There’s a war going on
I know I’m stating the obvious to anyone who is paying attention. If the first thought that comes to your mind is AI, you’re on the ball. If not, you’re either living under a rock or watching too much TV. The war over AI is not brewing, it’s getting heated. Google and Microsoft are 2 giants that have the power and ability to change our reality in many ways, but right now AI is the key technology that will change the way we do things. Search is just one of the ways our lives will change, but it’s an important part of our lives as we perform billions of searches every day. Search has not changed in the last 20 years and we are starting to see the effects of SEO keyword stuffing, which at times leads to poor search results, and this trend will continue.
Let the numbers speak
Statistics show (will you find an image below?) that around 40% of people search for something and immediately return to search after landing on the page (bounce rate). This means that what they found was not what they’re looking for, despite all Google’s efforts to show you the most relevant links first.
40% means nothing on its own, but when you put it into perspective, you understand the magnitude of it. Google processes over 8 billion search queries a day. If you’re like most people who have access to the internet, you do 3 Google searches a day or more. And again, if you are part of the majority, you have not looked at the second page of Google search results in ages. Less than 1% of people click to get to the second page. People will use a different search term before clicking on the second page. Google is failing in its mission to give you the best possible information on the first click.
Gigantomachy
It’s undeniable that AI can answer your questions much better than Google or any other search engine. Instead of clicking on a series of links and working your way through a lot of unwanted information, you get your answer instantly. We will see Microsoft Bing challenge Google with its own AI tool for search. The war between Microsoft and Google will mean the end of search engines as we know them. Search engines will be indistinguishable from AI chatbots like ChatGPT. We can see this happening in real time as Bing integrates OpenAI technology and Google develops and eventually integrates Bard into its search engine technology. Note that Microsoft claims to be using the latest OpenAI technology, which is more advanced than ChatGPT in its current version 3.5. If you have not tried the Bing chatbot yet, you really should.
When I say the war is getting heated, I mean it. There’s a possibility that Microsoft will dethrone Google. That sounds crazy right now, but we’ve seen crazier things happen before.
Are we witnessing a losing battle?
In early February, Google presented Bard and to say it was a botched presentation is not an exaggeration, which is saying something for a presentation by a company of this magnitude. There are two reasons that make this statement seem justified: first, the fact that just a few hours before the live event, Bard was tweeted in a Google ad with an incorrect answer to a question posed by a user, and secondly that Bard was not given enough screen time, especially considering what Microsoft is doing. We saw that the executive and team were not prepared to carry out the presentation because there was no phone available to do it, causing Google shares to fall by about 7%-9% (~$100 billion) during and shortly after the presentation.
If you want to watch the entire presentation, here’s the link to Google’s Youtube video. You will notice that the comments are turned off, which is not common on the Google’ account. You can watch the videos of the previous events and see that the comments are switched on. If you want to watch Microsoft’s highlights, it’s definitely worth it, as it lashes out the latest developments in search that you didn’t get from Google… You didn’t get any pictures, but were showered with videos.
Let me entertain you
Will users want to read blogs if they can get quick, direct answers and avoid all the flowery language that permeates the writings of most mundane blogs? I know, you’re looking at me and nodding. I smirk back at you.
If you read my last blog and are in any way interested in how you can use AI to write blogs that rank well on SEPRs, stay the course. First, I will explain how I prepared ChatGPT to write an article that will rank well and hopefully drive traffic to your website/landing page. Secondly, I will tell you about the actual results I achieved with my last blog. And finally, I will share my lessons learned. As usual, I should conclude with a hint on what I plan to write about next.
No rest for the wicked
Before I opened ChatGPT, I wanted to decide what topic I wanted to write about. I could have easily asked ChatGPT for ideas, but I was too embroiled with my own thoughts around my oversight that I didn’t think to ask for help and didn’t stop until I had the brilliant idea to write a blog about SEO. I wanted to test my hypotheses about how effective the article would be in terms of SERP ranking, which is exactly what SEO exists for. Could I have come up with a stupider idea? More on that later.
A recipe for success
Preparation is key
Once I know what I want to write about, the first thing I do is to search for the most important keywords for that topic using Google’s Key Planner. If you don’t have access to it, you can create an account without a campaign, switch to expert mode and click on Tools and Settings / Keyword Planner.
