How we verify

This site publishes a narrow set of facts and tries very hard not to publish anything else. This page explains where the facts come from, so you can decide how much weight to give them.

What we do

We query Apple’s public store lookup endpoint once per App Store storefront, for each app we track:

https://itunes.apple.com/lookup?id=<appleId>&country=<storefront>

An app counts as available in a storefront when that request returns a result for that country. The response is also where every specification on this site comes from: version, download size, minimum iOS, languages, age rating and advisories, seller of record, bundle identifier, release and update dates, and the store’s own rating.

Three outcomes are kept apart, because they mean different things:

• available — the app is listed there.
• absent — the storefront exists and the app is not in it.
• not a storefront — Apple rejects the country code outright. Several ISO country codes are not App Store storefronts at all. These are excluded from the denominator rather than counted as absences, which would inflate it.

A fourth outcome, could not check, covers requests that failed after retries. Any count we publish states how many of these there were, even when the answer is zero.

The whole procedure is one script, scripts/fetch-app-snapshot.mjs, in this site’s repository. The storefront list, the retry behaviour and the definition of “available” are all in it and can be read. Its output is committed as src/data/app-store-snapshot.json, and every run is additionally kept under its date in src/data/snapshots/, so the series can be compared over time rather than only asserted.

Nothing from that file is retyped into an article. The pages read it directly. A number that is never typed by a person cannot be mistyped by one.

What we do not do

We do not install these apps. Both apps we currently track are restricted to a single storefront each — one in Nigeria, one in Tanzania — and we are in neither. We have not created accounts, deposited money, placed bets or requested withdrawals. Any site that tells you how an app “feels to use” while covering apps it cannot legally reach is describing something it has not seen. We would rather have a shorter page.

We do not score apps out of five. We publish Apple’s store rating with the number of ratings next to it, and where that number is small we say so. An average over eight votes is arithmetic, not evidence.

How we score

Each app page carries a four-axis assessment out of 10. These are our opinions, set by hand. They are not measurements, and they are not the App Store’s rating.

• Availability — in how many of the storefronts we queried the app is actually listed. This is the one axis that follows directly from the data, and for both current apps it is low because each is listed in one storefront out of 183.
• Build — what the store’s own metadata suggests about the app as a piece of software: the declared minimum iOS, whether there is an iPad build, whether it is localised, and whether it is still being updated. We have not run either app, so this axis cannot speak to how it feels to use.
• Transparency — how much we could establish about who publishes and operates it. An App Store seller that appears by name in a national gaming regulator’s register scores high here. A seller we could not connect to the brand at all scores low.
• Usefulness — whether a reader can do anything with what we found, or whether the honest answer is “this is not available to you and we cannot tell you much about it”.

The headline number on each page is the plain average of those four. It is computed, not written, so it cannot drift away from the axes it claims to summarise. The bars beside each axis are a reading aid — every value is printed as a number too, and they are all the same colour, because a red bar on a page about gambling would read as a safety warning and we are not issuing one.

We do not publish device compatibility lists. Apple returns a supportedDevices array, but for both apps we track it is the same 127-entry list, and it includes hardware that cannot run the minimum iOS the same response declares. It is boilerplate. We publish the declared minimum iOS version and, where we can map it reliably, the oldest iPhone that runs it.

We do not explain facts we have only measured. One app is 1.9 MB and the other 93.5 MB. We report both figures. We do not tell you what that implies about how either app is built, because we have not opened either one.

We do not resolve contradictions by choosing. Where sources disagree about an operator’s licence or corporate identity — and they do — we publish the disagreement and name each source.

Registers we check, and one mistake we made checking them

Where a licence is claimed, we go to the regulator’s own register rather than to another review site. So far that means the Curaçao Gaming Authority’s online gaming licence registry, the Anjouan Gaming Authority’s public register, the Gaming Board of Tanzania’s operator lists, and state-level Nigerian regulators’ published lists.

A correction worth recording, because it changed a published claim: an earlier draft of this site said the Curaçao register was returning HTTP 403 and could not be read. That was our fetching tool being blocked, not the register being unavailable. Requested normally, cert.cga.cw returns HTTP 200 and the registry PDF downloads fine. We had reported a limitation of our own equipment as a fact about someone else’s website. It has been fixed and the pages that relied on it were rewritten.

What we could not establish

Honest gaps, as of 6 August 2026:

• Several operator-owned domains do not resolve on our network, which runs a DNS filter that blocks gambling domains. Where that is the case we rely on regulators’ registers rather than the operator’s own account of itself — which is the better source anyway.
• We could not confirm company registration numbers in national company registries, only in gaming registers.
• We could not establish whether the App Store seller of record for 1win holds any Nigerian gaming licence, because it appears on neither the licensed nor the unlicensed list we checked.

Where any of this changes, the affected page changes with it, and its date changes too.

Corrections

If something here is wrong, it is worth more to us to fix it than to defend it. Every claim on this site carries either a source you can open or a line in a data file you can regenerate, which is the point — you should not have to take our word for any of it.