Take Back Your Bookings: Driving Direct Sales with Smarter Digital Marketing

Tour operators ask about OTAs and direct bookings more than any other subject, and most keep paying the commission anyway.

An OTA takes 15 to 30 percent of every ticket, and it takes the customer too. After the tour it markets to your guest, and it hides their real email address behind a temporary fake one, so you cannot follow up. The alternative is advertising to people who already know you, using the guest records sitting in your booking system and the traffic already coming to your website.

Carla Sage, Nick Kogos and John Marshall from Datafy join Peter Syme to work through how that data becomes an audience on Meta and Google, how the two platforms differ, and how much of a budget belongs behind each kind of audience. They also spend time on the number that decides whether any of it is worth doing: what a customer costs to acquire, measured against what that customer is worth across every booking they ever make.

Resources:

• Datafy: mobile location data, nationwide credit card data, audience building and campaign management under one provider
• John Marshall, Datafy: john@datafy.com, the contact for operators with questions
• Meta Ads Manager (Facebook and Instagram): imported customer audiences, website and app audiences, and lookalike audiences with a 1 to 10 percent similarity slider
• Meta pixel: the site tag that builds website visitor and cart abandoner audiences
• Google Ads: the second platform covered throughout
• Google Audience Manager: where imported customer lists are built in Google, including automated connections from many booking and POS systems
• Google Analytics: connects directly to Google Ads for remarketing audiences
• Google Tag Manager: the faster route to URL-based remarketing rules, described as easier to set up than Meta’s equivalent
• Google Demand Gen campaigns: the only Google campaign type currently running lookalike audiences, covering YouTube, Gmail and Discover placements
• Google Things to Do: the booking ads that sit above Google search results, available only to businesses with a booking engine attached
• YouTube: channel and video interaction audiences, available through Google because Google owns YouTube
• Apple Pay and Google Pay: named as a checkout requirement, not an optional extra
• Gemini: suggested for reading a Google Ads report and answering acquisition cost questions from it
• Your local DMO or CVB: free source of top domestic origin markets and top source countries for your destination
• Viator, GetYourGuide and Expedia: the OTA channels operators are trying to reduce their dependence on

