Method · Evidence standard
How We Verify Live Casino Information and Operator Facts
This site publishes what it can document and marks the rest as unknown. That sounds simple, and most of the work is in the details: which sources count, what a verification state actually claims, when a fact was read, and where the limits of the method sit.
This page sets all of that out in one place. It is a description of how the site is built, not a rating system — nothing here produces a score, and the site does not rank operators.
Eligibility: an active UKGC licence record
An operator appears on this site only if the UK Gambling Commission's public register shows an active remote casino licence for it. That is a condition of being covered at all, and it is deliberately not worth anything beyond that.
A licence tells you that a company is regulated in Great Britain, which company is accountable, and when the current licence record began. It does not tell you that the operator is good, better than another, or recommended by us. Since every operator we cover holds one, it cannot separate them either — it is the entry condition, not a mark of quality. Where an operator's licence record ends, the operator leaves the list rather than sitting there with a stale badge.
Which operators clear that condition is published rather than asserted. The operator directory lists every licence record read for this site, each entry naming its licensee and its licence number, so the same check can be run against the register by anyone who wants to run it.
Where the facts come from
Four tiers, in order of preference. A fact is recorded at the highest tier that actually establishes it, and we do not claim a tier we have not used.
- 1. The regulator
- The UK Gambling Commission publishes its public register as a daily download: licensees, licence numbers, registered domains, licence dates and domain status. Every operator fact on this site is read from that register, and every operator entry links to its record so the reading can be checked. A register pass reads more licence records than the site ends up publishing — a brand is only listed here once its record has been read and matched to the licensee behind it, so the pass is always wider than the index it produces. Applied to every operator we publish — register pass of 21 August 2026
- 2. The operator’s own published pages
- A statement an operator publishes about itself — its terms, its help pages, a description of where a table is streamed from. Used where the operator is the origin of the fact rather than a party repeating it. This is the source behind the single UK venue-stream entry on the homepage.
- 3. Official game and platform documentation
- Where a fact belongs to a game rather than to an operator — the rules of a table, the structure of a format — the document published by whoever produces it is the source. We consult this tier, and to date it has not produced a per-operator fact we could publish, because studios do not publish which sites carry which tables.
- 4. Other primary sources
- Anything else only where the source is the origin of the fact, not a report of it. A search result that describes a page is not that page: it tells us where to look and never counts as the evidence itself.
The three source states
Every fact we hold carries one of three states behind the scenes, and the state decides whether it is published at all. They describe the evidence, never the operator and never the product.
- Source verified — on a stated date
- The specific fact shown has a traceable source and the date we read it, and the page shows that date next to the fact rather than in a footnote. It means exactly that and nothing more: not that the operator is good, not that the table is running now, not that anything was tested, and not that the source has been watched since.
- Not verified
- We do not currently hold adequate evidence for the claim. The claim may well be true. We have simply not read a source that establishes it, so we decline to state it as fact.
- Not published
- The information is not published anywhere we can point to — most often because nobody has published it. The field stays empty of claims rather than being filled with an estimate.
- Source verified — on a stated date
- Not verified
- Not published
Why unknown is not the same as nothing
This is the rule the rest of the method depends on. An empty field, a crossed-out icon or a zero all read as a statement about the operator: it does not offer this, it has none of those, it scores badly here. None of that is what our missing data means.
Not verified and not published are statements about us. They record that we have not read a source, and nothing whatsoever about whether the thing exists. A table we have not documented is not a table that does not run. An operator whose payment page we cannot read is not an operator with poor payments. So where a fact sits in either state, we do not render it at all — no blank cell, no dash, no greyed-out marker that a reader could take for an absence. The question is left where it belongs, which is open, and the pages say in plain words which questions those are.
Dates, and what happens when we look again
Every fact carries the date it was read, not the date the page was published. The operator entries on this site were read from the register on 21 August 2026, and that date is shown with them. When a fact is re-read the date moves; until then it stands as what was true when we looked.
