We regularly update our data depending on the country and the available public data. Most western countries are updated once a year. Also, the distribution of names across genders may change over the years. For example, if a name was used exclusively by women in the 1970s, it may be used predominantly by male people decades later. That's why we also use birth data for several countries, such as the US, to enrich the results. Our models try to take all these factors into account as well as possible. The returned accuracy value is regularly adjusted to reflect the actual usage in the population as accurate as possible. The value is supposed to provide a reasonable evaluation based on the last decades and may expresses a possible fluctuation. We do look at about the last 4 decades of the data. If you run a query and repeat the same query a year later, you may get a different result because our system might have dropped old data, and new data might have been added. Every time new data is added, we perform so-called integration tests afterward. In this process, our API is tested against a previously validated data set to ensure quality. This data set is continuously expanded with names and test cases to achieve the broadest coverage possible.