Methodology
Last updated 3 August 2026
These studies are published so that they can be checked. This page explains where the data comes from, how figures are verified, and what we will not publish, no matter how interesting it would be.
Where the data comes from
Three distinct datasets sit behind this section, and every study says which one it uses.
The structural dataset is a maintained audit of online casino brands: their licences, bonus terms, payment providers and game suppliers. Our editorial team compiles and checks it as part of reviewing operators, and it covers brands whether or not they are commercially relevant to us. Studies built on it are censuses rather than samples.
The observed-demand dataset records how visitors to our comparison sites behave: which operators they look at, compare and click through to, by market and by date. It reaches back to January 2017. This dataset is powerful but partial, and its main limitation is described below.
The tracking dataset records the automated confirmations sent when a player registration completes — timings only, with no commercial quantity attached. It carries a caveat the other two do not: it is analysed separately rather than through the process described below, so its figures cannot be checked back against source in the same way. Any study drawing on it says so plainly in its own source note.
What the demand data is, and is not
Demand observed through comparison sites is not a national market. It over-represents people who actively shop between operators and under-represents those who go straight to a brand they already know. Shares reported from it are shares of a particular slice of demand and should not be read as national channelisation rates.
It also carries an observer problem worth stating plainly: the set of sites contributing observations changes over time, and a site opening or closing in a market moves a measured share without anything happening in that market. Studies using this dataset hold the contributing panel fixed across the window they measure, and report what that correction is worth. It is also why we do not publish multi-year trend lines from this source — the correction is small inside a tight window and large across a long one.
Standing rules
Each study restates the ones that bind it most tightly.
- Every published figure is re-queried before publication: Numbers that arrive from an intermediate step (a summary, a note, an earlier draft) are not published. They are re-run against the source data first. This has repeatedly changed published figures and, on two occasions, reversed a finding entirely.
- Every figure is traceable back to how it was produced: The classification rules, exclusions, and judgement calls behind each study are recorded when it is built and described on the study itself. We will supply the underlying tables on request. Where a figure could not be reproduced, we cut it rather than publish it.
- Charts use fixed scales, not data-driven ones: Axis ceilings are set deliberately and stated in the caption. A ceiling derived from the data would make identical findings look different between rebuilds, and would let a chart exaggerate a change without anyone choosing to.
- Every chart ships the table behind it: No figure appears without the underlying numbers being available on the same page. If a chart cannot be checked, it should not be persuasive.
- Limitations are stated at full length: Each study carries a section describing what the measure does not capture, which results are unresolved, and what would have to be true for a finding to be wrong. Results that survive only one specification are labelled as such.
- Null results are published: Findings of no effect are reported with the same prominence as findings of an effect. A body of research that only ever finds something is not measuring; it is selecting.
What we do not publish
These studies are industry research, not a report on our own business. Anything that would disclose commercial performance is excluded by rule rather than by judgement:
- No monetary totals, revenue figures or commercial terms.
- No traffic, click or deposit volumes. Every demand figure is published as a share or as a difference between two shares.
- No named casino brand or operator in any performance context.
- No breakdown by individual comparison site or commercial account.
Third-party suppliers (game studios, payment providers, licensing jurisdictions) are named where the finding is not checkable without naming them, and where their presence is publicly observable by visiting the operators concerned.
Corrections
Published figures are corrected when they are found to be wrong, and the correction is logged rather than made silently. The log is at /research/corrections.
Using this research
Figures, charts and tables may be reproduced with attribution and a link to the study they come from. Permission is granted in advance and does not need to be requested. The full terms, what we can supply on request, and how to reach the analyst for comment are on the media page.
Independence
This research is published by a commercial casino comparison site that earns commission from some of the operators it covers. That conflict is real, and it is set out in full — along with the findings we have published against our own interest — in about this research. Our lead analyst, Antti Virtanen, leads the work and is responsible for the method and every published figure.