Now I have a list of keywords to work with. I choose the keyword that seems most relevant and do a Google search for it. When you do a Google search, on the first page you will find a section labelled “More to Ask”. If you click on some of the questions displayed, more questions will appear. These are the most frequently asked questions that users search on Google. Once you have a good list of questions, you will shortlist it, to a number you’re happy with. Between 5 and 10 should be enough, but you may want to write a more comprehensive article and select a few more questions.
Prepping
Now armed with the questions I want to answer in my article and the keywords I want to use, I open ChatGPT and start the prepping. My aim is not to have ChatGPT start writing the article immediately, but to prepare it for writing the article later. Below are the prompts I used. I have to paraphrase as ChatGPT does not show the list of previous chats at the moment. As ChatGPT saves your chats; you can find them on the left side; I wanted to copy/paste the prompts I used, but I get the message “Not seeing what you expected here? Don’t worry, your conversation data is preserved! Check back soon.”
- I want to write an article about SEO. Do you understand? ChatGPT will reply that it understood your request. It usually returns what it understood from your request.
- My brand, BestSites 0x, does not offer standalone SEO services, but creates websites using SEO best practices. Do you understand?
- I want you to use the normal format for a blog, like H1, H2, H3 and bullet points. Do you understand?
- I want you to answer the following questions: “List of questions” Do you understand?
- I want you to use as many keywords from this list as possible: “List of keywords” Do you understand?
Choose your battles
Then I spent a lot of time trying to get ChatGPT to embed a few links to some keywords, but that did not work and since I didn’t have that many keywords/links, I decided to do it manually. I spent this time not because of this particular blog, but because I was thinking of future blogs. I found a workaround with the Apps Script extension on Google Docs, but will work with ChatGPT again until I find the right prompt, as it shouldn’t be hard for the AI to do it. I’m pretty sure the problem was my prompts and not ChatGPTs capabilities.
Kneading, blending and cutting
Can you please write an outline for this blog post? Sometimes I feel the need to treat the AI with the courtesy it deserves and feel bad about all the previous interactions where I let those good manners slip. It keeps happening though, it’s only code after all.
ChatGPT will create an outline. At this stage, you tell ChatGPT what should be added and/or removed.
When you’re happy with the outline, you can proceed with writing the blog.
There are 2 ways to do this. I used the prompt “Write the blog and don’t forget the questions and keywords”
Then I had to revise the whole blog, paying attention not only to the message but also to the use of the keywords and questions I wanted to use.
A better method is a slightly different one.
You can ask ChatGPT to write each section in turn. For example, “Write the introduction.” You then review the introduction and ask ChatGPT to rewrite it if required, to add a particular keyword or message you want to convey.
Recipe for disaster?
So, did I end up on page 1 of the Google’ SERPs? I never thought it was possible when I started, but to be honest, I wasn’t sure what to expect either. What was the result of using this process? 0, nada, rien, zilch. I searched to the depths of the SERPs for my article, at least until page 10 and no trace of it. It’s as if it didn’t exist.
My articles on Bitcoin and Why Small Businesses Fail fared much better with over 2 dozen reads each compared to 6 reads that I know of for this last article. Does it faze me? No, I was rewarded even before I hit publish.
That’s not to say that increasing my reach is not my goal, but I know it takes time, persistence and perseverance. Writing a blog is not marketing, but it definitely has those same requirements.
I don’t want to think that it failed because of the quality of the content if no one read it, but the truth is that there are already some articles that have been read, and if readers do not come back, it’s definitely because of the quality of the content. I keep that in mind.
Another thing to consider is the snippet of text that is displayed before you open the article. The quality of this excerpt is important, and being critical of my work, I have to admit that some of these excerpts could be better to entice you to open the article. This specific blog included, unless I change it before I publish it.
Intelligent failure?
Do I think this flop has anything to do with the process itself? No. However, the process needs a little tweaking and, above all, the person who wants to use it needs to put in a little more time and effort than what I put in. I learned several lessons. There are 2 main reasons why this article didn’t rank.
