Creating SEO text via ChatGPT: tips, practices, prompts
Today, technologies related to the use of artificial intelligence are developing very rapidly. Every day, new areas of its use appear. As an example, virtual bloggers, which are based on artificial intelligence, are already actively working on the market. Here, everything is so natural and dynamic that not everyone who sees their posts and stories guesses that they are subscribed to an AI, not a real person. But still, one of the first, but very important and widely used areas of application of neural networks in practice was the generation of text content.
When this technology first appeared on the market, it began to be used almost everywhere, replacing texts written by copywriters with such materials. But after a short period of time, it became clear that Google, so to speak, does not particularly “respect” generated text content. If there was too much of it on the site, the search engine lowered such resources in the search results. That is, as a result, there is a serious risk of losing those positions that were won for a long time and with great difficulty through competent optimization and promotion.
But at the same time, not everyone is ready to part with a convenient and easy-to-use tool. So, is it possible to use artificial intelligence to generate text content without the risk of running into restrictions and sanctions from the search engine? This is what we will talk about in today's review. We will tell you what formats and types of SEO texts can be created using modern neural networks, in particular ChatGPT. We will provide a number of practical recommendations that will allow you to get good results, we will describe in detail all the activities that will need to be implemented as part of preparing the generated text for publication. We will provide a number of additional recommendations that will allow you to create sites through ChatGPT that neither Google nor any other search engine will perceive as low-quality.
What types of SEO texts can be created using AI
If you analyze the total volume of texts generated by artificial intelligence and presented on the Internet today, you can highlight those groups where you can find quite a lot of them:
- Category pages. This is the main page of any online store, which contains direct links to shopping groups. Once the user clicks on the relevant category, they will immediately be redirected to the product catalog, where they can select the appropriate option and place an order.
- Filter pages. This is a search result obtained by narrowing the general list in a category by certain parameters. By optimizing such pages and organizing correct work with them, the search for the necessary products is significantly simplified, and interaction with the site is improved.
- Various information pages. This includes all sections of the site that present interesting and useful information for the user audience. This also includes blog pages, the "About Us" section, and "Contacts" etc.
- Materials that are being prepared for placement on external sources, including as part of outreach, cross-marketing etc.
- To create fan content. These are various materials that are created and distributed by fans on a particular topic. No matter how unrealistic it may sound, but the same ChatGPT is able to instantly issue a questionnaire on any topic, such as “What bouquet to give a girl on the first date” and immediately provide a system for calculating answers.
In practice, texts generated by artificial intelligence can be posted on sites of any direction. These can be marketplaces and smaller online stores, representative offices of companies that specialize in cargo transportation, logistics, repair of buildings, houses, apartments, household appliances, construction, resources of medical centers and more.
Among all this, medical topics stand out in particular, because here we are talking about sites that search engines classify as YMYL. The fact is that such a business can have a direct impact on health, and in some cases even on people's lives. By the way, search engines also classify other sites related to people's security or finances in this category.
Google will impose stricter requirements on all resources that have such a mark. Among other things, it will evaluate them according to the EEAT algorithm. That is, during the check, the expertise, experience, authority and reliability of the materials presented in the publication will be assessed. But if you analyze the traffic coming from medical and banking sites that contain texts generated by neural networks, you can see that the search engine does not impose any additional sanctions on them.
What you need to know about ChatGPT as a tool for generating SEO texts
Among all the neural networks that exist on the market today, ChatGPT is one of the most famous. Moreover, for many, it has become a household name for artificial intelligence in general. But still, the text content that it creates does not always occupy leading positions in the TOPs. This is especially true for materials that are prepared specifically for organic website promotion. And one of the most serious reasons here is primarily the version of the neural network itself.
If you use free ChatGPT, then even with all the desire you will not get texts that will be perceived by the search engine as high-quality, authoritative. Outdated and unreliable information may be used here. Therefore, the first recommendation for those who would like to use ChatGPT when preparing text content is to restore the selection to the most current paid version of the neural network. In its original version, there were quite a lot of errors, but with each subsequent update they are eliminated, the service as a whole is improving.
If you have already had to work with version 3.5, that is, an incomplete solution, then you probably already noticed that such ChatGPT very often writes what you want. That is, there is a banal ignoring of user queries. And even with considerable effort, it was impossible to achieve something high-quality from AI. Users noticed spelling errors here when writing both Russian-language and English-language texts. It was almost impossible to get it to add key queries to the texts. And if this worked, then basically all the keywords were written in one row separated by commas. As a result, all this had to be manually modified, modified, and the necessary phrases included, which took a lot of time, effort, and, for the most part, nerves.
But one only has to look at the next version, the same 4.0, to notice striking differences. And here you don’t have to be an expert to understand that this is a fundamentally new product that has little in common with its predecessor. Yes, you will have to pay for such a solution, and not as cheaply as many would like, but in the end you will be able to get high-quality materials that meet even the strict requirements of modern search engines.
