How Blogo reads the pages that already rank

When the title options are generated, Blogo requests a live Google results page for the main keyword in your site’s market. It keeps the top ten ordinary results, the people-also-ask questions shown with them, and the featured answer box if there is one. The top ten is shown to you beside the title options.

Blogo then opens each of those pages itself. Your own site is left out, since it is not a competitor. A page that yields fewer than 150 words of main content (a consent wall, a blocked page, a document viewer) is kept in the list but treated as unread.

It keeps the structure and the numbers, not the article text.

Code groups similar headings across the pages into topics. This step uses no model, so the same set of pages always produces the same result.

A topic that at least two pages and at least 40% of the pages read include is treated as one most ranking pages cover. A topic found on only one or two pages is kept as a thinly covered angle, a place where your page might say something the others do not.

Outline. The model proposing outlines is told that every commonly covered topic must be addressed, in its own words and order, and that each outline needs at least one section built on a thinly covered angle or on knowledge only your business has. The proposal that covers more of the common topics is shown first. If the outline you confirm still leaves one out, the writer is told to cover it inside an existing section.

Length. The suggested range runs from 0.8 to 1.25 times the middle length of the pages read, within fixed lower and upper bounds. It is guidance. Length is not checked, and the writer is told to stop when the evidence is used up.

Questions and FAQ. The questions you are asked are written with the covered topics and the thin angles in view. An FAQ section is included when at least three people-also-ask questions share words with your topic, and it answers those questions as asked.

New information. After writing, Blogo counts sourced figures in the draft that none of the read pages state, plus sentences built on your answers. Fewer than three holds the draft. The count is applied only when at least three ranking pages were read.

The domains of the ranking pages are added to the article’s do-not-link list. Three kinds are left off: recognised authority sites (government and education domains and a fixed list of institutions and publications), sites already approved as sources for this article, and your own site.

You see the list at the writing-setup step and can untick any entry. For the ones left ticked, the writer is told never to link there, addresses on those domains are removed from the research notes before the writer sees them, and a link to one that still appears holds the draft.