How we verify.

The three kinds of claims on this site

Everything published here falls into one of three buckets, and we try to make it obvious which one you're reading:

1. Sourced facts. Statistics, prices, dates, rules, and history. These link to a source — ideally a primary one — and the load-bearing ones carry a date showing when we checked. If a fact can't meet that bar, it doesn't get stated as fact.
2. Desk verification, described exactly. Much of the practical material on this site — which bars are still open, which vending machines still exist, which brewery tours are bookable in English — is verified at the desk: operators' own sites and booking pages, Google Street View, and dated Japanese-language sources. When we've done that, we say that, in those words. A "Last verified" date on a listing means "the date of the most recent source we could confirm" — not a promise that someone stood in front of it this morning.
3. The editor's voice. Opinions, taste calls, and the occasional story. This is the part of the site that sounds like a person, because it is one. It is never dressed up as measurement — when a ranking comes down to an editor's call rather than a published comparison, the article says so.

Where facts come from

In order of preference:

• Official and primary sources first. Government and industry-body statistics (the vending-machine numbers on this site come from the JVMA, the industry association), operators' and makers' own pages (climbing fees from the official Mt. Fuji site, eSIM plans from the providers' own plan and policy pages, brewery hours from the breweries).
• Wire services and established journalism second — Kyodo News, The Japan Times, and the like — cited by name, for claims a primary source doesn't publish.
• Enthusiast databases and dated traveler reports third, used mostly to confirm that something still exists, and always as "a source dated X says", never as settled fact.

Other travel blogs are competitors, not sources. Where a third-party review is the only evidence for something — real-world coverage of a mobile network, say — the claim is attributed and linked, not absorbed as our own.

Numbers we can't verify get removed, not softened

The tempting move with a shaky statistic is to hedge it — "reportedly", "some say" — and keep the impressive number. We don't. In our August 2026 audit, national ekiben variety counts and market-size figures that couldn't be traced to any reliable source were deleted outright, and a Japanese term we could not confirm actually exists was removed from an article rather than kept because it sounded good. Where uncertainty is honest and useful, we label it instead — some smaller-prefecture entries in the ekiben atlas are marked "representative pending verification" rather than presented as settled.

Prices go stale — here's how we handle it

• Prices are given as ranges where the real world varies (¥900–¥1,100), with the year.
• Articles carry an updated date, and price-sensitive pages get re-checked against the official pages when they're revised.
• When a vendor doesn't publish a fixed price sheet — some eSIM packages only show prices at checkout — we say exactly that instead of quoting a number we'd be guessing at.

What we don't claim

We don't claim verification that didn't happen. No "we tested" without a test, no on-site visit claims for desk research, no speed numbers we didn't source. Our eSIM comparisons, for example, are built from the providers' official listings and policy pages plus attributed third-party reviews — the articles say so plainly, because what a plan page commits to in writing (networks, throttles, tethering, refunds) turns out to be most of what a traveler needs to know anyway.

Corrections, publicly

When we find an error, we fix the article and correct the record rather than quietly deleting. Real examples from the August 2026 audit:

• The "one vending machine per 23 people" figure — repeated across the travel internet — turned out to be a peak-era count. Corrected to roughly one per 48 people (about 2.6 million machines) per JVMA data, across every language version of every affected article.
• The Toge no Kamameshi cumulative-sales figure was updated from 170 million to roughly 190 million pots, per a November 2025 Kyodo News report.
• The Mt. Fuji trail fee was corrected from the ¥2,000 introduced in 2024 to the ¥4,000 charged in 2026, per the official climbing site.

Spotted something wrong? Tell us on the contact page. We'll check it against a primary source, fix it, and note the correction. Readers who catch real errors are doing this site a favor.

One boundary worth stating

This page is about how facts are checked. How the site makes money — and why that never buys a recommendation — is documented separately on the Affiliate Disclosure page. The two systems are deliberately kept apart: sources decide what's true; disclosure explains who pays the hosting bill.