Key Takeaways

• The commission is the visible cost of an OTA, and the guest relationship is the expensive one. [00:02:24 to 00:03:58] Carla Sage lists what comes with the 15 to 30 percent: the OTA emails your guest after the tour to sell them other local activities, holds your money for weeks past the tour date, and substitutes a temporary fake email address for the real one so you cannot follow up. Listings also sit beside cheaper competitors, and algorithm changes can drop your visibility with no warning and no recourse.
• Moving a quarter of your bookings from OTA to your own site changes the business, not just the margin. [00:03:58 to 00:04:43] Carla Sage puts the direct booking margin at roughly 97 percent against the heavily cut OTA equivalent. The compounding part is the contact record: a direct booking gives you the email address and phone number, which means you can invite that guest back next season or sell them an event.
• Use location data to target three specific zones, not a radius around your town. [00:04:43 to 00:06:14] Carla Sage names tourist hubs first: resorts, airport terminals and cruise ship docks, where anonymized mobile signals show devices gathering in real time or having gathered previously. Then points of interest, meaning dining and entertainment districts where visitors spend time. Then competitor zones, which lets you advertise to people standing at a competing attraction or departure point right now.
• Your booking system is a targeting database, and the export is the first job. [00:06:14 to 00:07:00] Pull guest emails, phone numbers and ZIP codes out of the reservation software, then segment rather than uploading one undifferentiated list. Carla Sage names the segments that pay: top spenders, repeat guests, large family groups, and people who bought the premium experience.
• Re-contact past guests eight or nine months after their trip, not at the anniversary. [00:07:00 to 00:07:47] Carla Sage times this to when the guest starts planning next year’s vacation rather than when they took last year’s. The list you already own is the cheapest audience in the session.
• Site tags separate the browser from the shopper, and the shopper is where the money goes. [00:07:00 to 00:08:32] Carla Sage describes tracking specific page views on individual tours and pricing pages, capturing visitors who selected dates and started a booking without finishing, and measuring engagement depth. Somebody who spent several minutes comparing prices and schedules is a different target from a casual browser and should not receive the same ad.
• Match the message to the data source that produced the audience. [00:08:32 to 00:09:17] Carla Sage pairs geofenced traffic with immediacy copy like “top rated activity near you today” or “book direct and save 10 percent,” and pairs ZIP code and CRM records with neighborhood copy like a local pass or a weekend family discount. Warm website visitors who left without buying in the last 14 days get the retargeting line: still planning your tour, secure your spots before they sell out.
• Meta gives you three audience types, and the third only works if you build the first two properly. [00:10:03 to 00:11:37] Nick Kogos separates imported data, meaning anything with contact information including past visitor lists, email lists, sales lists and trade show lists, from website and app traffic audiences, from lookalikes built off either. Once a list is imported you can narrow it inside Meta by location, interest or behavior, so a full previous-season visitor list can be cut down to one city without re-exporting anything.
• Write the pixel rules around content differences, then serve different ads to each bucket. [00:12:23 to 00:13:09] Nick Kogos says the setup work is deciding who goes in which group: everyone who hit the homepage, people who completed a booking, people who started and did not. The test for whether a split is real is whether the content differs. Kid-friendly activity visitors and couples package visitors should not see the same creative.
• Keep lookalike audiences from competing with each other and with their source lists. [00:13:09 to 00:14:41] Meta scans the profiles in an uploaded audience and finds similar people on a 1 to 10 percent slider, where 1 percent is closest to the original and higher percentages broaden it. Nick Kogos warns against running lookalikes built from unrelated source audiences alongside each other, and says to keep each lookalike grouped with its original audience or separated entirely.
• Clean the list before uploading it or the platform will do it for you, badly. [00:13:54 to 00:15:28] Nick Kogos calls list hygiene the real hurdle: remove duplicates, strip the dummy email addresses and placeholder names staff entered at the point of sale. Meta needs an email address or a phone number at minimum, and every extra field improves the match rate.
• Build a full funnel and exclude your lists from one another. [00:14:41 to 00:15:28] Nick Kogos wants people pushed down the funnel deliberately rather than every audience receiving every ad. Somebody who has already booked should be receiving an ad for a different experience, not the one they bought.
• On Meta, split the daily budget 60/30/10. [00:15:28 to 00:16:14] Nick Kogos puts 60 percent on imported audiences, 30 percent on website and app retargeting, and 10 percent on expanding through lookalikes. The weighting reflects intent: the imported lists are your actual customers and leads, not bought eyeballs.
• Split audiences three ways that operators usually collapse into one. [00:16:14 to 00:17:48] Nick Kogos names season, where a summer visitor gets served winter creative to convert them a second time, and locals versus visitors, which he sees grouped together too often when locals should be getting staycation and local discount messaging. The third split is funnel position: greatest hits content for people who joined the email list but never visited, and niche content for repeat guests you are trying to move onto a different activity.
• Google’s list import is stricter than Meta’s but can connect to your POS automatically. [00:17:48 to 00:19:21] Nick Kogos points to Google Audience Manager for custom audiences built from buyer, email, sales and trade show lists, and flags the automated connections available from many booking and POS systems, which removes the manual export entirely. Google demands more than an email and a phone number: expect to supply ZIP code, country or first and last names before a list is accepted.
