You have Google Ads running. Someone posts on Instagram most weeks. A newsletter goes out when a colleague remembers. An SEO agency sent a report last quarter that nobody opened. Every piece is defensible. Together they do not reliably turn strangers into departures.
That gap is not laziness and it is not a small budget. It is structural, and the structure belongs to travel. Consideration windows run for weeks or months, so the thing that caused a booking has usually stopped running by the time the booking lands. Demand is violently seasonal, so the same channel can look broken in October and brilliant in February with nothing changed but the calendar. Ticket values are high and margins are thin, so a small move in enquiry quality is worth more than a big move in click price. And often an OTA sits between you and the traveller, owning a relationship you paid to create.
This piece gives you a way to see marketing as one system with four parts, a method for finding the single constraint worth fixing this quarter, the arithmetic that should govern every budget decision, and a way to plan around seasons and lead times.
Why channel-by-channel thinking quietly fails in travel
Most travel marketing plans are organised by channel because that is how agencies and job titles are organised. One person owns paid, one owns social, one owns email. Each reports on their own numbers, and each number can improve for two quarters running while bookings stay exactly where they were.
That arrangement works where the purchase is fast and cheap: see ad, click, buy, done. Travel does not behave like that. A family choosing a three-week Namibia self-drive reads your itinerary in February, comes back in April through a branded search, sends an enquiry in May and pays a deposit in June. Which channel gets the credit? All of them and none of them.
So you get predictable behaviour. Paid leans on brand terms because they convert. Email sends to people who have already travelled with you because open rates look healthy. Content chases destination keywords with no path to a bookable product. Everyone hits their number and the business does not grow. The fix is not better attribution software. It is changing the unit of management from channel to system.
The four systems that actually produce a departure
Whatever your size, your marketing does four jobs. Naming them separately is what lets you find the broken one instead of spreading budget evenly across all four and hoping.
- Demand generation: creating intent among people who do not know you yet. SEO, destination content, paid social, partnerships, PR, the trade.
- Conversion: turning attention into a qualified enquiry or booking. Itinerary pages, pricing clarity, departure and availability display, photography, the enquiry form, the booking engine.
- Sales operations: everything between the enquiry and the deposit. First response time, quoting, itinerary revisions, follow-up, and the system that stops leads going cold.
- Retention and referral: repeat departures, past-traveller reactivation, reviews, referrals, and repeat volume through agents.
Sales operations is the system most operators ignore
For a DMC or tour operator the enquiry is the halfway point, not the finish line. If a tailor-made request sits unanswered for two working days while a competitor replies within two hours with a rough route and an indicative per-pax price, no amount of paid media fixes that. You are buying enquiries and pouring them into a bucket with a hole in it, then blaming the tap.
Accommodations and campings run the same four systems under different names. There the neglected one is usually the family who books the same pitch every August and would keep doing it for a decade if you asked them once, in January, before they start browsing.
Fix one constraint at a time, and prove which one it is
Systems have a binding constraint: one place where the whole thing is limited, and where adding capacity anywhere else changes nothing. If your enquiry inbox is the bottleneck, another ten thousand euro of paid social buys you a longer queue of people waiting on a quote, not more departures. Most travel businesses skip this diagnosis, because fixing everything a bit feels like progress and fixing one thing feels like neglect.
Say your enquiry-to-booking rate is 8 percent. Doubling ad spend buys twice as much of the same disappointment: twice the enquiries, the same share going cold, twice the invoice. Moving that rate from 8 to 16 percent doubles booked revenue on the media budget you already have, and it does not require the market, the season or the exchange rate to cooperate.
Be wary of anyone who hands you an industry-standard conversion rate to measure that against. The only comparison worth anything is your own rate, split by source and by product line, because a group departure enquiry and a tailor-made honeymoon enquiry are not the same event and never convert at the same speed.
Finding your constraint is a Monday-morning task, not a philosophy. Open last season's numbers, walk the four systems in order, and stop at the first place where they stop making sense.
- Too little qualified traffic in your booking season, but what arrives converts normally. Constraint is demand generation.
- Plenty of relevant traffic, very few enquiries. Constraint is conversion.
- Plenty of enquiries, very few deposits. Constraint is sales operations.
- Healthy new bookings, almost no repeats or referrals. Constraint is retention.
Do the per-booking maths before you touch a budget
The number that should govern every budget decision is gross profit per booking, not revenue per booking. Most of what a travel business invoices is pass-through: flights, accommodation, ground handling, guides, park fees. Two operators can sell the same 4,000 euro trip and have completely different amounts of money available to win it, because one keeps 15 percent of that and the other keeps 35. Revenue per booking cannot tell you which one you are.
Here is a worked illustration, with numbers I have made up to show the mechanism rather than to describe your business. Say a tailor-made trip sells at 4,200 euro on an 18 percent gross margin, so 756 euro of gross profit per booking. At a 20 percent enquiry-to-booking rate, each booking consumes five enquiries. Buy those at 60 euro each and acquisition costs 300 euro, leaving 456 euro before overhead. Workable.
Now change one variable. Enquiry quality drops and the rate slips to 12 percent. You now need roughly 8.3 enquiries per booking, so at the same 60 euro each, acquisition costs 500 euro against 756 euro of gross profit, before anyone has been paid a salary. Your cost-per-lead dashboard has not moved a cent, because the cost per lead genuinely did not change. That is how a travel business loses money while the reporting looks stable.