Domain Authority
1 — Website Domain Authority (DA). Domain authority is the strength of a domain. It predicts how likely a piece of content hosted on that domain is to appear in the SERPs.
DA ranges from 0 to 100 (Moz will say 1 to 100). DA is calculated using several factors, but backlinks and linking root domains are the most important as far as I can tell. The strength of these links is also important. For example, it’s better to have one backlink from a domain with a high DA than several from domains with very low DA values.
Google does not consider the DA value of a website, but it considers all factors that influence the DA value of a website. Your website has a DA value referencing the likelihood that it will be ranked, but its ranking is not influenced by the DA value it has. I’m doing my best to make this clear without complicating it, but I have this itch that makes me think I’m kind of failing here.
My website’s DA has a value of 0.1 (that’s why I say it’s from 0 to 100). When search engines evaluate information on a certain topic, they prioritise domains that are considered trustworthy and have high-quality content. How do search engines know if a domain is trustworthy and offers high- quality content? When other websites with good DA values link to the content of that domain.
Cutthroat competition
2 — SEO. I had the brilliant idea to test my first article written specifically for SERP ranking to target the keyword SEO and its related keywords. Why was this brilliant? I’m assuming you can read the sarcasm in my writing.
SEO is pretty much the business of the masters of the internet search business. I mean that! The best in the field can make a website rank in the first page of SERPs in a matter of a couple of months. I still believe that any domain with a DA of over 20 will rank on the first page of Google with articles written with this recipe, except when writing about a few highly competitive topics, which include SEO. Every day about 7 million blogs are published. Having your blog on the first page of the SERPs is an amazing achievement and a goal I would love to reach one day.
An opportunity to breed luck
Those were the 2 main reasons, but they weren’t the only ones.
Do the leg work
- I didn’t put enough time and effort into actually writing the article. Both while working with ChatGPT and afterwards.
Garbage in garbage out
- I fed ChatGPT a very long list of keywords (I’m embarrassed to say how many, that’s how many). The result was that the AI used the ones it “thought” should be used. When I wanted to check the keywords used, I was overwhelmed due to the number of total words in question. The lesson here is to use a moderate keyword list. 20 to 30 should suffice, and make sure they’re all used. In case it wasn’t clear yet, I’ll say it now: The quality of the work an AI produces is only as good as the quality of the input it receives.
Trade-offs and creative thinking
- Be smart about the topic you choose to write about (or have the AI write about), if the goal is to rank and drive traffic to your site. You won’t get away with writing content that has nothing to do with what you sell/offer, but you can use niche topics that have something to do with your product/service. For example, a pet shop could write about ant keeping even though they don’t sell ants. A web designer could write about a new film that uses the latest technology for real-time rendering.
- Choosing a topic that is either just trending or is important but not discussed much can make or break your chances of ranking on page 1 of the SERPs. Focus on your niche and look for topics that you are most likely to add value to. You might need to focus on keywords that are searched for with much less frequency. It may be that only 5 people are searching for that exact keyword every week, but these people will most likely click to read your article. Please note that by keyword I also mean long-tail keywords or keyword passphrases
What to expect next
I’ll save you the conclusion or summary, you don’t need to read the same thing again in possibly different words. I’ll just add this, not as a conclusion but as a forewarning. Be mindful of the AI biases. Put in the effort to make sure the message in your blog reflects your own thoughts (and biases) but not the AI ones. I’m a great advocate of good sense but not of common sense. I don’t think they’re one and the same. This is probably a good topic to write about but possibly in a different blog. AI’s biases, good sense and common sense.
I plan to “write” an AI article from time to time and will use the process I just expounded with some adjustments to implement my findings. I will probably share the results in terms of reach.
However, there are a few topics I want to write about first. Web design trends and inspiration. Crypto and responsible investing. Definitely Bitcoin part 2. There are a few important things I did not cover in the first article and a few topics I could have done a better job on in the same article.
My next venture into AI is to try and code a widget that I want to use on one of my mockup websites or possibly for a customer’s one. I want to create something I have not seen yet (which is not to say it does not exist, because I can not know everything that’s out there). I will document the journey and make it available along with the widget if I manage to code one.