How this site is researched

Most cycling sites will tell you Sa Calobra is 9.4 km, or 9.5 km, or 10 km; that it gains 668 m, or 654 m, or 682 m; that it averages 6.3%, or 7.0%, or 7.1%. All of those numbers are published somewhere. Almost none come with a statement of how they were arrived at — which is exactly why they disagree, and why there is no way to tell which to believe.

This page is that statement, for every figure on the site. It is linked from the foot of every page, so no single page has to keep explaining itself.

What measurement gives you

Everything here is worked from measurement and from published, checkable sources — public datasets, official publications, operators’ own pages. The power of it is consistency: the geometry of a road, the altitude of a pass, the distance from a town to the foot of a climb, and the rank of one climb against every other are all got right the same way, for every road on the island at once — and then re-checked next month. Almost nobody else even attempts it.

It is a sharper instrument than a ride report, not a blunter one. Where a page here tells you what a climb is, it is telling you what the road measures — its true gradient, its real steepest kilometre, where it stands against every other climb on the island — rather than how it felt to one person on one Tuesday in March.

Where the numbers come from

Every climb figure on the site is produced by one script, run over the same two public datasets:

• Road geometry — the centreline of the actual road, from OpenStreetMap (ODbL). Not a GPS trace of somebody’s ride, so no rider’s wander, stop or lost signal is inside the number.
• Elevation — Spain’s national 5 m digital terrain model, IGN Elevacion4258_5 (CC BY 4.0 scne.es). Not a phone barometer.

The rules the script applies are published in full on the climbs atlas, because a figure is only comparable with another when both were made the same way. In summary: the inventory of climbs is queried from OpenStreetMap rather than written by hand; roads tagged closed to the public are removed before anything is routed; elevation is sampled every 10 m; tunnels and bridges are interpolated between the portals, because a terrain model reads the mountain above a tunnel and the valley beneath a bridge; height gained is summit minus the lowest point on the road, not a running total of every rise; and the steepest kilometre is the largest gain over any 1,000 m window, sliding 10 m at a time.

Every departure made by hand from what the query returned — each rename, re-anchoring and exclusion — is listed on the atlas with its reason, row by row. An inventory nobody can audit is a list of opinions about which hills matter.

Checked against the world

A consistent method is worth nothing if it is consistently wrong, so the check is published too. Every pass here carries an elevation tag in OpenStreetMap set by somebody who is not us, and three summits were separately measured from the Mallorca 312 organiser’s official GPX — barometric and GPS elevations recorded on the road, sharing nothing with a LiDAR terrain model except the mountain itself. The full comparison table shows every difference, including the three that fall outside ten metres and the one place the measurement disagrees with every published source.

How a judgement is computed

Some things here are not measurements — rankings, the difficulty index, which side of a pass to take, which climbs are worth a day. Every one is computed from criteria published beside it, never assigned by hand:

• The difficulty index is height gained squared, divided by ten times the length in metres. It is printed rather than described so it can be argued with. An unpublished ranking is an opinion with a number next to it.
• A climb page’s verdict leads with whichever of three measures — height gained, steepest kilometre, length — that climb ranks highest on against every other climb-side here, and says so plainly when a climb sits mid-table on all three. It cannot see a commercial relationship, because none of its inputs is a commercial fact.
• Editorial calls — what a region is best for, what the riding is like — are labelled as ours where they appear.

Still right next season

Every fact here is stamped with the day it was checked and how often it is re-checked — a currency guarantee almost no cycling site offers. Past that interval a page shows its age and points you at the source for the current status, and re-checking — re-reading sources, fixing link rot — is core work here. It is why a site built this way can be trusted to age: the numbers do not rot while nobody is looking.

A date on a check answers when did somebody last look — and there is a second question behind it that a check date cannot reach: how long was the thing good for in the first place? An offer runs to a deadline, a timetable to the end of its season, a fares scheme to the end of its year, and every one of those pages goes on sitting there afterwards, unchanged, saying what it always said. So the dates a source puts on itself are read and kept alongside it, and when the day you are planning for falls outside them the page says so beside the figure rather than letting the figure stand. The race is in April 2027; a discount code advertised for bookings made by June 2025 is worth trying and worth knowing the age of, and you get both.

What cannot be kept true from here is never copied into prose that will be wrong in a month. Opening hours, live prices and road status are handed to a live source instead — an official feed, an operator’s own page. And where two authorities genuinely disagree, both are published, dated and named: the Formentor traffic restriction ends on 18 October according to the DGT and 15 October according to the Consell de Mallorca, and you get both — because choosing one silently would be inventing a certainty that does not exist.

How the site makes money

Some links here are affiliate links: if you book a hire car or somewhere to stay through one, we may earn a commission, and it costs you nothing extra. Every one is marked at the point of the link, and the disclosure appears above the first on any page that carries them.

What commission does not touch is worth stating plainly: no ranking, comparison, verdict or difficulty score on this site takes a commercial relationship as an input — the criteria above are the whole of it, computed from the measured record. Where there is no partner in a category, the page says so rather than invent a recommendation. The full terms are on the affiliate disclosure page.

Corrections

Spot a wrong number? Every figure is derived from a script and a dataset, so one correction usually fixes the whole set rather than a single page. Email hello@mallorcavelo.com.