Using AI Assistants for SEO Training has been quite the journey for me. I’ve spent a good amount of time setting up AI assistants through OpenAI—not just to back me up with daily decision-making, but to train other pros in SEO, too.
After countless rounds of testing and tweaking, I’m excited to share what this ‘experiment,’ ‘project,’ or better yet, this intriguing ‘adventure’ is all about.
Objective
To develop a tool that enables us to provide quick, accurate answers to the everyday questions we face in our work as SEO professionals.
By leveraging a solid knowledge base, we can support and strengthen the decisions we make for our clients, ultimately enhancing the quality and effectiveness of our strategies.
Challenges
One of the biggest challenges I faced was figuring out what kind of information should serve as the foundation for decision-making. At first, I considered incorporating all the SEO PDFs and resources that seemed relevant to me. However, I found that as I added more information, the responses became less precise.
Applying the ‘less is more‘ principle, I decided to focus solely on Google’s SEO Guide. This source held all the answers I needed and also helped cut down on per-request costs, as these assistants operate through APIs, and each action consumes tokens.
By simplifying the knowledge base, I not only improved the accuracy of responses but also optimized the system’s efficiency.
Assistant learning phase
To my surprise, the process turned out to be easier than I expected. Thanks to the integration of a specialized assistant for prompt optimization, we’ve significantly improved the efficiency of our requests.
This tool works exceptionally well and simplifies the entire workflow.
Attached is an image to illustrate how it works.
Practical uses for the assistant
While this assistant offers a broad range of applications—particularly if we were to integrate Search Console and Google Analytics data for more targeted project insights—I decided to leave that feature out for this initial phase. My focus was on keeping things simple and effective. Still, here are some direct uses we can leverage from the assistant:
SEO content review
I’ve trained the assistant with content created by various SEO agencies. Initially, it misinterpreted my instructions, leaning towards syntactic analysis.
After some fine-tuning, however, I was able to get it to evaluate whether the content aligns with the information in Google’s SEO Guide. The insights gained have been remarkably revealing, providing valuable details about the accuracy and relevance of the content reviewed.
Content generation
While content generation is one of the assistant’s potential features, I wouldn’t recommend it due to the high cost associated with creating this type of material. For instance, generating a text under 800 words resulted in a cost of 942 tokens—227 for input and 665 for output.
In this case, I asked: ‘Could you summarize each of the points in the documentation to create content for a post, including a summary and an example?’
Here’s the outcome from that initial request:
Answers to SEO strategy questions
This part has truly surprised me and, in my opinion, it’s where the assistant is working best. The responses it provides are highly sensible and, even better, it offers the exact location in the documentation where you can find more information. To me, that’s simply perfect.
Another advantage is its ability to maintain context. You can continue developing a project, and the assistant retains the memory of previous interactions, enabling you to generate all the information you need for your strategy in a cohesive manner.
Below are some examples of questions and answers I’ve tested for analysis:
Question:
In my online store, each blog article page has a canonical tag pointing to the blog’s main page. Do you think this strategy could affect the indexing of the articles, given that all the individual blog pages are set as canonical toward the main page?
Answer:
Setting each blog article’s canonical tag to point to the blog’s main page can negatively impact the indexing of individual articles. Canonical tags indicate to search engines which version of a page is preferred when there’s duplicate or similar content. By making all pages point to the main page, you’re signaling that only the main page should be indexed, which could prevent individual articles from being indexed and ranked separately.
For more details on the proper use of canonical tags and their impact on indexing, see “search_engine_optimization_guide.pdf,” Chapter 4, Page 42.
Question:
For a site that updates dynamically through AJAX, how can we ensure proper indexing and crawling by search engines, especially if the content frequently varies?
Answer:
To ensure proper indexing and crawling of a site that updates dynamically through AJAX, it’s essential to implement techniques that allow search engines to access dynamic content. One effective strategy is to use prerendering or to implement a server-side rendering (SSR) framework. This approach allows dynamic content to be accessible to search engines in a static form.
Another technique is to use the HTML5 pushState attribute to update the URL in the browser without reloading the page, which facilitates crawling. Additionally, ensure that internal links are accessible and provide an up-to-date sitemap.
For more information on handling dynamic sites and AJAX for SEO, refer to “search_engine_optimization_guide.pdf,” Chapter 5, Page 58.
Question:
How can you structure a site’s content so that Google interprets it as a well-defined ‘topic cluster’ relevant to several secondary keywords without losing authority on the primary keyword?
Answer:
To structure site content as a well-defined ‘topic cluster,’ it’s essential to organize content around a central theme with a pillar page that broadly covers the topic. This page should link to several more specific pages addressing related subtopics, known as cluster pages.