The Commission republishes the register as a structured download on a daily rhythm. The licence values here come from one dated snapshot of that download, and they move when the snapshot is taken again — a comparison run against the register as it stands that day. What follows depends on what changed: where the register has only revised a licence number for the same licensee on the same account, the operator layer is brought forward as a whole and the date on every entry moves with it; where the difference touches who is accountable for what — a different licensee, an account that no longer carries a brand's domain, a licence that is no longer active — the entry stays as it is until a person has read the record. Anything taken from an operator's own pages is re-read when we revisit that entry. The date shown is the date of the snapshot behind it: a reading taken then, and the record stands on that reading until the next one is taken.
What happens when sources disagree or disappear
Everything above describes the ordinary case, where a source exists and says one thing. These are the awkward cases, and how each one is settled.
When two sources conflict, the more authoritative and more current primary source wins, and the older value is replaced rather than defended. A figure does not earn standing by having been published here first. Where an operator's own page and the register disagree about the company behind a brand, the register is the authority on that question and the operator's page is a claim about it.
A fact can also travel backwards. If something already documented can no longer be verified — the page is gone, the wording has changed, the figure has quietly moved — it does not stay here as current truth on the strength of having been true once. It returns to not verified, or to not published, and the site stops stating it. These states are not a ladder that only goes up.
Not being able to look is not the same as there being nothing to see. An operator page may be unreachable because of a security filter, because of where the request came from, or because it needs an account first. None of that is evidence about the product, and "we could not read it" never becomes "the operator does not offer it".
A page answering successfully is not evidence either. An address can return a perfectly healthy response and still contain nothing that bears on the question: a frame that loads before its content arrives, a holding page that answers every address on a domain the same way, an address that no longer exists but responds as though it does. So the test is never whether a source loaded. It is whether the thing we went there to read is actually in it, identifiable, and about the operator or the game in question.
A search result is a way to find a candidate page, never the finding itself. Where the page behind it cannot then be read, the snippet does not stand in for it. And a date here records one event — a reading taken on a day — which is the whole of what it claims and the reason it is printed beside the fact rather than in a footnote.
The life of a fact on this site
Every line published here has been through the same five steps, and can go back through them.
The order matters more than it looks. A claim is proposed first and the source is sought afterwards, never the reverse — which means a statement that cannot find a source does not quietly become a softer statement, it becomes a marked state. Writing the sentence first and hunting for something that supports it is how unsupported claims get published everywhere else.
The dating step is what keeps the system honest over time. A fact is not true here in the abstract; it was read on a particular day from a particular document. If the source changes tomorrow, the line does not become a lie — it becomes an old reading, visibly dated, until somebody reads it again.
Nothing in the loop upgrades itself. A not-verified line does not drift into verified because it has sat on the page a while, and only a fresh reading moves anything. It is also why the site publishes so little compared with its subject: most interesting questions about live casino products stop at the second step, because no source answers them.
Regulatory facts and product facts are not the same thing
A regulatory fact has one authoritative publisher, a defined format and a fixed address. The UK Gambling Commission publishes its register as a structured download; there is exactly one correct answer to "who holds this licence", and anybody can check it in the same place we did. Facts of that kind reach the verified state consistently, which is why the operator layer of this site is the part that looks complete.
A product fact has none of those properties. Which live tables a site carries today is published by nobody in a durable form, changes without announcement, and is often visible only behind an account. Two people asking the same question can get different answers on the same day, and neither can point at a document.
So they are handled differently here. Regulatory facts are stated with their source and their date. Product facts are stated only where a document supports them, and marked unknown where none does — which is currently most of the time. The visible asymmetry between a fully sourced operator layer and the absence of any table-availability list is not an unfinished site; it is the method showing its work.
The format pages sit on the first side of that line. How a wheel is spun and called, how a dealt round resolves, which bets a table offers and how a presented round is structured are properties of the games themselves — documented, stable, and readable without an account. They are set out on live roulette, live blackjack, live baccarat and live game shows. Which of those tables a given operator has running this afternoon is the product fact, and it is the one left unknown.