• Google Tag Manager and Google Analytics make website remarketing faster to build than Meta’s version. [00:19:21 to 00:20:06] If either is already installed, Nick Kogos says the connection is direct and the URL rules for remarketing groups are set up inside Google. He describes it as a considerably easier system than building the same groups in Meta.
• If you post to YouTube, you already have a Google audience you are not using. [00:20:06 to 00:20:52] Google owns YouTube, so you can build audiences from people who interacted with your channel or individual videos and target them through Google Ads. Nick Kogos calls this the standout advantage for operators with an active YouTube presence.
• Google lookalikes only run in Demand Gen campaigns right now, which is why the Google budget split differs. [00:20:52 to 00:21:38] Lookalikes are unavailable in search and display, and are limited to the Demand Gen mix of YouTube, Gmail and Discover placements. Google told Nick Kogos it plans to open lookalikes to all products later this year. Until then he moves the website traffic share up to 40 percent and holds imported audiences at 60 percent.
• Build a negative keyword list before you build the campaign. [00:21:38 to 00:22:25] Nick Kogos starts with free, jobs, Wikipedia and salary, which pull in job seekers and researchers rather than buyers. His destination example: Durango, Colorado is a different search from Durango, Mexico, which is different again from the Durango car, and all three will spend your money if you let them.
• Bid and build for a phone, then run the campaign when someone can answer it. [00:22:25 to 00:23:17] Nick Kogos prioritizes mobile device bids and asks whether the vertical video ad works without sound, because if it does not, it needs captions. He also schedules campaigns to run during open hours so a guest who sees the ad and wants to speak to a human can reach one.
• Getting people to the site is half the job, and the checkout drop-off will eat the other half. [00:23:17 to 00:24:51] Carla Sage sets a concrete bar: a guest picks a date and checks out on mobile in under 60 seconds. Refund and weather policies belong next to the pay button so the guest is confident at the moment of payment. Apple Pay and Google Pay need to be enabled, because typing a 16 digit card number on a phone is where bookings die.
• The learning curve costs less than the commission, and the training is free. [00:26:23 to 00:27:56] Carla Sage acknowledges OTAs are easy and do sell tickets, and says the trade is the guest relationship plus the margin. Facebook and Google are built for ordinary people to use, and there are thousands of free training videos covering the simple version. Her point to operators: you do not have to become a specialist to make this pay.
• Operators are losing 10 to 35 percent of their Google discoverability, which raises the stakes on everything else. [00:30:13 to 00:30:59] Peter Syme says nothing in the session is new, but the audits he has seen over the past year show traffic to operator sites down significantly. Destination visitation can be up while an individual operator’s digital traffic falls, which makes owned audiences and paid targeting more load-bearing than they were.
• A new operator with no customer list starts with in-platform targeting and feeds the list as it grows. [00:30:59 to 00:31:45] Nick Kogos answers an operator with few existing clients: you already know roughly where your bookings come from, so use Meta and Google’s built-in targeting options for that geography and age range. Then keep re-uploading the client list as it builds. He describes it as a slow game rather than a difficult one.
• Ask your DMO or CVB for the visitation data instead of guessing your source markets. [00:31:45 to 00:32:31] The Datafy team suggests starting from an ideal customer profile, then checking it against the top ten domestic origin markets and top ten source countries the destination organization already tracks. Attraction operators nearby are a second free source of the same information.
• Target who actually books you, not the visitor you pictured when you started. [00:33:17 to 00:34:02] The Datafy team pushes back on persona work done in isolation. Past guest lists and website traffic show who is buying the product, and that is frequently not who the operator imagined. Build the lookalike and prospecting audiences off the real buyer.
• Sketch the customer as a person and put the sketch on the wall. [00:34:02 to 00:34:50] Nick Kogos builds a physical profile, his example being Amy from Alabama: what she wears, what she is into, which activities she picks. When creative gets built, somebody is looking at Amy rather than at a demographic range.
• Google Things to Do puts operators above search results in a space DMOs cannot enter. [00:34:50 to 00:35:37] The Datafy team explains that these booking ads sit above Google’s organic search results and require a booking engine attached to the tour or activity. DMOs mostly do not have one on their site, so they are being pushed further down the page while operators can buy the top position. This is the specific opportunity created by the traffic losses everyone is feeling.
• A saturated, expensive destination makes owned-data targeting more important, not less. [00:37:08 to 00:37:54] Answering a Greek operator on Santorini ad costs, the Datafy team argues the operator has no control over when, how or to whom an OTA serves their listing, and complete control over that in Meta and Google. Both platforms let you enter at a much lower budget than an OTA’s advertising products cost.
• Advertise to the group booking, not the couple, and let the couple book anyway. [00:38:39 to 00:39:24] Peter Syme never ran an ad that looked like it was chasing one or two people. He targeted groups of four and up, because a group of 15 on a $75 tour carries a far larger advertising budget than a pair does. Couples still booked as a side effect, as long as the landing page made clear that two people were welcome.
• Fund the audience paying your bills first, and reach the hard audience somewhere cheaper. [00:40:10 to 00:41:45] Peter Syme describes ads that printed cash targeting the UK for a Spanish operation while a US West Coast audience never worked. Carla Sage’s response is that the distant audience will probably never pay for itself the same way, so the question becomes where you meet those visitors instead: at the airport when they land, or at another attraction mid-trip, rather than spending to convince them to build a whole trip.
