser treats those as two unrelated visits, so the moment somebody moves from yoursite.com to bookingsystem.com, the journey starts again from scratch. Nothing carries across unless somebody sets it up. That setting is called cross-domain tracking.

Of the 417 operators where we could test the handover, 41 had it working. 9.8%. In 2024 the figure was around 10%.

Now the part that gets missed, and it is the reason this matters more than any other line in this report.

Your booking system already tells you whether a booking came through an OTA or came direct. That is useful, and most operators stop there, because it feels like the answer.

It is one split, on the last step, and it is the smallest version of the question.

What working cross-domain tracking tells you is which sites sent you the booking. Which of your regional tourism organisation's listings produced revenue and which produced nothing. What your destination management organisation's website is genuinely worth to you, in money, this season. Whether the booking showing as "direct" was really direct, or was somebody who found you through search three weeks ago and typed your name in later. Which pages, which campaigns, which partners.

Your booking system cannot tell you any of that, and it never will, because by the time the booking reaches it, everything that happened before has been thrown away, if the system has not been configured correctly.

That is the difference between knowing you took 400 bookings and knowing what to do next year. Nine in ten operators do not have it.

There is more on the practical side of this in our guide to tourism analytics and conversion tracking for operators, and setting up cross-domain tracking between a website and a booking engine is the piece of work this finding is really about.

Finding 2: having analytics installed is not the same as measuring anything

Two thirds of the operators we audited have Google Analytics 4 installed. 325 of 489.

A third have nothing at all. Google Analytics 4 is free. It takes minutes to install. In 2026 a third of trading tourism businesses have no visibility of their own website whatsoever. Not poor visibility. None. If that is your business, you have no way of knowing how many people visited your site last month, which pages they read, or whether the money you spent on advertising brought anybody at all.

The more revealing number is what the other two thirds are doing with it.

Of the 325 operators with analytics installed, 253 are not recording how many phone clicks or email clicks they received. The booking user journey is broken on 77.8% of them, because there is no cross-domain tracking between their website and the booking system.

In plain terms: the tag is on the page and it counts visits. It will tell you how many people came, and roughly where from. It records nothing at all about the three things a tourism business needs to know, which are whether somebody rang you, whether somebody emailed you, and whether somebody went on to book.

Your analytics can tell you that 4,000 people visited last month. If it is not configured, it cannot tell you the booking path any of them took. If the clicks are not tracked properly, it cannot tell you how many people tried to contact you, or which channel they came from.

So whilst over two thirds of operators have Google Analytics installed, the vast majority of them do not have it configured to record the data that matters. That is the data that drives your conversions and your optimisations. What is running is the basic measurement. It is the extra configuration that gives you the useful part, the part you can act on and get real insight from.

The problem is not that operators refuse to measure. Most have gone and installed the thing. The measurement stops at the point where it would start being useful, and nobody has told them.

Finding 3: the contact details are clickable now, and almost nobody records the clicks

There is one genuine improvement in this study, and it deserves naming.

Tourism websites have got much better at basic usability. Two thirds now have a phone number you can tap, and three quarters have a clickable email address. In 2024 that was a minority. On a phone, this is the difference between a customer calling you and a customer giving up.

Then the measurement falls over at exactly the same point as everything else.

We would not claim the phone call is the booking for most operators. That overstates it. For a great many tourism businesses it is a substantial part of the enquiry flow though, particularly for anything involving a group, a custom itinerary or a high price, and it is invisible in almost every account we look at.

The value of recording those clicks goes well beyond the count. Here is a real example from our own client work.

We set up phone click tracking for an operator and found something we were not looking for. A large volume of calls was being made by people who had already clicked through into the booking platform. Not from the operator's own website. From the booking step itself, part way through paying. That volume dwarfed the calls coming from their own site.

We had suspected the booking process was confusing. That data proved it, and it pointed at exactly where. People were reaching the point of paying, getting stuck, and picking up the phone. Without click tracking, that is invisible, and it looks like nothing more than a slightly disappointing conversion rate.

The only reason that insight existed is how the tracking was built. We did not record a generic "somebody clicked a phone number" event, which is what most setups do and which would have shown a single meaningless total. We recorded phone clicks on the operator's own website separately from phone clicks inside the booking platform, so the two could be compared. We also split them by landline and freephone number, which we did out of interest rather than necessity.

