Most advice about search rankings skips the part that matters: what a search engine is actually doing when someone types a query. Understanding the pipeline makes the tactics obvious instead of magical.
Three stages, not one
Search works in three distinct stages, and a page can fail at any of them.
Crawling. A bot has to find your URL. It does that by following links from pages it already knows, reading your sitemap, and revisiting pages it has seen before. If nothing links to a page and it is not in your sitemap, it may as well not exist.
Indexing. Once fetched, the page is parsed, rendered, and stored. The engine decides whether the page is worth keeping. Thin pages, near-duplicates, and pages blocked by a noindex tag get dropped here. A page that is crawled is not automatically a page that is indexed.
Ranking. Only indexed pages compete. For a given query, the engine pulls a candidate set and orders it using hundreds of signals combined by machine-learned models.
What the ranking signals are really measuring
The signal list is long, but nearly all of it reduces to three questions:
- Is this page about the query? Handled by text matching, semantic embeddings, entity recognition, and the anchor text of links pointing at the page.
- Is this source trustworthy? Approximated through links from other trusted sites, brand signals, topical consistency, and site-level quality history.
- Will the user have a good experience? Page speed, mobile layout, intrusive interstitials, and whether searchers who click tend to stay or bounce back to the results.
Queries are classified before they are answered
Search engines guess what type of result a query wants. “Running shoes” wants product listings. “How to lace running shoes” wants a tutorial. “Nike” wants a brand homepage. If you publish a 3,000-word guide targeting a query that returns product pages, you will not rank regardless of quality, because you built the wrong format.
The practical test: search your target query and look at the top ten results. That is the format the engine has decided satisfies the query. Match it, then beat it on substance.
Freshness is query-dependent
Some queries have a freshness requirement built in. “Best laptops” needs current models. “How photosynthesis works” does not. Publishing schedules should follow that logic: refresh the pages where staleness costs you, and leave evergreen pages alone unless the facts changed.
What this means in practice
Before chasing tactics, confirm the basics in order. Is the page crawlable? Is it indexed? Does it match the intent behind the query? Is there any reason a search engine would trust your domain on this topic? Most ranking problems are diagnosed by walking down that list, and most of them stop before the third question.