Entity salience
How central a thing is to a piece of text, scored from 0 to 1. One of the very few genuinely measurable terms in this vocabulary.
Checkable · the test is below
What it means
Given a page, a language model can score which entities it is actually about rather than which merely appear. Google's Natural Language API returns exactly this, per entity, as a number.
Why it matters for an HVAC company
It settles arguments about what a page is actually about. Contractor sites are full of pages a machine reads as being about an equipment manufacturer, a service area list or the company itself, because those blocks are longer and more repetitive than the service copy they surround. Salience puts a number on that instead of leaving it to opinion.
What would settle it
Run the page text through an entity analysis API and read the salience scores. No interpretation needed, and it is one of the only numbers on this site you can generate rather than observe.
A worked example
An AC repair page returns salience of 0.38 for “Trane” and 0.31 for “Carrier”, against 0.09 for “AC repair”. The cause is a brands-we-service block listing eighteen manufacturer names, and to a machine reading the page, the page is about Trane. Moving that block to its own brands page leaves the repair page about repair, and “AC repair” comes back at 0.27. Nothing about the repair copy changed.