That level of detail is a decision somebody makes at setup, before there is any data to look at. Get it right and a question you had not thought to ask gets answered a year later. Get it wrong and you have a number that goes up and down and tells you nothing.

That is what measurement is for. Not a number on a dashboard. A specific, fixable problem that was costing real bookings.

And is the phone number itself even right?

One limit on what we measured here. We checked whether the number was clickable and whether the click was recorded. We did not check, across 545 sites, whether the number was written in a format that actually works.

That is its own problem, and we run into it constantly. A number can be perfectly clickable and still fail when somebody taps it, because the format written into the page does not dial. The country code is missing, or there are spaces and brackets in it that the phone cannot interpret, or it is written for a domestic caller in a way that breaks from anywhere else.

This is not only an overseas problem, which is what most people assume. It goes wrong domestically too, for the same reason.

We ran into it sitting in Galway, trying to phone a five-star hotel in the same town. Tapped the number on their own website, and it did not dial. Number not recognised.

The number was written in a format that worked for neither domestic nor international dialling. We happened to be on an international mobile at the time, but a domestic one would have hit exactly the same wall. That was a business with rooms to sell, a website that looked the part, and a phone number on it that did not work when a customer used it the way it was presented.

Somebody in that position has to work out the correct number themselves and dial it manually, at the exact moment they were ready to talk to you. Most will not bother.

The safe answer is to write every number in full international format. It costs a domestic caller nothing to dial with the country code included, so there is no downside. Somebody on a desktop can read it and know exactly what to type. Somebody on a phone taps it and gets through, whether they are in the next town or the next hemisphere.

Finding 4: Meta measurement went backwards, and a pixel on its own does very little

The Meta pixel is installed on 30.5% of the operators we audited, down from 34.9% in 2024.

The pixel, which Meta now calls a dataset, is what lets you see whether the money you spend on Facebook and Instagram produces anything. Fewer operators have it than three years ago. Some of that will be sites rebuilt without carrying the tracking across, which is a theme running through this whole report.

Installing it is only the first half of the job. The pixel on its own records that somebody visited your website. To get anything useful out of it, you have to configure it to send the events that matter: phone clicks, email clicks and bookings. Put the pixel on and configure nothing, and all you have done is watch people browse.

There is a second consequence that almost nobody realises, and it works against you.

If you have a pixel collecting data on your website and you are not advertising on Meta yourself, you are still teaching Meta about your audience. Meta now knows that a particular set of people are looking at accommodation like yours, in your area, at this time of year. If your competitor down the road is running Meta ads and you are not, that knowledge is available to their campaigns and not to yours. You are supplying the training data and they are spending the budget.

We saw this from the customer's side recently. Looking for an overseas SIM card, we spent time on one provider's website. In the days afterwards we were advertised to by a string of their competitors, and never once by the company whose site we had actually been on. They had the pixel. They were not using it. Everybody else in that market was targeting the audience properly, and that provider had effectively handed them the introduction.

More on getting this right in our guide to running travel Facebook ads properly.

Finding 5: 96.9% of tourism websites fail Google's own speed standard

We measure page speed with Google's own tool, the same way each year, so this is the least interpretive number in the study.

The median tourism website takes 9.8 seconds to display its main content on a mobile connection. Google's threshold for "good" is 2.5 seconds.

One in five tourism websites takes more than twenty seconds to show its main content. The slowest site in the study took over two minutes. Thirteen sites out of 455 scored 90 or above on Google's overall performance rating, which is 2.8% of the industry meeting a standard Google has been publishing for years.

These are not obscure businesses. The slowest sites in our data include hotels in capital cities and established operators with real advertising budgets.

You will probably not recognise these numbers from using your own site, and that is the trap. You visit your website on good office wifi, from a device that has already cached half of it, and it feels fine. This is Google's measurement, taken the way Google takes it, on a mobile connection, from a first-time visitor's position. That is the version your customers get.

A slow site costs you in more places than most people expect:

• Search rankings. Speed is a ranking factor, and it has been for years.
• The cost of your advertising. Google rates the experience people get after they click. A poor one raises what you pay for the same position, which is one of the first things we look at when we take on Google Ads for a tourism operator.
• Your conversion rate. People leave. The longer the wait, the more of them leave.
• Your average order value. Slow booking journeys push people towards the simplest, cheapest option rather than the one with the extras on it.
• The bookings you never hear about, from the visitors who closed the tab before your page appeared.

