Detecting Co-Occurrence Confusion As Google BERT and its equivalents increasingly impacts understanding in queries and content, and whilst optimising for ‘clarifying algorithms’ is not necessarily within the grasp of organisations, seeing how search engines are understanding and disambiguating queries and ‘word’s meaning’, in SERPs and rankings utilising natural language understanding and co-occurrence does provide us with some strong clues as to areas where conceptual misinterpretation is occurring. This can provide us with hints which allow us to improve consistent clarification signals within a site (e.g. wrong pages ranking from within a property or site section). We can look at the factors which could be impacting wrong page ranking and steer conceptual understanding via crawl and word sense disambiguation and clustering. Often the conclusion is drawn that there is not ‘enough authority’ for x page, but it is not necessarily the correct conclusion to draw. It is often just legacy cruft and noise which is clouding understanding when combined with concepts and semantics. We’ll discuss what might be happening here and what brands can do about it to cut through the exhaust data generated over time to make the right targets crystal clear.