November 10, 2026
Two pages using the words Fortune Gems need not describe the same software. Adding Playtime to a search does not resolve that ambiguity or establish an operator catalogue. Comparing versions should therefore begin with identification, not with claims that one version pays better. A matching thumbnail is weak evidence because artwork can be reused, copied, or shown beside an unrelated description.
Build an identification record
Write down the exact displayed title, the developer or publisher if explicitly named, a version identifier if available, the date of the information, and the source address. Leave a field empty when it cannot be established. A missing developer name should not be filled from the first search result that happens to use similar colours.
Next compare the actual rules. Useful fields include the unit used for awards, the meaning of a stake, qualifying symbol arrangements, feature conditions, and limits on awards. These are comparison categories, not claims that a particular title contains any given mechanism. A discrepancy in one field is a reason to investigate, even when every screenshot looks familiar.
Do not compare unlike measurements
Suppose one hypothetical help page expresses an award as twenty credits while another says twenty times the total stake. Those statements are not equivalent until the credit value and stake basis are known. Likewise, a reported return percentage does not directly describe the chance of a feature or the size of a typical short session’s loss.
The British Gambling Commission technical standards provide a primary regulatory reference for game information within the UK framework. They do not verify the identity, rules, or approval of a product found through this search. Use them to understand why precise descriptions matter, not as an endorsement.
Resolve conflicts without guessing
If a review and a help page disagree, preserve both statements with their dates. An old review can be accurate for its own version and still be unsuitable for the version currently described elsewhere. An operator’s label can also be incomplete. Seek a clear answer from an independently authenticated source rather than deciding that the more generous statement must be correct.
A finished comparison can legitimately conclude that the versions cannot be matched. That is more informative than a ranking built from incompatible numbers. Avoid testing the uncertainty with real spending: a handful of results cannot identify hidden probabilities or prove that two implementations use the same mathematics. Identification, stated rules, and unresolved differences belong in separate columns.