A good site, but nobody could see what it delivered

The front end was in order: the promise was immediately clear and the route to an enquiry short. What belongs underneath was missing entirely, which left no way to see what visibility returned.

Visit the website

The result

AI referral traffic: +75%
AI bot traffic: +54%
AI citation share: +24%
AI visibility: +11%

Measured in the client's own analytics and server logs, not an estimated score.

The situation

Tropikal Rentals rents cars on Curaçao to travellers who plan their holiday weeks ahead, most of them from the Netherlands. That search starts online, and increasingly with an AI being asked for a recommendation rather than with a list of search results.

The site itself held up well. The title said in fifty characters what the business does and where, there was one clear main heading, and there was a working form on the home page. Someone who was convinced could act straight away.

Underneath that there was nothing. No analytics, no ad pixel, no sitemap, no robots.txt and no structured data. Every euro spent on visibility disappeared into the dark, and an AI model that wanted to know what kind of business this is and where it sits had to infer it from running text.

Where it started

52 / 100

Measured with Oddbird's findability scan on 8 September 2026. Anyone can repeat it on the same domain.

Message clarity100
50-character title, one main heading, meta description present
Conversion path90
Form on the home page, visible actions towards contact
Search visibility50
No sitemap, no robots.txt, no canonical
Trust and proof40
Only a phone number; no reviews, address or registration details
AI readiness30
No structured data
Insight0
No blog, no analytics, no pixels

The plan

In order of leverage.

01
Set up measurement
Without measurement every next spend is a guess, and afterwards there is no way to say what it returned.
Week 1
02
Technical foundation: sitemap, robots.txt and canonical
What crawlers need to find every page, instead of inferring it from links.
Week 1 to 2
03
Structured data for a local rental business
So Google and AI models know what kind of business this is, where it sits and what it rents.
Week 2 to 3
04
Make proof visible: reviews, address and registration
Someone booking ahead from the Netherlands with an unfamiliar company on an island wants to know who they are dealing with.
Week 3 to 6
05
Content around the questions travellers ask beforehand
The layer AI models quote from when someone asks where to rent a car on Curaçao.
Ongoing

Where the gain was

The biggest gain was not in the site itself but in what was missing underneath: measurement and visibility in AI. The story was already there, and that saved a phase.