Every one of those is a cost you are already paying, and none of them shows up as a line item anywhere.

Put performance in the requirements. One of us spent a previous career as a software developer in banking, where no project specification got approved without a performance line in it. Website projects almost never have one. We have never seen website performance written into a set of requirements unless we put it there ourselves for a client.

Without it, the temptation is to keep adding bells and whistles, and every one of them costs you speed. You end up with something that costs you money for years afterwards.

Finding 6: one in five changed booking platform, and the tracking rarely survives the move

Of the operators where we could identify the platform in both years, 21% are on a different booking system than they were in 2024. Movements we saw repeatedly included Rezdy to FareHarbor, BookDirect to SiteMinder, and Preno to BookDirect. Again, that is what our sample shows, not a statement about which platform is winning.

Changing booking system is a reasonable business decision, and operators change for good reasons: price, features, support, a system that finally handles groups properly.

What almost nobody accounts for is that the tracking does not come with it.

We have seen this at close range. A client pays to have conversion tracking set up properly. Every booking attributed, every channel measured, the whole thing working end to end. They then change booking system, for perfectly sound reasons, and nobody in that conversation mentions the measurement. The cross-domain configuration pointed at the old platform. The conversion actions were built around the old platform's confirmation page. On the day the new system goes live, all of it stops.

They have paid for that setup once. They now need to pay for it again, and in the meantime they have lost the year-on-year comparison they were paying for. The history does not come back.

So put it in the requirements. When you are choosing a new booking system, the tracking questions belong on the list alongside the operational ones. Operational ability will usually win, and it should, because a booking system that does not run your business properly is no use however well it measures. But the questions still have to be asked, and asked at the point where you can still walk away, which is before you sign.

The specific questions are at the end of this article. Ask them of every provider you shortlist. Enough operators asking them is the only thing that will make this industry's booking platforms take conversion tracking seriously, and a great many of them currently fall short.

A note on FareHarbor, and this one is opinion

FareHarbor came up repeatedly in the moves we saw. They pushed hard during the pandemic as the lower-cost option, on a model where the operator pays little or nothing up front and the booking fee is charged to the traveller instead.

We do not know the detail of individual contracts, and this is a comment on a pricing model, not on how well any platform supports tracking. We still find it interesting that so many operators are moving to FareHarbor, because there is nothing wrong with the platform, but almost every operator we work with is trying to move bookings off the OTAs and onto their own website. The OTA's commission is invisible to the customer, and they book. Book direct, and the customer watches a fee appear at the checkout.

For the operator it makes commercial sense, because the cost comes off their own books. As marketers we think it works against the message. You are asking somebody to book direct and then charging them for doing it, at the moment they are weighing your price against the OTA's.

Finding 7: one in thirteen has vanished from the address they used in 2024

42 of the 545 businesses, 7.7%, no longer respond at the web address they were using three years ago.

Some have closed, which is what three years in this industry does.

Others have changed name or rebranded and did not carry the old address across. A name change done properly keeps the old address alive and forwards everybody to the new one. Not doing that throws away every inbound link, every bookmark, every listing on a partner site and every year of search history the old address had earned. From the outside, a business that rebranded badly and a business that closed look identical.

Finding 8: the things we found while looking for something else

None of the following was what we set out to measure. They turned up while we were checking other things.

When we run a digital review or a digital audit for a client, we never do it in one sitting. That is deliberate. The obvious problems surface on the first pass. The expensive ones surface on the second and third, usually while you are looking for something else entirely. An audit done in an afternoon finds the things that were easy to find.

An Australian motel with somebody's browsing session pasted into its Book Now button. Four booking links on the homepage, all carrying an analytics parameter written straight into the page source. Decoded, it was one visitor's session from November, with a counter showing they had been on the site 141 times. Somebody opened the booking system, copied the address out of the browser bar, and pasted it in as the link. That parameter has to be generated fresh at the moment of the click. An expired one is discarded, so every genuine customer arrives at the booking system as a stranger. The site looks like it has tracking. It has a decorative copy of tracking.

Two operators who pasted a Facebook click identifier into their own website. One links to its own Instagram page, another to press coverage about itself, both carrying the tracking code Facebook appends when you click through to something. Harmless in those two cases. What it shows is that links are going onto these websites without anybody knowing what is inside them, and at any scale that is how conversion tracking quietly breaks.

