SEO Trends and Expectations

AI use for SEO

• LLM is not an AI from science fiction. It is just a well-tuned predictor of the most relevant words we need in a specific context. Thus, it is able to create both perfectly logical, useful texts and bizarre hallucinations. 
• ChatGPT is a new Excel. A powerful tool to enhance productivity in many fields. SEO is not an exception. But the best results it brings are not in SEO specifically, but in lightening the load of accompanying stuff like dealing with emails, memos, brainstorming, and data analytics.
• LLM could offer reliable help with some code-related tasks, like generating schema markup, but it is not a game-changer.
• I strongly believe that modern LLMs are not able to generate SEO strategy or competitor analysis.
• LLM may be used for content generation. There are a lot of examples where this content ranks well. But it does not mean any site could successfully use this content. Remember survivorship bias. I have many examples where AI-generated content was hit pretty hard by Google Helpful Update.
• LLMs bring better results when we get a combination of 3 factors: a) we use a lot of our own data to provide context, b) we know exactly how the result should look, and c) an expert reviews the output before we use it.

Unhelpful content update

• If someone was in doubt whether or not to use AI content head-on, Google resolved the matter. Straightforward AI content is not for sites. User gets it in SERP (I am talking about SGE).
• There is no one definitive reason or common trait of sites that were hit.
• I think Google just started to assess website quality by more broad lists of criteria. There is nothing actually new here, the rules were just made stricter. 
• To restore after the update we, first of all, need to carefully study our main target keywords and make sure that relevant pages are really relevant for the user’s intent. Such audit could be very cumbersome, of course.

SGE (now AI overviews)

• The product itself is very raw. A lot of people hate it. But de-rolling is very unlikely, so SEOs need to adapt to it.
• There are reasons to believe that the impact on the betting and gambling affiliate marketing industry will be less significant than on average, at least for now. 
• Google claims that links in AI answers bring traffic of higher quality because those users are associated with higher dwell time. I think it is true, but it is not good news. It is to be expected. The sites get only the users who want to do deep research. It is not about improving the click’s quality. Those who would spend less time just do not click. Here is the analogy: a 1×1 inсh piece from the middle of a watermelon is more juicy than a whole watermelon. But it does not mean that the tiny piece is more valuable!

The reign of Reddit

• I do not think the situation with Reddit-dominated SERPs will last long. I am sure relatively soon we will observe regression toward the mean. 
• Reddit still will rank incredibly well. But some queries will be occupied by successfully adapted websites; in other cases Google will get enough user behavior data and re-balance SERPs.
• Of course, SEOs could use the current situation by creating content that suits their marketing goals on Reddit and other social sites.
• In some cases UGC content really provides the best answer possible, because no one else tried to do so. It is a source of valuable content ideas.

Parasite SEO and Authority Abuse policy

• Danny Sullivan said that Google has not gone live with algorithmic actions on site reputation abuse.
• I monitor 120 affiliate-related keywords on 10 markets. It actually looks like only manual actions here and mostly on tier-1 markets. 
• I do not believe parasite SEO would die. Google starts to fight it exactly because it is efficient. I think parasite SEO will lose some of its simplicity, but there will always be successful cases.
• I myself will not invest in parasite SEO. I prefer to develop my own website and accumulate my own authority to use it wisely.