Rankwell

The context

A company with a hundred and forty blog posts and a single marketing person doesn't have a writing problem: it has a choosing problem. There are eight hundred queries the site already ranks eleventh for and nobody looks at them, because finding them means exporting Search Console, crossing it with what's published and opening the search engine ten times per topic. And on the other side is the temptation to solve it by publishing forty generated posts a month, which is how a site gets ruined.

What the system does

• Topics come from Search Console and not from a keyword database: they're the queries this site already has impressions for, ranked by the clicks to be won and not by search volume
• Sometimes it says no, and explains why: page one is all government bodies, the intent is transactional, or two of your own pages are already fighting over that query
• It prioritizes improving a page sitting at position 9 over writing a new post, because it pays more and costs a fraction
• The brief is the deliverable: the top ten results, what each one covers and —the part that matters— what none of them covers
• Every claim in the draft carries its source, and the ones without it are underlined in amber: reviewing stops being reading fifteen hundred words and becomes looking at two
• It detects cannibalization between your own pages, which is the side effect of the tool itself once you go from three posts a month to eleven
• It never publishes on its own. Human approval and the flagging of unsourced claims are locked rules that can't be switched off
• The cost of each run sits next to what it did: forty-two cents for a draft, and fourteen dollars seventy for the month