Eleven operators whose websites throw a security warning. Their certificate has expired, or does not cover the address people actually type. Visitors get a browser warning telling them the connection is not private. Most close the tab, and it reads as a business you should not trust, not the small technical fault it is.

Five whose site only works if you type the address one particular way. With the www. but not without it, or the reverse. The same goes for http against https, which matters because older external links pointing at you are often still written as http and will not redirect properly. To anybody who hits it, this looks like a broken website. It is usually a broken redirect, and normally a fairly quick fix. This is a fundamental redirect problem, and it is the same class of fault that destroys search rankings during a website migration or a change of address. It also means you have no control over how your own web address behaves when somebody else links to it, prints it, or types it from memory. The owner never notices, because they always arrive the same way.

None of these is exotic. Each is a half-hour job for somebody who knows where to look, and each is currently costing more than that.

What we cannot tell you, and will not pretend to

Reading this, the natural question is whose fault it is. We cannot tell you, and neither can anybody else looking at these sites from outside.

Everything in this report was gathered without a single login. No access to anybody's analytics account, no access to anybody's website, no permission asked for and none needed. It is all publicly available information, sitting in the open on every website, for anyone who knows what they are looking at. Knowing what to look at is the part that took fifteen years of doing conversion tracking and digital marketing in tourism, hospitality and leisure.

What that does not give us is intent. Tracking gets set up by one of four people: the booking vendor, the owner, an agency, or an agency that was not very good at it. From the outside all four produce a website that looks the same. We can see whether the measurement works. We can never see who was responsible.

That rules out the comparison everybody wants, which is a table of which booking platforms track well. A good platform does not give you good tracking. Some of the best-built booking systems in this study sit behind sites measuring nothing, and some of the smaller ones are set up perfectly. Your vendor is selling you a booking system, and analytics is usually not what they are being paid for, so some will help you and some will tell you it falls outside their scope. Which one you get often comes down to how your own website was built, which is nothing to do with them either.

So we are not publishing the table of which booking platforms support what level of conversion tracking and digital marketing. That would be a fascinating study to run, and it is not the point of this article. The data we have cannot support it either, and getting it wrong would cost somebody business.

What we can say is where it usually comes from. Most often it is an operator who did not know that part existed, and so never asked the provider about it. Or it is a straightforward oversight, or a gap in experience, on the part of whoever built the site or ran the marketing. You have to be working with somebody who knows this side of it inside out, and that is rarer than it should be. We have watched big agencies miss it and small agencies miss it. We have watched world-class tourism operators miss it and one-person operations miss it, and every combination of the two.

Why this should matter to you

You cannot grow what you cannot see, and the specific thing most tourism operators cannot see is which of their marketing produces money.

That has three consequences, and we watch all three happen every season.

Budget goes to whatever is easiest to measure, which is usually the channel that reports on itself most loudly, rather than the one producing the most revenue.

Genuinely good marketing gets cut, because it cannot prove what it did. The work that produces a booking six weeks later, the partner listing, the content that gets found in search, all of it shows up as nothing when the measurement stops at the booking system's front door.

And you cannot make the case for more. Whether you are asking a board, a bank or a business partner, the argument for spending more on marketing is evidence that the last lot produced revenue. Without measurement, that argument does not exist. We have written separately about how to justify your marketing budget with real numbers and about the direct bookings versus OTA decision for tourism operators, which is the choice this data affects most directly.

Owners and internal marketing teams should both be paying attention here, because the quality bar across this industry has to come up, and every gap in this report is costing somebody real money right now. It is also why tourism marketing is the sector we have chosen to specialise in.

And it puts a question to the booking platforms. If your booking system does not support this level of tracking, or supports it but never makes that visible to you when you are being set up, that tells you where their interest ends. They are focused on taking the booking. Your marketing is somebody else's problem. For us the two things go together, because there is no value in a booking system you cannot measure.

The questions to ask your booking system

Ask these at implementation, or when you are shortlisting a new provider, while you still have a choice. We ask them of every new client's existing platform in our first engagement, and the answers tell you a great deal very quickly.