The register establishes
- the current remote casino licence record, and whether it is active
- the licensee — the company legally accountable for it
- the licence number
- the brand-to-domain relationship, where the register supports it
- a direct or white-label arrangement, where the register supports it
The register does not establish
- which live tables an operator carries
- which studios or providers it carries
- how many games it offers
- whether a given table is available now
- stake limits
- seat availability
- stream quality
- dealer quality
- payout speed
- what the site is like to use
If a page presents a licence number and a live-table count with equal confidence, one of the two has not been checked.
What our evidence covers, and where it stops
The method is documentary: it reads published material rather than exercising the product. That boundary decides which questions the published record answers, so it is stated in the same place as the sources themselves.
Outside the current dataset
- deposit handling
- withdrawal performance and payout speed
- identity-check duration
- customer support quality
- dealer conduct at a table
- stream latency and uptime
- stream picture quality
- seat availability at a given hour
- stake limits, which vary by table and by player
- the complete live-table catalogue at any site
- lobby contents across the market
Every field in that list needs evidence of a different kind: an account with money moving through it, or a lobby observed over time. Our published operator evidence comes from documentary sources — the Gambling Commission register, the GAMSTOP participating-companies list, platform listings and operators' own published pages — and what those establish is who is accountable for a site, under which licence, since when.
So the published record is documentary evidence of the fields it shows, and it should be read that way. How fast a withdrawal really clears, how a dealer handles a table, whether a stream holds up on a Saturday night: each sits on the far side of that boundary, stays in an open state on the page, and would need its own evidence before it could be published.
The trade is deliberate. A documentary method gives up first-hand experience and gets back something a reader can check: every fact points at a source, and anyone who disagrees can open it.
Why there is no ranking
A ranking is a claim that operators have been compared. To make that claim fairly you need the same information about a large majority of them, gathered the same way, from sources anyone can revisit. We do not have that.
What we can read consistently is regulatory: licences, licensees, domains, dates. That is uniform across the operators we cover, and precisely because it is uniform it cannot rank anybody. What would differentiate operators — which live tables and studios each one carries, what its live categories look like — is not publicly readable in a consistent, repeatable way. The gap is not small, and it does not close by trying harder.
So the operator layer is listed alphabetically and carries the facts we can source. We would rather publish a short, checkable set of facts than a number that implies a comparison nobody performed.
What the register does separate is structure, and structure is worth reading rather than scoring. Coral Casino and Ladbrokes Casino are recorded under one licence number held by one licensee, and appear here as two brands on a single record. That is a statement about who is accountable, not a judgement of either brand, and it is the kind of detail a numbered list has nowhere to put.
What would have to be true first
If a scoring system is ever introduced here, these are the conditions it would have to meet. None of them is met today, and none of this is scheduled.
- the same information becomes readable for a large majority of the operators covered, not a handful
- the criteria and their weights are fixed and published before any operator is looked at
- the rubric is public, so a reader can see what separates one band from the next
- the evidence behind each value is shown next to the value
- unknown stays unknown and never becomes a deduction dressed up as a measurement
- no commercial arrangement can change a value or an order
Until then the site does what it does now: it records documented facts, dates them, names their sources, and stays silent where it has nothing to point at.
Common questions about our method
Does an active UKGC licence mean you recommend an operator?
No. A licence record establishes that an operator is regulated in Great Britain and identifies the company legally accountable for it. It is a matter of eligibility, not of quality, and it says nothing about how good a site is to use. Every operator we list holds one, so it also cannot distinguish between them.
Why does the site not list which tables each operator carries?
Because that kind of statement cannot currently be sourced to the standard set out on this page. Listing a table against a site means stating which category and format it belongs to, which studio produces it and where that information came from. We can read licence records consistently; we cannot read table-level availability consistently, and it changes without notice. So the site explains the formats, which are stable and published, and does not publish an availability list it would have to invent.
Do you treat an operator’s marketing as a source?
Only as a claim, never as an observation. If an operator states a figure about its own product, that is the operator saying it, and it is recorded as such if it is recorded at all. A number repeated from a promotional page does not become a fact by being quoted here.
Will you ever score or rank operators?
Only if the evidence changes enough to make a fair comparison possible, and there is no timetable for that. The conditions we would hold ourselves to are set out above. Until every one of them is met, publishing a score would mean inventing the difference between operators rather than measuring it.