ChatGPT will be especially convenient to use where it is necessary to create large volumes of texts in a short time. As an option, for example, to launch a website for a small medical center, you will need:
- approximately 25-30 texts for the main information catalog;
- the same 25-30 texts for service category pages;
- approximately 30-35 texts for posting on blog pages;
- approximately 100 texts for external posting.
What do we have as a result? The need to write 180-195 texts in the shortest possible time and stretch this period over six months, then it will be impossible to launch the site. And this means that patients will not be able to find the relevant information and use the services, the clinic itself will not receive clients. By the way, if you prepare the prompts correctly, then it will take about half an hour on average to create one text. This means that in a working day (8 hours) you can create about 16 texts, and it will take you about 2 weeks to prepare the entire volume. For comparison, a real person can write this amount of material in about a month and a half, even if he works very intensively and practically without days off. As you can see, the difference is very significant.
General recommendations for creating prompts for AI for generating texts
Prompt is a sequential record of commands that will set the neural network to perform a particular task. Since in this case we are talking about generating SEO text, you should have a written technical task indicating all the points that should be present in the future material, its structure. Based on this information, it will be much easier to create a prompt. Otherwise, you will need to build the most appropriate structure of the future material based on key queries, that is, select headings and subheadings of different levels in order to understand what information should be presented in the article.
Alternatively, you can also involve a neural network in performing this task. That is, you can ask ChatGPT to provide you with several options for the plan of the future article, and then send them to the customer for approval. When there is a choice, he will be able to determine the option that will best meet his requests. This means that in the end it is more likely to accept the finished content without any problems.
Also, another point that we want to draw your attention to is whether you want to create the entire text at once or prepare it in separate blocks. The first option can be used to generate small texts. Those whose volume does not exceed 2-3 thousand characters. The second option involves sending step-by-step commands to ChatGPT, on the basis of which the neural network will create sections devoted to a specific issue. In practice, such a solution should be used when preparing voluminous materials that involve a complex structure and a large amount of diverse material.
As an option, let's assume that you need to write an article about solar panels. It may contain such blocks as a description of the solar power plant itself, the advantages of its use, the types of solar batteries that can be used in its arrangement, additional components necessary to ensure operability, features of choosing a place for installation, a method of fixing the panels. You can also add recommendations that will improve the efficiency of electricity generation or extend the service life of such a power plant. In this case, it is necessary to generate separate blocks, and then collect the finished material from them. Otherwise, the neural network will not be able to create a text that would include all these sections in a clearly defined sequence and with all the questions that need to be answered.
You should also be prepared for the fact that ChatGPT gives fairly concise answers to a number of questions. It may turn out that an entire section will consist of literally a couple of sentences. And this means that you will have to either supplement it yourself or ask the neural network additional queries. That is, there is a lot of work ahead. It will be necessary to write detailed prompts for each separate block, indicating the requirements that the material must meet. If you want a particular block to have a bulleted or numbered list, then this must also be indicated. Don't forget to additionally write the introduction and conclusions in order to obtain a completely logically constructed structure containing both the introduction and the call to perform the target action.
You must also pay attention to all the requirements that high-quality SEO content must meet, namely:
- adding key queries to the material, and also to the headings, subheadings of different levels in such quantity that there is no spam;
- a structure thought out to the smallest detail, including logically and consistently constructed headings, lists, where possible — tables explaining images;
- the material itself should be interesting, informative and relevant to your keywords, the contents of each block should reveal its name.
Detailed instructions for developing a prompt for ChatGPT
We would like to draw your attention to the fact that the success of text generation in any neural network, including ChatGPT, is a correctly posed task. To make it easier for you to prepare a high-quality prompt, use our step-by-step recommendations:
- If you already have a prepared text structure, write it down. If you don't, ask the neural network to create this very structure in several possible versions.
- Specify that you need an introduction that would captivate readers and encourage them to study the entire material, as well as a logical conclusion, including a call to action.
- Explain for whom this material is being created and in what context the topic should be disclosed. If we return to solar panels, then here you can ask to describe the topic from the point of view of a person who is just considering the feasibility of arranging such a power plant at home. There is no need to go too much into technical issues. The main thing here is to let the person understand what investments and difficulties with the implementation of the project await him, what benefit he will receive from performing such work, indicate the service life and payback period.
- Specify what type of page or site you are preparing the content for. Alternatively, in the case we are considering, it could be just an introductory block or a website of a company that designs and installs household solar power plants. Here, the text itself will be different, since in the second case it will already contain an unobtrusive, but commercial note, like they read it, realized that it was profitable and ordered turnkey work from specialists.
- Be sure to write down the keywords. Here, you should definitely indicate those phrases that you need to include in the text in direct entry. That is, you can even write down the case.
- Indicate in the prompt additional words (LSI) or phrases that will help to reveal the topic and not miss those moments that are really interesting to the user audience. In practice, it often happens that the same ChatGPT can ignore additional phrases. If this happened in your case, you can add LSI words to the list of main keywords.
