Google uses several hidden systems to decide which websites appear at the top of your search results. For a long time, people thought Google only looked at the words on a page or the links pointing to it.

However, the recent US Department of Justice legal case and some leaked internal documents have proved that Google relies heavily on how real people behave.

Google Glue

Google Glue is essentially a “Super Query Log” that covers everything on the search results page beyond the standard blue links. If you think of the search page as a digital shop window, NavBoost is watching which products people buy, but Glue is watching which items they stop to look at, which posters they read, and which aisles they walk down.

  • Glue monitors “Universal Search” features such as Map Packs, Video Carousels, Featured Snippets, and Image Boxes.
  • While NavBoost is famous for tracking clicks, Glue captures “micro-interactions” like hovers, scrolls, swipes, and even how long you look at a specific module without clicking anything at all.
  • Glue helps Google understand if a user wants information in a specific format. If Glue sees that users on mobile devices always scroll past the text to find the map for “coffee near me,” it will eventually move the Map Pack to the very top.

How Glue, NavBoost and Popularity interact

These systems work together in a massive feedback loop to create what Google calls the Popularity Signal (P)*. This signal is then used to re-rank the initial results provided by the classic search algorithm.

You could think of it as NavBoost handling the traditional web results (the 10 blue links), while Glue handles the rest of the page features. They are effectively two branches of the same system that communicate constantly to ensure the whole page is satisfying.

This then interacts with the Popularity Signal.

Every search query has a “Popularity Score” attached to it, and this score is built from thirteen months of data collected by Glue and NavBoost.

Positive Signals are “good clicks” (staying on a site) and “longest clicks” (the last site a user visited before finishing their search).

Negative Signals, however, are “bad licks” (clicking and immediately hitting the back button) or “impressions without clicks” (seeing a result but choosing to ignore it).

The next phase is the re-ranking phase.

Google first uses its standard “quality” signals (like keywords and links) to find a list of possible results. Then, it applies the Popularity Signal from Glue and NavBoost to re-order that list. If a page is in position five but has a “Popularity Score” much higher than the page in position one, the systems will swap them.

The weightings and impact of positive and negative user interactions aren’t known, as Google’s systems have a high level of imponderable complexity.

What does this mean for SEO and AI Search?

The discovery of Glue proves that brand power is your greatest asset. When users recognise your brand name on a busy search page, they are more likely to click it or hover over it.

These small actions feed into Glue and NavBoost, telling Google that your brand is the popular choice. This creates a defensive moat that makes it very hard for lesser-known competitors to outrank you, regardless of how much they spend on SEO – as SEO can’t fix a broken brand and user experience.