Looks like you won’t have to tell clients “it depends” as often as you used to…
Google released a guide containing descriptions and updates of all ranking systems in a centralized hub for users to “better understand how these systems work.”
New, old, retired: The centralized page will reflect real-time tweaks to existing systems. It will also publish future updates and sunset the outdated ones.
It also describes Google’s AI system categories, which are responsible for different aspects of content:
- Bidirectional Encoder Representations from Transformers (BERT) helps Google understand how combinations of words express different meanings.
- Multitask Unified Model (MUM) is a system capable of both understanding and generating language. It’s responsible for creating things like featured snippets.
- Neural matching helps Google understand representations of concepts in queries and pages and match them to one another.
- RankBrain system allows Google to understand how words are related to concepts.
You can see the full list of all current and retired systems here.
Why the change: Besides the need for a centralized hub, Google also wanted to differentiate between “systems” and “updates to those systems.”
For instance, Google’s most recent “Helpful content update” was a brand new system. Calling updates to this system, “update to Helpful content update” is terribly confusing.
Why we care: Having a clear understanding of which ranking systems exist, how Google defines them, and which are still in use is obviously helpful for several reasons.
For example, you might refer to this document when you’re explaining to clients why something works or doesn’t work. Nice, right?