• Choose partners who are good where you are not, and hand the expensive audiences to them. [00:41:45 to 00:44:47] Peter Syme has asked thousands of operators at in-person events how many sales partners they have. The answer is usually under five, and finding one with 30 or more happens in a single-digit percentage of cases. His own partner network was built by function: OTAs brought international fillers he could not target profitably, while he kept the audiences he converted well. The Datafy team agrees the split makes sense, because the money not going to a 30 percent commission on the audiences you already win compounds.
• You cannot set an ad budget until you know your current cost of acquisition. [00:44:47 to 00:46:20] Peter Syme answers the budget question with a question, every time. An operator calling 30 percent expensive without knowing what a direct customer costs them has no basis for the comparison. Doing that arithmetic is pre-work, before any campaign is built.
• Run the acquisition math against the booking value before you decide the spend is too high. [00:47:05 to 00:48:36] Peter Syme describes a multi-day operator complaining about $230 to $300 per customer and looking for places to cut. Their average retail price was around $5,000, their average booking was two people, and their margin was about 35 percent. Spending $300 to land a $10,000 booking is not a complaint, it is a signal to spend more.
• Lifetime value can make the acquisition cost disappear, and only first party data lets you collect it. [00:48:36 to 00:49:21] The Datafy team extends the same example: the $5,000 customer becomes a $35,000 customer across five or six bookings, at which point the $235 acquisition cost stops mattering. Through an OTA you never see that guest again, because you are one commoditized option in their algorithm and they get fed something else locally.
• A guest who visits your city once has almost no lifetime value, and that is a business structure problem. [00:49:21 to 00:50:52] Peter Syme says the worst customer in tourism is a tourist, because inbound one-time visitors come to Denver or San Francisco once and never return. Referral systems help, but an operator fully exposed to one-time tourists carries a structural exposure that no amount of digital marketing fixes. What the lifetime value is changes what you can afford to pay to acquire them.
• Platform geo-targeting gives you the audience but not the data behind it. [00:51:39 to 00:52:25] Meta and Google let you exclude locals and target visitors in an area, but nothing comes back about where those visitors came from or how long they stayed. That gap is the difference between targeting a market and knowing one.
• Real location data work needs a data broker, a media buyer and a visualizer, which is why operators rarely do it alone. [00:52:25 to 00:53:12] John Marshall is upfront that this is the sales-y answer. Datafy owns its US-wide mobile device panel rather than reselling it, which lets them build custom audiences from it, and matches it against nationwide credit card data. Combined with an operator’s booking data and first party records, that is three separate vendor relationships an operator would otherwise have to assemble.
• Let the weather pick the creative, and shoot content so you have a version for every kind of day. [00:53:59 to 00:56:15] Peter Syme’s water-based adventure business posted live reels of guests in the sun on good days and saw next-day bookings jump. On bad days they showed the covered activities, aerial adventure or paintball, rather than horizontal rain. The Datafy team adds the reverse case from a lower Manhattan museum, which ran ads specifically when it rained, and warns that a content library where every photo is two people in sunshine cannot support any of this.
• Last minute bookers decide at breakfast or over dinner the night before, and your ad has to be in that moment. [00:54:44 to 00:55:29] Peter Syme describes the actual decision: nobody lands in a destination with the museum tour, walking tour and food tour already chosen. They are sitting there wanting to do something, and every other signal around them is shaping the choice. Geographic ad drops only work if the creative reads as one of those influences.
• Set weather and time-of-day ad rules in advance instead of posting reactively at 5am. [00:56:15 to 00:57:00] The platforms allow budget to be held back and released on conditions: serve this set when it rains, this set when the sun is out, this set at a particular hour. The Datafy team frames it as the version of live posting that survives an operator who does not have the time or the expertise to do it manually.
• Measure acquisition cost per channel as separate lines rather than looking for one blended number. [00:57:46 to 00:58:31] Answering an operator juggling direct, Meta, Google, B2B partners and OTAs, the Datafy team says having multiple channels is not the complication it looks like: it is one graph with five lines and a different cost on each. That is also the only view that tells you whether the OTA’s 30 percent is beating the 15 percent you are paying inside Meta.
• The ad platforms already calculate your acquisition cost, and AI will read the report for you. [00:58:31 to 01:00:04] With conversion tracking live and the booking system connected, Meta and Google compute acquisition cost and conversion rate per platform and will report it. The Datafy team suggests handing the exported Google Ads report to Gemini with the question attached, and having it build an ongoing spreadsheet from the same data.
• Importing your customer list does not mean advertising to your customers, unless you choose that at setup. [01:00:50 to 01:01:36] An operator in Mauritius who ran Meta ads to the UK market got inquiries and no bookings, and asked whether uploading their client list would re-target the same people. Once the list is in the platform, you decide: serve the list directly, or build a lookalike from it and serve people who resemble them. Both options stay open after the import.

About the partner

• Website: datafy.com
• One-sheet: Datafy one-sheet
• Talk to John Marshall, Director of Business Development: john@datafy.com