Booking.com reviews of stays on Lanzarote, 2022 to 2026
The island scores 8.6. Then you read what people wrote.
Every guest leaves a number. Two in three also leave a few sentences. This is what those sentences add to the number.
The corpus
0
reviews of properties in localities, written by guests from countries over five years.
The score
0out of 10
on average. % of stays score 8 or more, and the single most common score is a perfect 10. Read on its own, the island looks uniformly happy.
Distribution of review scores, all stays 2022 to 2026.
The problem with the score
0%
of established properties average below 7. Nine in ten sit between and . A destination cannot manage with a number that cannot tell its properties apart.
Average score per property, properties with at least 20 reviews.
What guests wrote
0
complaints, written into % of all reviews. This is the layer the average hides, and unlike the score it can be read topic by topic.
Booking.com asks separately what the guest liked and disliked. Each answer is split into sentences and tagged with the topic it talks about, in eight languages. Twenty-three topics, validated against a language model on a sample of 1,200 complaints.
The untouchable asset
0%
of reviews with text praise the location. Cleanliness, staff and the pool follow. And it is praised everywhere: no locality on the island falls below 29 mentions per 100, and the top of the range belongs to two fishing villages, Punta Mujeres and Famara, not to the resorts.
Out of every 100 reviews that contain written text, how many mention the topic. Green: praise. Red: complaint. Location is read from the praise side, where the wording is unambiguous; its complaint side is the least reliable of the 23 topics.
What breaks
The five most common complaints are all inside the room.
Beds, cleaning, the bathroom, wear and tear, the kitchen. Not the beach, not the food, not the staff.
Out of every 100 reviews that contain written text, how many complain about the topic. Top ten. One review can mention several topics, so these do not sum to 100.
What hurts
0out of 10
is the average score when a guest says the listing did not match reality. Pests, an absent host and damp follow. A small room or a noisy street costs about a point and a half less.
Average score of reviews that complain about each topic. Island average shown for reference.
What stars buy
Stars fix the building. They do not fix the bill.
Complaints about wear and tear fall steadily from one star to five, and cleanliness falls faster still. Complaints about price do the opposite, and five-star guests are the most likely of all to say they were overcharged.
Out of every 100 reviews that contain written text, how many complain about the topic, by the property's star rating.
Eternal spring, in theory
0times
more heat complaints in August than in winter. Cold complaints mirror them in December and January. Same buildings, opposite failures, and neither is what the island sells.
Out of every 100 reviews that contain written text, how many complain, by month of stay. Complete years only.
Four complete years. The finding.
The score did not move. The complaints did.
From 2022 to 2025 the average went from to . Underneath, complaints about food and staff fell by a fifth or more. Complaints about the fabric of the buildings rose.
Change in the complaint rate between 2022 and 2025. Service topics fall, building topics rise.
The clearest signal on the island
0% damp and mould
Complaints about damp, mould and smells rose by half in four years, faster than any other topic. It is also one of the most reliably measured: the words are unambiguous in every language, and it agrees with an independent model of the time.
Out of every 100 reviews that contain written text, how many complain about damp, by year of stay. 2026 covers January to June only, which is the wet half of the year, so the hollow bar is not comparable with the four solid ones.
How long people stay
0% stay exactly 7 nights
One number towers over the rest. The package week is still the unit of demand here, with a second, smaller island of city breaks at two and three nights.
Reviews by number of nights stayed, capped at 21.
Two islands
0vs nights
Spanish guests stay four nights, a third of them in summer. British, Irish and German guests stay seven, mostly outside it. They are not on the same holiday, and they do not complain about the same things.
The twelve largest origin markets, ordered by how many reviews they write. Share of stays in June to August covers complete years only.
Where all of this happens
0%
of every review on the island comes from three coastal strips. % comes from just 100 properties out of . The reviewed island is a very small part of the island.
Left: every property with at least fifteen reviews, sized by how many, settling where it actually is. Right: share of all reviews by locality, top twelve.
The honest null
0%
of reviews with text complain about crowds, and the share fell over the four complete years. Whatever the island's carrying capacity debate says, guests do not write about Lanzarote as full. Their rooms, yes.
Crowding is measured on the same footing as every other topic, and it is one of the least reliable of the twenty-three, so treat it as directional. A low number here is still a finding, not a gap.
What a destination can do with this
Manage the text, not the number.
Track topics, not scores. A complaint rate per topic per month moves when something changes on the ground. The average does not.
The stock is ageing faster than the service. Food and staff complaints fell for four straight years while damp, bathrooms and kitchens rose. That is a capital problem, not a training one.
Season the buildings. Heat in August and cold in January are the same failure: rooms sold on a climate they cannot deliver indoors.
How this was made
Source: Booking.com guest reviews for Lanzarote properties, check-in years 2022 to 2026, reviews after deduplication. 2026 ends in June and is excluded from every trend and seasonal curve. Scores are Booking's 1 to 10 scale; % of reviews carry free text and the rest are one-word labels, excluded from all text rates.
Each review body is split into its positive and negative side, and the negative side into sentences. Sentences are matched against a multilingual topic lexicon of 23 topics in 8 languages, built by reading stratified samples. It tags % of complaints.
A sample of 1,200 complaints, 200 in each of six languages, was independently labelled by a language model and used to score every topic. Ten topics reach an agreement of 0.75 or better and carry the figures above; six are usable with caveats; seven, including crowds, location and accessibility, are directional only and are marked as such in the explorer. The full scoring table ships with the project.
Nothing in this story identifies a property or a guest.
Built for NECSTouR. Analysis and design by Josep Ferrer. Data: Booking.com reviews, generated .