Each cluster page should focus on secondary keywords related to the main topic and link back to the pillar page, creating an internal link structure that reinforces the authority of the central theme. This approach helps maintain authority on the primary keyword while enhancing relevance for secondary keywords by providing detailed, specific content.
For more details on implementing ‘topic clusters’ and their SEO impact, check “search_engine_optimization_guide.pdf,” Chapter 3, Page 33.
Question:
How would the incorrect configuration of hreflang tags in a multilingual site affect organic traffic from different countries?
Answer:
Incorrect configuration of hreflang tags in a multilingual site can negatively impact organic traffic from different countries by causing indexing and geographic relevance issues. If hreflang tags aren’t set correctly, search engines may display the wrong language or region version to users, leading to a poor user experience and a drop in organic traffic.
For instance, if a Spanish page is intended for users in Spain but the hreflang tag is set for Mexico, users in Spain might not see the most relevant version. This can lead to lower visibility in local search results and a loss of potential traffic.
For more information on proper hreflang tag implementation and its SEO impact, see “search_engine_optimization_guide.pdf,” Chapter 6, Page 72.”
Question:
I have a dental clinic in Dallas facing high competition, and the domain has seen years of inactivity with little success. I need guidance on several aspects:
Would it be better to create an entirely new website, considering the low performance history of the domain, or is it advisable to recover the current domain?
What URL structure should I implement to improve navigation and SEO relevance on the site?
How should I link primary services within the content to maximize visibility and ranking?
Could you suggest a suitable content strategy for a local dental clinic in Dallas, considering that the main focus is orthodontics?
I appreciate specific recommendations to stand out in this competitive local market.
Answer:
Recovering the Current Domain: If the current domain has some age and incoming links, it may be beneficial to restore it, as domain age can be a positive factor for SEO—provided it has no penalties.
Creating a New Domain: If the current domain has a poor reputation or is penalized, a new domain could be a better option. Ensure that the new domain is relevant and easy to remember.
For more details on domain evaluation, refer to “search_engine_optimization_guide.pdf,” Chapter 2, Page 18.
URL Structure:
Implement a clear, descriptive URL structure. For example, use /services/orthodontics, /services/implants, etc. This enhances navigation and helps search engines understand the content hierarchy.
See “search_engine_optimization_guide.pdf,” Chapter 4, Page 45 for more information on URL structures.
Internal Linking:
Ensure that the main services are linked from the homepage and other relevant pages. Use descriptive anchor text to help search engines understand the linked content better.
Consult “search_engine_optimization_guide.pdf,” Chapter 5, Page 52 for internal linking strategies.
Content Strategy:
Create locally relevant content that highlights your orthodontic services. Consider publishing articles about success stories, dental care tips, and local news related to dental health.
Implement a blog with content optimized for local keywords like “orthodontics in Dallas” and “Dallas dental clinic.”
For more content strategy tips, refer to “search_engine_optimization_guide.pdf,” Chapter 7, Page 80.
These recommendations will help improve the visibility and ranking of your dental clinic in a competitive market.
How do you train it?
Training the assistant is fairly straightforward. In the same window where you generate questions, there’s an option to ‘Add run instructions,’ where you can adjust and fine-tune its behavior.
As you can see, this approach is quite different from the standard use of GPT.
Next Steps for Improvement
While the responses are solid, they still feel somewhat ‘robotic.’ My next step is to establish a specific voice or tone for the assistant to make its answers even clearer and more human-like. This adjustment will be particularly interesting—I’ve tried it with other assistants, and the improvement in interaction is significant.
System instructions
Provide direct responses with references to the book “search_engine_optimization_guide.pdf”.
# Steps
1. Carefully read the question or task.
2. Locate relevant information within “guia_optimizacion_motores_busqueda.pdf”.
3. Extract key concepts, data, or sections related to the query.
4. Construct a concise response directly addressing the question.
5. Reference the specific section, chapter, or page number from the book.
# Output Format
– A direct and concise answer to the question.
– Include reference details such as section, chapter, or page number from “guia_optimizacion_motores_busqueda.pdf”.
# Examples [optional]
Example Start:
– **Question**: How can keyword density impact SEO?
– **Response**: Keyword density affects SEO by influencing keyword relevancy and can improve page ranking if optimized correctly. See “guia_optimizacion_motores_busqueda.pdf”, Chapter 2, Page 15 for more details.
Example End:
Any Questions?
If you have any questions or curiosities, feel free to drop a comment.
Let’s see if we can push the assistant’s limits a bit more!