The OTA question answers itself with the same arithmetic. Take a 180 euro day tour sold through a marketplace on, say, a 25 percent commission: that is 45 euro handed over on every sale, every season, indefinitely. If you can acquire the same booking directly for 22 euro, you keep roughly 23 euro plus the customer record, and the record is the part that can pay you again next year. Use your own contract rate rather than a number from an article, including this one. And be honest about the trade: the marketplace only charges you when it sells, while your ad budget spends whether anyone books or not.
Plan around the season you are selling, not the month you are in
The most common structural error in travel marketing plans is a budget calendar built on the calendar year while demand runs on a booking curve. If your peak departures are July and August and your median lead time is fourteen weeks, the money that fills July is spent in March. Spend it in June instead and you are bidding for the last-minute residue at the worst prices of the year, against every operator with unsold beds.
Platform tooling will not rescue you from that. Google's own Ads Help documentation on seasonality adjustments says they are ideal for short events of one to seven days and may not work as well if you use them for extended periods of more than fourteen days at a time, because Smart Bidding already accounts for seasonal events. A whole shoulder season is not a bidding-lever problem. It is a budget-and-structure problem you solve months earlier.
- Plot two or three years of bookings twice: once by departure month, once by booking month. Different curves, and only one of them tells you when to spend.
- Set budget against the booking curve, with a deliberate ramp before the curve steepens, rather than a flat monthly figure.
- Decide in advance what happens to shoulder capacity: discount, repackage, shift source market, or accept lower occupancy. Deciding in week three of a slow September is a reaction, not a strategy.
- Record lead time per product line. Group departures, tailor-made FIT and short breaks rarely share a booking window, and averaging them hides both.
Measure booked revenue, not leads
A lead is not revenue, and in travel the gap between the two is measured in weeks or months. That gap is where most marketing reporting quietly falls apart.
First, feed real commercial outcomes back into the platforms instead of counting form fills. Google's own Ads Help documentation on value-based bidding says the values you report can be real economic values, like revenue, or proxy values, like a lead score, and that Target ROAS should have at least 15 conversions in the last 30 days at the conversion tracking level. Send booked value back and the bidding starts hunting for the fourteen-night trips. Leave it counting enquiries and it will faithfully find you the cheapest enquiries on the market, which in travel usually means people idly pricing a trip they will take in three years.
Second, respect the attribution window, because travel routinely runs past it. Google's offline conversion import documentation says a conversion can be uploaded against a Google click ID so long as it is no longer than 90 days after the click, and 63 days for enhanced conversions using personally identifiable information. If your median enquiry-to-deposit time is 100 days, the honeymoon that ad genuinely caused can never be credited to it. That is not a bug to fix. It is a limit to design around.
So pick an intermediate event that sits inside the window and correlates with revenue: a quote sent, a consultation held, a deposit link issued. Give it a value equal to its historical expected booking value, then upload the actual booked value for the deals that close in time. Google's own guidance is to upload at least daily, or failing that on a consistent regular basis. Holding the click ID against the enquiry record and pushing value back as the deal moves is a job for the system your sales team already lives in, not a spreadsheet someone updates on Fridays.
The mistakes I see most often, and what to do instead
- Judging campaigns on cost per lead. A cheap enquiry from someone shopping for a 600 euro city break is not progress when you sell 6,000 euro safaris. Instead: report gross profit per booking, split by product line.
- Rebuilding the whole website as the answer to everything. It pauses every other kind of progress for six months, usually before anyone has identified the constraint. Instead: rewrite the three highest-traffic itinerary pages and watch enquiry rate for six weeks.
- Publishing destination content with no commercial path. An article on the best things to do in Lisbon that never connects to a bookable trip, departure or room is a hobby with a content calendar. Instead: link every piece to a specific product a reader can actually buy.
- Treating the enquiry inbox as administration. Instead: measure first-response time weekly and treat anything over four working hours in season as an incident with a name attached to it.
- Discounting by default to fill the shoulder season. It teaches your best repeat guests to wait, and they will wait next year too. Instead: change the package or the inclusions so the comparison is not like for like.
- Buying brand terms and calling it demand generation. Defending your name in the auction is sensible, but people searching for you already know you exist. Instead: judge demand generation only on non-brand enquiries, reported separately.
What to do first, and what to leave alone
Ninety days is enough to change one system properly and not enough to change two. In travel you also have to choose which ninety days: rebuilding how enquiries are handled is worth far more in the weeks before your booking curve steepens than in the fortnight it peaks, when nobody has time to change anything anyway. Treat the quarter as a single question, sequenced so the answer is unambiguous at the end of it.
Write the not-doing list at the same time: the channel you are not launching, the rebrand you are deferring, the trade show you are skipping. In week six a competitor will publish a beautiful new itinerary page or start bidding on your brand name, and someone will want to respond that same week. Without the list, that response becomes an unbudgeted project and the quarter ends with four half-finished things and no answer to your question.
None of this needs new tools or a bigger budget. It needs you to know which single number you are moving, and to leave every other number alone while you move it. Here is the order to start in.
- Week one: build the booking curve. Export 24 to 36 months of bookings with enquiry date, booking date, departure date, revenue and cost of sale. Everything else depends on that one file.
- Week two: calculate gross profit per booking by product line, and enquiry-to-booking rate by source. Rank sources by profit contributed, not enquiry volume. Then name the constraint in one sentence, somewhere the team can see it. If two people in the room would name different constraints, you are not ready to spend money.
- Weeks three and four: capture the click ID on the enquiry record, get booked value flowing back, and agree the intermediate conversion event.
- Weeks five to twelve: work the constraint and nothing else. Set one weekly number and look at it every Monday.
- End of quarter: keep what moved the number, kill what did not, then pick the next constraint.