Conversation

Did you know Google updated their search quality evaluator guidelines this week? We pay attention to these because they can give us clues as to what Google wants to accomplish with their search algorithms. Here are five things I found interesting with this revision. 👇
11
301
1) YMYL is now more clearly defined. Understanding this is important because Google says in their guide to how they fight disinformation that they give more weight to their understanding of EAT for YMYL topics.
Image
2
19
The main question to ask re whether your content is YMYL is whether the topic, or inaccurate information on the topic has the potential to harm. (They also changed many instances of "YMYL pages" to "YMYL topics.")
Image
1
12
With the QRG update, they gave us a load of examples to help us understand whether our topics are YMYL, not YMYL or somewhere in between. I would argue that a hot sauce challenge at our house could significantly harm someone's life😂🔥🌶️
Image
2
32
For every piece of content you have, you should ask yourself whether your content has the potential to cause people significant harm. If yes, your topics are YMYL and you need to pay close attention to E-A-T.
1
10
3) Some more hints at Google using Machine Learning to weight ranking factors. These wording changes make it sound like the raters are teaching computers a good result from a bad one to train ML algos🤔...Or maybe I'm reading too much into this?
Image
2
15
4) The two most important factors the quality raters rate are a) Page Quality ➡️What is the purpose of the page? ➡️Is it harmful? ➡️Rate the page quality b)Needs Met ➡️Determine user intent ➡️Rate the page in how it meets user intent
Image
3
17
Replying to
This is probably a good place to mention our book on the QRG has been reduced from $99 to $20. My team and I use this as a big chunk of our site quality reviews. Teaches you how to assess EAT like a quality rater:
3
8
Replying to
Human raters are often a component of good AI systems. Raters can look at a set of results and essentially label them as good/bad and then AI can take those labels and figure out how to modify the algo to replicate that more broadly. This is interesting:
1
2
Show replies