Methodology

The score

Every provider gets a single editorial score out of 10. The dataset is the foundation, but it does not calculate the result: the score expresses our whole view of the product, company, deal, risks, execution, and insights that do not always fit into a field. A provider can look like a 10/10 feature checklist and still be a 6/10 choice in practice. Payment never affects the order.

The six investigative lenses

How we assign the number

The editor investigates the structured fields, product behavior, operating history, commercial reality, market feedback, contradictions, and risk signals, then assigns one overall score. The six lenses keep the investigation broad; they are not a formula and do not constrain the conclusion. More populated fields or more advertised features do not automatically produce a higher rating.

What we score, and what we ignore

We score the things that change what you ship and what you pay. We don't score brand buzz, ad budgets, or how slick a sales deck looks. If it doesn't affect the build or the economics, it doesn't move the score.

How each value is structured

Every field in the framework declares what kind of value it holds: a yes or no, one option from a fixed list, a number with its unit, a money amount with its currency and period, a date, a list of items, or written analysis where the answer is genuinely a judgment rather than a value. Each provider value is stored in that shape. A catalogue size is a number rather than a sentence, a fee is an amount in a named currency, and a capability list is a set of comparable items.

Where a value needs a qualifier — the product it covers, the market it applies to, or the boundary that is still open — that wording sits beside the value as a scope note instead of being folded into it. On a review the note appears next to the value; in the comparison workspace it travels with the cell.

Twenty-six of the lists use one controlled vocabulary across all eighteen providers, so the same capability reads the same way in every profile. Where a list names markets, regions or languages we use one spelling for each: the market licensed by the Gambling Commission is Great Britain, US states are spelled out, and languages are named rather than abbreviated. Lists that carry company, product or licence names stay open, because a fixed vocabulary would force real records into the wrong box. That is what makes a side-by-side comparison a comparison rather than a spelling exercise.

How confident we are in each data point

A populated value can carry one of three confidence labels, and the “% verified” readout is our editorial confidence in the conclusions. Missing information has no separate confidence state: we keep investigating until we can publish a defensible value or must leave the field empty.

How we build the picture

We publish one current conclusion for each field. Competing accounts, incomplete evidence, and contradictions are resolved inside our research process rather than replayed in the published result. Each result states the value, its scope, its current status, and any material boundary that remains unresolved. Verified conclusion, supported conclusion, and editorial inference labels describe how strongly we hold the result. They never calculate the provider score.

How we separate permission, testing, and market access

A supplier license belongs to a named legal entity. A product test belongs to a named system or version and a technical standard. An ISO certificate belongs to a defined entity, site, or management-system scope, while PCI DSS, SOC 2, and ISAE evidence has its own assessed subject and period. Market access then binds a provider, product, supplying entity, permission, technical artifact and production deployment to one jurisdiction. We do not let one layer prove the others or let a group-level conclusion pass to every product and affiliate.

The normalized boundaries are published in the provider license register and the product certification register, the provider market-access register and the platform requirements ledger. The requirement ledger defines selected operator duties; it does not assign implementation status to a provider.

What this model doesn't capture

No score is the whole story. We're explicit about the edges:

• It scores the platform and the deal, not the deep internals — we don't independently test sportsbook odds-trading quality, live-dealer latency, or day-to-day UX.
• Most commercial terms are quote-only, so the commercial lens weighs established economics, transparency, and unresolved boundaries — not the number you will ultimately be quoted.
• Scores are our editorial judgement applied consistently, not a lab certification or a paid audit.
• This is a B2B view for operators choosing a platform — it is not a player-facing casino rating.
• The market moves fast; a score reflects the last review, and we re-score when the facts change.

No pay-to-rank

No provider pays to appear, to rank higher, or to change what a review says. We don't sell placements or sponsored slots to the companies we cover. Our Advertising Policy has the full rules.

Corrections, updates, and provider responses

We publish our best current conclusion, not a frozen snapshot. A correction is made when our published record was materially wrong. A material update is made when the underlying reality changed after publication. A clarification improves the wording without changing the conclusion. We do not erase the distinction between them.

Anyone may challenge a fact. Providers may also submit a concise, on-record response. We verify that an official response comes from the provider or an authorized representative, then publish it separately from our analysis when it materially addresses the review. A response is never a paid placement, does not overwrite our wording, and does not change a score by itself. New evidence can change our conclusion; pressure, payment, or disagreement cannot.

Material decisions appear in our Editorial recordwith the date, what changed, why it changed, and any effect on the score. Minor spelling, formatting, and layout changes are not logged. To challenge a record, email us at sup@igamingplatformproviders.com.

Keeping it current

Provider data changes, so we revisit it. This model and the data behind it were last reviewed on September 6, 2026.