How Arizona Places ranks attractions

Three measurements, kept apart

"Most visited" and "most popular" are different questions. Lake Mead National Recreation Area records more visits than Grand Canyon National Park; the Grand Canyon is the more significant destination. Collapsing those into a single number would lose the distinction a reader came for, so this site measures three things separately and only then combines them.

1. Visitation

How many people are recorded visiting, from official published counts only. National Park Service figures come from the agency's Visitor Use Statistics data package, which is in the public domain and reports monthly recreation visits per unit. If no agency publishes a count, this axis is empty — never estimated, never inferred from anything else.

Counts are compared on a logarithmic scale between 10,000 and 7,000,000 annual visits. A linear scale would compress everything below half a million visitors into the bottom of the range and make it unsortable.

2. Popularity

How much public interest a place attracts, measured as human (non-bot) readership of its English Wikipedia article over a full year. Scaled logarithmically between 1,000 and 600,000 annual pageviews.

What this is not: it is not a visitor count and not a search-volume figure. Google review counts and ratings are deliberately absent — the Google Maps Platform terms prohibit storing them, and a statically built page would have to. Wikimedia pageview data is published under a public-domain dedication and can be stored and cited.

It is specifically encyclopaedic readership — a count of people who looked a place up, which is not the same population as the people who know it. The two diverge most where fame is visual and recent, because a photograph travels without ever sending anyone to an encyclopaedia. This axis is therefore a reliable floor and an unreliable ceiling: a high number is real evidence of public interest, while a low or absent one may mean a place is obscure, or only that its fame does not take the form of reading.

3. Significance

How important a place is as a destination, derived from what it officially is rather than from an editor's impression. Each category carries a baseline, and each official designation adds to it. Both tables are published below, and both are the same data the code runs on.

Combining them

When all three are known, the combined score weights visitation at 45%, popularity at 25% and significance at 30%.

Three rules then keep the result honest:

1. A missing signal is never treated as a zero. Scoring an unmeasured attraction as though it measured zero is estimation in disguise, and it punishes exactly those places whose data is hardest to obtain. Missing axes are dropped and the remaining weights are renormalised.
2. Fame cannot reach the top band. An attraction with no official visitor count is capped at 82 out of 100. The top of a visitation ranking is a claim about measurement, not about renown.
3. Tourism relevance is applied last, to the combined score, so no single axis can route around it.

Why a busy shop does not outrank a national park

This is the failure the weighting exists to prevent, and it is a real one rather than a hypothetical. Measured over 2025, a large Arizona shopping centre drew roughly 15,200 Wikipedia readers against the Desert Botanical Garden's 8,700. On public interest alone, the mall wins.

Rather than hand-correcting individual records, the judgement is made once per category and published: retail carries a tourism relevance of 0.15, against 1.0 for genuine attraction categories. A shopping centre can win every measurable signal and still rank near the bottom, because it is being measured as what it is.

Category baselines and weighting

Designation bonuses

A designation not in this table adds nothing rather than causing an error. The list of things a place can be designated is open-ended, and an unrecognised one should be inert.

Confidence

Every ranked attraction carries a confidence level derived from its provenance, not from judgement:

• High — an official visitor count exists for the attraction itself.
• Medium — the count covers a wider area than the attraction, or there is no count but public interest data exists.
• Low — neither is available.

The "wider area" distinction matters. Glen Canyon National Recreation Area's figure covers 1.25 million acres across two states; presenting it as a measurement of any single overlook inside it would be false, so it is labelled and held to medium confidence.

The current database

• 724 attractions.
• 48 with an official published visitor count.
• 676 for which no official count is published at all.
• 449 with public interest data.

Known limitations

These are the things this ranking cannot do, stated rather than buried.

• Pageviews measure fame, not travel intent. People read about the Hoover Dam far more than they read about the Desert Botanical Garden, and visit the garden without reading about it at all. This is why popularity is weighted below visitation and never substitutes for it.
• The popularity axis underrates socially famous places. Because it counts encyclopaedia readers, it sees a place in proportion to how much has been written about it rather than how many people have seen it. Devil's Bridge is the clearest case in this database: it is among the most photographed hikes in Arizona, it draws crowds large enough that Sedona manages access to it, and it has no English Wikipedia article at all — so this axis records nothing for it and it is ranked on significance alone. It is in the database only because Wikivoyage indexes what travellers look for rather than what is encyclopaedically notable. 275 of 724 attractions here have no readership figure, and that number should be read as a limit on the measure rather than a verdict on the places.
• Federal attractions are over-measured relative to private ones. Agencies publish counts; private museums and tribal attractions mostly do not. A high-confidence ranking is therefore skewed toward federal land, and the confidence label is how that shows.
• Recreation visits are not unique visitors. The National Park Service counts an entry, so one person entering a park three times in a day counts three times.
• A missing signal scores better than a weak one. Because missing axes are dropped and the remaining weights renormalised, an attraction with no public-interest data is scored on visitation and significance alone, while one with genuinely low public interest carries that low number into its average. The alternative would be to invent a number for the missing case, which this site does not do, so the asymmetry is disclosed rather than papered over.
• An attraction with nothing measured is scaled down, not ranked on category alone. 272 of the 724 attractions here have neither an official visitor count nor a readership figure. Their combined score would otherwise be the significance baseline of their category and nothing else — a number derived entirely from what the place is, with no observation behind it. Because measured attractions score a median of 44.3 on the combined scale, and category baselines run from 45 to 62, every one of those records would outrank the median measured one purely for lacking data. Their scores are therefore multiplied by 0.44, which keeps the ordering they genuinely have — a national monument really is more significant than a roadside marker — while placing the set below the median measured score. This hides nothing; it stops an absence of evidence from reading as evidence.
• Some published years are unusable. Eleven Arizona park units report zero recreation visits for October 2025 — a reporting gap during a federal funding lapse, not a measurement. Those units use 2024, their last complete year, and their pages say so.

Sources

• Visitation (federal): National Park Service Visitor Use Statistics data package, public domain, monthly recreation visits 1979–2025.
• Visitation (state): Arizona Office of Tourism Monthly State Parks Visitation Report, from Arizona State Parks and compiled by the Arizona Hospitality Research and Resource Center at Northern Arizona University. Transcribed by hand from the primary report, which is published as a PDF and cannot be fetched automatically.
• Public interest: Wikimedia pageviews API, human traffic only, public-domain dedication.
• Coordinates, official websites and designations: Wikidata, public-domain dedication.
• County assignment: U.S. Census geography, resolved from each attraction's coordinates.

Each attraction page lists the specific sources behind its own figures, with the date the record was last verified.