You linked website states also that
But, if you're pleased with your score, you might want to consider taking a properly administered and supervised IQ test.
A formal IQ test is supervised by a trained psychologist who will not only keep track of your performance, but also how you reach that goal. There are many aspects of an IQ test, including working memory, analytical skills, logical thinking, but also verbal skills. Verbal skills can for instance be asked with open questions and the way people answer will affect outcomes of the test. For instance a 'describe the following words' test; people that will give examples of the words will automatically score less well than those that can give accurate descriptions not only consisting of examples of the word at hand, but will include unifying, overarching definitions or synonyms. Computers cannot really do this and cannot replace a trained psychologist.
Also, people may have questions on a test and without having that particular question answered prior to starting the test, they may be distracted from the test itself, not knowing what is exactly the purpose of the task at hand. Computers cannot easily replace a human being that can listen and interpret the questions. The questions themselves may also hold implications.
The psychologist will also assess whether the participant is nervous, anxious, tired, easily distracted or otherwise has a handicap that will interfere with performance. Computers cannot assess this easily, especially not online.
For these reasons formal, guided IQ tests are performed with a psychologist. When the averaged IQ outcome is not a representative reflection of the various sub-IQs (analytical skill set, verbal skill set, processing speed, general knowledge etc. etc.) this can be due to various reasons, one of them being widely varying scores on different tests. For instance, when some of the tests are being performed excellent/near perfect, while others are way below average, i.e., there are huge discrepancies being the various sub-IQ outcomes, it is a red flag and can point toward underlying reasons those underperformed tests failed to meet up with the others, and are reason for further/deeper follow-up psychological testing as too the underlying reasons for the discrepancies. in other words, overall subscores generally agree to an extent; if they don't, the overall IQ outcome may not be valid. A simple online test will simply average the sub scores and present the average. A formal test can easily take the psychologist weeks to assess and interpret.
The Wikipedia page for instance says:
Outside influences such as low motivation or high anxiety can occasionally lower a person's IQ test score. For individuals with very low scores, the 95% confidence interval may be greater than 40 points, potentially complicating the accuracy of diagnoses of intellectual disability.