Sabine Heller

I take a result page and work out what it is telling you before looking at any of the listed pages. The type of page that ranks is the most useful signal on the page, and it is read from format rather than from wording. A query wanting a definition gets a definition. A query naming a tool gets a tool page. A query with a version number gets documentation. If the pages ranking are a different format from yours, the ranking problem is upstream of anything you can do to the text. The features take up space and change what the snippet looks like. A featured snippet removes the click entirely, which makes ranking first less valuable for those queries and worth knowing before you optimise for them. Other features push the organic results down the page. Localisation changes the whole picture, so I cover reading a result page for the country you care about rather than your own. The variability of the page matters too. The same query returns different results for different people and different days, so a single observation is an anecdote. I say how many observations a claim needs. My practical rule is that a page format has to match the intent of the query before it competes. Matching the wording of a competitor whose format is wrong is a way to lose to a competitor whose format is right. I keep a note of what the page looked like on a given date, because a claim about who ranks is only meaningful alongside when it was observed and for which country and language.