
In recent years Google have been moving away from keywords towards semantic search and question answering:

Many of the old keyword tricks are ineffective. The semantic meaning of content is now key. But how can webmasters understand whether their content is semantically similar to a search query? (in Google's eyes)
We know, from published material that behind the scenes Google are using a form of AI known as T5 (actually MUM which is derived from T5). We also use T5 for our platform so...
What if we enable webmasters to test a query against their content using T5? “See your content, as Google sees it”. Webmasters would be able to evaluate how semantically close their content is to a given query.
Of course we don’t actually know the ranking signals Google use, just the underlying technology. Nevertheless I think it has to be an improvement over all the keyword analysis tools still out there.
Thoughts?
It sounds very interesting .. and also very complex!