Overview

Paid advertising works for experiences, but not in the way most operators are first told it does. This guide covers the approach: what each channel is good at, how to set a budget you can defend, and what to measure so you know whether it worked.

If you already know you want to run Meta (Facebook and Instagram) Ads, read this first and then go to Meta Ads: Campaign Setup & Playbook.

1. The one rule that matters

Optimise for bookings, not for clicks.

Everything else in this guide follows from that. Clicks are cheap, easy to buy, and a large share of them are not people — bots, scrapers and accidental taps on low-quality placements. An Ad platform will happily give you as many clicks as your budget allows, and if clicks are what you asked for, that is exactly what you will get.

Bookings cannot be faked. A completed payment is a real person with a real card. When you tell an Ad platform to optimise for purchases, its algorithm goes looking for people who behave like your past buyers, and the bot traffic stops being profitable for anyone to send you.

This has one practical consequence worth knowing up front: purchase-optimised campaigns need conversion data before they work well. Expect a learning period. A campaign with fewer than about 20–30 conversions in a week is still guessing.

2. Which channel does what

The single most common mistake is treating all Ad channels as interchangeable. They are not.

Channel	What it does	Best for
Google Search / Things To Do	Captures demand that already exists — someone is searching “sunset cruise Lisbon” right now	Evergreen products with steady search volume; well-known destinations; high-intent, bottom-of-funnel
Meta (Facebook / Instagram)	Creates demand — nobody was looking, but the right person sees something they want	Time-bound events, new launches, seasonal pushes, visually striking experiences, local audiences
TikTok / Snapchat	Creates demand with a younger audience, heavily creative-dependent	Social, group and nightlife experiences; strong video
Affiliates / partners	Borrowed trust — someone else’s audience recommends you	Anything where a recommendation beats an advert

Search captures. Social creates. If people are already searching for what you sell, start with search — it is cheaper per booking and needs less creative work. If nobody is searching because they do not know your experience exists, or because it only happens on three dates in October, search has nothing to capture and social is where you belong.

Most operators end up running both, for different products.

3. Set the budget from the booking, not from the Ad

Work backwards from what a booking is worth to you.

Average booking value. Take your last 90 days of paid bookings, ex-tax.
What you can afford to pay for one. If your average booking is 100 and your margin after supplier costs is 30, then 30 is your absolute ceiling per booking and something closer to 10–15 is a healthy target.
That target is your CPA (cost per acquisition). Your ROAS target is simply average booking value ÷ CPA — a 100 booking at a 15 CPA is a 6.7× ROAS.
Daily budget = target CPA × bookings you want per day. Do not start below roughly 10× your target CPA per day, or the platform will never gather enough data to optimise.

Two adjustments worth making:

Repeat and group bookings. If a typical customer books twice, or brings three friends, you can afford more for the first booking than the first booking alone is worth.
Lead time. Experiences are often booked weeks or months ahead. A campaign judged after seven days will look worse than it is. Judge on a window at least as long as your typical booking lead time.

4. What to measure

Ignore the vanity metrics. These four answer whether it worked:

Metric	What it tells you
Cost per booking (CPA)	The number you can act on. Compare it to §3’s ceiling.
ROAS	Revenue ÷ spend. Useful for comparing campaigns; useless without knowing your margin.
Bookings	The count that makes the other two meaningful. Ten bookings is a signal; two is noise.
Net contribution	Revenue − Ad spend − cost of sale. The only one that is actually profit.

Two cautions that catch everybody:

The Ad platform’s conversion count will not match your booking count. The platform counts conversions its own way, across its own attribution window, including views as well as clicks. Your booking records are the truth about money. Use the platform’s number to steer the campaign; use your own to decide whether to keep spending.
Never compute a conversion rate by dividing bookings by Ad clicks. Different populations, different counting. It will always look wrong, in an unpredictable direction.

5. Set-up checklist before you spend anything

Advertising into a broken measurement setup is the most expensive mistake available.

[ ] Your own pixel or tag is connected. In Spotinga: Organization → Settings → Integrations → Advertising. Without this, your Ad account sees no conversions at all and cannot optimise for them.
[ ] Test a real booking end to end and confirm the purchase event arrives in the Ad platform’s event tool, with the right value and currency.
[ ] Your product pages load fast and answer the obvious questions — duration, what is included, meeting point, cancellation policy. Paid traffic is unforgiving: a visitor who has to hunt for the price leaves.
[ ] Availability is real. Nothing wastes budget faster than Ads for sold-out dates.
[ ] UTM parameters on every Ad link, so you can tell campaigns apart in your own reports rather than trusting the platform’s.
[ ] A single landing page per campaign, matching the Ad’s promise. If the Ad says “sunset kayak”, the page must open on the sunset kayak, not on your homepage.

6. Mistakes that quietly waste budget

Optimising for traffic or engagement. You will get traffic and engagement. You will not get bookings. See §1.
Broad targeting with no purchase signal. Early on, before the platform has conversion data, broad targeting is mostly a way to buy impressions from the cheapest available inventory. Start narrower, widen once conversions are flowing.
Running Ads to a homepage. Every extra click between the Ad and the booking form costs you a share of the traffic you just paid for.
Changing the campaign every day. Edits reset the learning phase. Give a campaign at least a week and a meaningful number of conversions before judging it.
Too many campaigns on a small budget. Five campaigns at 10/day each will all stay in learning forever. One campaign at 50/day will not.
Advertising products that do not convert organically. Paid traffic amplifies what is already there. If a product converts badly with free traffic, paid traffic buys you the same bad conversion rate at a price.
Ignoring cancellations. If a channel drives bookings that cancel, its real ROAS is much worse than reported. Check refund rates per campaign.

7. A sensible starting sequence

Weeks 1–2 — measure. Connect tracking, confirm a test purchase is recorded, tag your links. Spend nothing.
Weeks 3–6 — one channel, one product, one campaign. Pick your single best-converting product and the channel that suits it (§2). Optimise for purchases. Budget per §3.
Weeks 7–10 — read the result honestly. Cost per booking against your ceiling. If it is close, scale the budget slowly — roughly 20% at a time. If it is far off, the problem is usually the offer or the landing page, not the bid.
Then widen. Add a second product, or retarget people who viewed but did not book — almost always the cheapest bookings you will buy.