About
How we know
Every “best of” site claims to know Aruba. Here's the actual machinery behind ours — what we read, how often we refresh it, where AI helps, and where it's not allowed to touch the steering wheel.
How a place earns a spot
Resorts, restaurants, beaches, and tour operators don't get listed because they asked. Every page starts as a hand-curated pick, then has to keep earning it against real traveler signal: its TripAdvisor rating and ranking, weighted by how many reviews back it up. A 4.9 from twelve people loses to a 4.6 from two thousand — small samples get small trust.
On top of that sits what we actually know about the place — location, price tier, noise level, who it suits and who it doesn't. We refresh review data weekly, and if a place starts slipping, its standing slips with it.
One thing that never enters the math: money. Some outbound links are booking-partner links (the full story is on the disclosure page), but partner payouts are not a scoring input, and a place with no partner link competes on exactly the same terms as one with ten.
How the planner picks for you
The trip planner is deterministic math, not an AI guessing. Your answers — who's going, budget, vibe, must-haves, dealbreakers — become weights, and every place gets scored against them: its attributes, its traveler signal, and penalties when recent reviewers complain about exactly the thing you said was a dealbreaker. Highest scores win. Same answers, same picks, every time.
AI shows up only after the picks are locked, to write the “here's why these fit you” explanation. It narrates the decision; it never makes it.
The Recent Pulse
The Pulse on each page answers one question: what are travelers saying about this place right now? Every week we pull fresh TripAdvisor reviews, then use AI to boil them down to what keeps coming up — the good, the bad, and whether things are trending up or cooling off.
Two rules keep it honest. First, every summary carries a confidence label — high, medium, or low — based on how many recent reviews stand behind it; a handful of reviews gets a hedge, not a headline. Second, everything is retold in our own words. We never reprint someone's review, and trend calls only happen when there's enough volume to be confident.
Community answers
The community section draws on public posts in large Aruba Facebook groups — thousands of travelers comparing notes in real time. Weekly, we collect those public discussions and use AI to find the questions that keep getting asked and the answers that keep holding up. Each guide shows how many traveler answers stand behind it, and every count comes from those actual answers — never an estimate.
Privacy rules are strict: anything with names, phone numbers, or other personal details is thrown out automatically, nothing is ever republished verbatim, and every answer is paraphrased so no one's post is traceable back to them.
Where AI fits, plainly
We use AI to read at a scale no human can — thousands of reviews and posts a week — and to draft summaries of what it finds. We don't use it to decide what's good. Rankings and planner matches come from explicit scoring rules a human wrote and can audit, and editorial picks are made by people. If a summary ever reads wrong, that's on us, and we want to hear about it.
Spotted something off — a closed restaurant, a stale price, a summary that doesn't match reality? Email hello@onehappyaruba.com and we'll check it against the source data and fix it.