t9xt.com for a AI text detector - might find cool at first but search engines heavily penalised us for it. but why? most search engines use a modified version of Damerau–Levenshtein algorithm. This is super useful for spelling errors, but also misclassifies some words that are intentionally spelt differently - penalising creativity and pushing for "standardization". We faced this on day 1, and day 90. Here's what happened: At first we thought it's because our domain is new, (we shifted over from slopyard.com) which had a good DR but was about to expire - we wanted a brand more meaningful. but (after buying another new domain to test our hypothesis, we became curious) what was the search engine doing? It was auto-inferring t9xt as text (as intended) but since we were t9xt and not text, we kept being skipped from search. Ok that's fine, it's just a new domain, what happened next? So search engines use a framework for "entity-recognition" under the hood, this takes input queries, extracts entities - matches the relevant entity from its index based on context, and redirects the search filters to return more relevant results to use (this is really useful) - but, not for us. By first auto-correct all queries "t9xt" as "text", the entity recognition engine skipped us completely. In fact, at first - even searching by domain would return a result of there is no active website by this name (AI). A clear mismatch with the index. A second reason we hypothesised, but now rely on it's own AI search tool's answer -> when documents were first OCRd, many OCR tools would mistake e and 9, leading to breakout pattern -> prompting the t9xt to text almost always.

The lesson we learnt here, cool is great - but no point if you're not going to be indexed properly. Thankfully there are many other ways of correcting this, but we wanted to share this little piece as a caveat before buying a cool domain that may have unintended interpretations.

Taken from https://t9xt.com/writings/why-search-engines-penalised-us-for-having-a-cool-domain