1. Do you have documentation on setting up analytics and conversion tracking with your platform? If there is no documentation, that is usually the whole answer. It is one of the clearest signals we get.
2. What level of conversion tracking do you support? Is it only a completed purchase, or can it record earlier steps: the search, the item added, the different stages of the booking? A platform that only reports completed bookings cannot tell you where people are dropping out.
3. Do you feed the same events to the Meta pixel? And to any other advertising platform we use?
4. How do you recommend this is implemented on our website? Specifically, do you support cross-domain tracking, and what code do we need on our site for it to work?
5. If you are recommending an embedded booking form, how does the tracking survive it? An embedded form on your own page is usually a warning sign for conversion tracking, because of how browsers handle cookies inside an embedded window. Ask them directly whether it is supported and ask to see the documentation that says what you need to add to your website to pass the visitor's session into it. We have had clients add an embedded booking form without mentioning it to us, and watched tracking that had been working perfectly stop dead.

Where AI actually fits

We started with AI because that is what everybody is asking about. Here is the honest answer, and it is about priorities rather than about the technology.

Three years of the same result tells you something a single year could not. This is not a lag, and it is not operators being slow to adopt something new. Cross-domain tracking, conversion tracking and clickable contact details were not new in 2024 and they are not new now. They have been free, available and documented for well over a decade. Two thirds of this industry has gone as far as installing the analytics and then stopped before measuring anything with it.

So if your attention is currently on AI while the fundamentals are missing, the order is wrong. Fix the measurement first. We have written separately on tourism SEO and getting found in AI search for when you are ready for it. That is a sequencing argument, not a reason to avoid AI, and the sequencing matters because of what happens when you skip it.

Point AI at your data when your data is wrong, and it will not tell you the data is wrong. That is not what it does. It will take the numbers you give it, find the pattern in them, and present you with a confident, well-reasoned answer built on a false picture. It is very good at giving you something that sounds right.

None of which makes this an anti-AI article. We use AI every day, on every piece of work we do. Our job is heavily data-driven and we get insight out of it we would not otherwise reach. We also watch it go wrong several times a day, at exactly the points where it needs somebody with subject-matter expertise looking over its shoulder. We know its benefits, and we understand its weaknesses.

The argument is about order. AI citations, AI references and the content that earns them deserve a high place on your list. Being able to measure your marketing comes before them.

If your reporting says paid search produced 12 bookings when it actually produced 60, because two thirds of the journey was discarded at the booking handover, then every recommendation that follows will be built on that. The recommendation will be clear, quick and wrong, and there will be nothing in the output to warn you.

The fundamentals in this report are what everything else depends on. Right now, for nine operators in ten, they are not there.

Where to start

1. Check the booking handover. Click through your own booking journey and see whether the tracking follows you. This is the one that matters most, and nine in ten get it wrong.
2. Record your phone and email clicks. Most operators have now made them clickable. Almost nobody counts them.
3. Check the phone number itself works from overseas. Full international format, every time.
4. Test your page speed on a phone, on mobile data, not on the office wifi. Over four seconds to show anything is a bigger problem than your ad copy.
5. If you are changing booking system, budget for the tracking rebuild inside the same project, and ask the five questions above before you sign.
6. Check your own web address works both ways, with and without the www., over a secure connection.

None of this is glamorous and none of it is AI. It is the layer everything else sits on, and for most of this industry it is not there.

Get your own setup checked

If any of this sounds like your business, the quickest way to know for certain is to have somebody look at it properly.

The Comprehensive Digital Review and Action Plan is exactly this study, run on your business alone. We go through your analytics, your tracking, the handover to your booking system, your website, your search visibility, your ads and your listings, and you get a written review of 25 to 50 pages with a prioritised plan, plus two hours to talk it through. It is how you find out which of the numbers above is you.

If you already know your tracking is the problem and you want it fixed rather than diagnosed, that is the Digital Foundation Setup, and $1,000 of the review comes off it if you have had one.

And if you would rather just talk first, the Growth Plan Session is a paid advisory call. We work through your business, your situation and your market, then tell you straight where the growth is and how we would go and get it.

This is the third year of our research into tourism digital marketing. We audited 545 operators worldwide in August 2026 against the baseline we recorded over late 2023 and published as our 2024 study, using only publicly available information. No business is named. We also run a full digital review for tourism operators for anyone who wants the whole picture on their own business, and single-channel audits for anyone who only needs one platform looked at.