How we check financial information (South Africa)

Three kinds of information, three different rules

Pages on this site mix three kinds of information, and each is handled differently. Broker terms (minimum deposits, leverage, dormant-account rules, competition conditions) are read from XM's own pages and its client agreement. Reader experience comes from published reviews, sampled by country. Research comes from academic papers on retail trading. A reader can tell which is which from the way it is written: terms carry a source and date, experiences carry a platform and a date, research carries a paper reference and an evidence level.

Broker terms: source, date, clause

A term is recorded with the exact page or document it came from, the date it was read, and, for the client agreement, the clause number. That record lives in the site's facts registry and is printed on the page in the facts strip and in the “Sources and verification dates” tables. When two pages of the broker disagree, the page says which one was used and why. When a page could not be reached, the term is marked as not captured instead of being copied from a third party.

Terms change. The check date tells the reader how old the reading is; the quarterly re-check and reader reports trigger a new reading.

Reader experience: how reviews are sampled

Review quotes come from a platform's public reviews filtered by the reviewer's country, collected on a stated date, and recorded as a sample with its size and star split. The site prints the size and the split so a reader can see how negative the sample is before reading the quotes. Quotes are chosen to show the recurring themes of the sample, not the best or worst review, and the ratings shown next to quotes roughly follow the sample's mix.

Counts such as “how many reviews mention withdrawals” are counts within that sample only. They are not a rate for all XM clients in South Africa, and the pages say so.

When a page says how many reviews mention something, the number comes from a keyword search over the full text of that sample, not from a reading impression. One review counts once per theme, and the word lists are these: login or access trouble: “log in”, “login”, “logging in”, “log into”, “failing to log”, “unable to log”, “members area”, “password”, “sign in”, “something went wrong”, “access my account”, “access to my account”, “can't access”, “cannot access”, “app is not working”, “application has not been working”; blocked or restricted account: “block”, “restrict”, “ban”, “banned”, “disabled”, “archived”, “suspend”, “terminated”, “red flag”, “closure”, “closed my account”, “frozen”; withdrawals: “withdraw”, “widraw”, “widawal”, “withdrawl”, “payout”, “paid out”, “cash out”; support responsiveness: “support”, “customer service”, “customer experience”, “live chat”, “no reply”, “no response”, “not respond”, “responsiveness”, “account manager”, “emails”; payment methods: “payment method”, “payment option”, “payment infrastructure”, “payment experience”, “ozow”, “skrill”, “card-based”, “bank transfer”, “binance pay”, “google pay”, “upi”, “deposit method”, “crypto”. A count like this describes the wording of the sample, not what every client experienced, and a review that describes a problem in other words is missed by design. If a number on any page does not match this method, the number is wrong and worth reporting.

Research: quoted with its limits

When a page cites a study, it names the paper, states the evidence level assigned to it during research, and states how far the finding transfers to XM and to the leverage available in South Africa. A study on other markets or at much lower leverage is quoted as a caution, never as a measurement of this broker. Papers without a clear mechanism or practical implication are not quoted at all.

Related XM pages