- Are there any spelling rules that you consider key when creating your text? Also indicate this in the prompt. For example, you can tell the neural network what kind of quotation marks you want to use, classic or in the form of "herringbones", what kind of dash you want to put - long or medium. Also indicate the requirements for lists here. How many of them should be in the text or in a separate block, how many points should be in each of them, what kind of list (numbered or bulleted) you want to see, etc.
- If you have an article from competitors that you really like, you can even add it to the neural network and ask to write in the same vein. But here we must not forget that high uniqueness is very important for each SEO text.
- An additional option that you can also use here is is to ask ChatGPT to immediately prepare a text with HTML markup for you. This will significantly simplify your work at the stage of preparing the material for publication.
- If the text requires images or tables, it is also important to indicate this to the neural network. In principle, ChatGPT copes with table generation quite well. However, it is not always possible to do this directly at the stage of text creation. In some cases, you will need to additionally create a table and then add it to the text. But ChatGPT still has a problem with creating images today. And here a more reasonable and appropriate solution would be to use other specialized neural networks for generating images.
In principle, creating a prompt is not as complicated a process as it might seem at first glance. Here it is important to simply clearly and consistently present all the information that the neural network will follow when preparing the material for you. The more nuances and details you take into account at this stage, the more detailed and high-quality the output text will be.
What work should be done with the already generated text
Regardless of what kind of text you got at the output, it will still require revision. You should be prepared for this even if you created a very cool prompt, all your wishes down to the smallest detail, and ChatGPT generated the content “excellent”. In any case, you need to perform a comprehensive check of the finished material. It will include the following work:
- Compliance of the presented material with the facts. You should understand that ChatGPT can not only make mistakes in some details, but also come up with content from scratch. If you are not very knowledgeable about the topic in which you will work, then you need to study several professional sites in this niche in detail in advance, or even seek competent help from a person who is well versed in this matter. In most cases, no topic can be completely trusted to artificial intelligence. Understand that this work is performed by a robot that can analyze the source material, but it is far from always possible to draw the appropriate conclusions from it correctly. Therefore, review everything carefully and, if necessary, make edits to it before handing it over to the customer for verification.
- Pay special attention to checking grammar. If you are using the latest paid version of ChatGPT, here you should be wary not so much of spelling errors as of spam, incomprehensible phrases, words, rants. This is especially true when generation is performed in less common languages.
- Keep an eye on active windows. The thing is that if you generate content in the same window all the time, the system will remember all your requirements and try to take them into account when creating absolutely every material. If today you write a text about solar power plants, and tomorrow we need you to prepare material for an online store "hunting and fishing", then nothing terrible will happen: you will receive more advanced material. This problem is more relevant in cases where a large amount of material is prepared on the same topic. As a result, the neural network will follow all your requirements as much as possible, which can make all the texts too similar to each other, monotonous.
- If you receive a text from ChatGPT that you don’t like at all, don’t rush to abandon everything and sit down to write it manually. Open a new dialog box and run the same prompt again. There is a chance that the new material will be much better than the previous one. In addition, you can combine different blocks if you think that this option will best meet your needs.
- If you have real examples of what you are writing about, implemented cases, user experience, then supplement the material received from the neural network with all this. Only a real person can do this. In addition, this way you will improve the overall quality of the text and make it one that will be appreciated not only by readers, but also by search bots.
Again, we draw your attention to the fact that the work is being carried out with a robot, albeit a trained one. That is, we cannot exclude the possibility that instead of an answer to your request, you will receive an apology from ChatGPT such as: "Sorry, but I can't write such a text." This happens especially often in topics where there are certain restrictions, such as alcoholic beverages, weapons, including for hunting, etc. But often such an error pops up and when it is necessary to generate quite ordinary content. In this case, you can try to re-run your prompt with the hope that the neural network will still offer you a text or at least throw in a few suitable blocks.
Another problem that you may encounter in practice is the incorrect interpretation of your prompt. Let's say you are writing about the same alcohol and receive material from ChatGPT about how good a product it is, what positive properties it has. And everything would be fine, but you needed a text about the harmful effects of alcohol on the body. In this case, you will need to stop the neural network and give additional instructions so that it can be rewritten, but from a slightly different perspective. And here there is a chance that the error will be corrected and you will receive relevant content.
We have already mentioned above that when working with unusual languages, grammar and spelling errors may be observed in the finished content. But it may also happen that ChatGPT does not listen to your recommendations at all and ultimately provides you with text in English. But here you can use a translator and make the appropriate adjustments to the material.
All this once again indicates that you should not rely entirely on neural networks, even such a popular one as ChatGPT. You must fully control the work of artificial intelligence and understand that at the output you get truly professional material, and not a set of random phrases containing key queries.
Let's summarize
To summarize everything we talked about today in the review, we will highlight a number of key points:
- Modern neural networks can be used to prepare SEO texts of various types, formats for placement on the official website of the company and on third-party resources in almost any niche.
- Use only reliable neural networks to perform work. Alternatively, you should not trust the free version of ChatGPT. Use a paid product and you will see how worthy articles can be obtained with its help.
- If you already have a technical task, a high-quality prompt, then you can get a pretty decent text in half an hour.
- When developing a prompt, describe in detail the structure of the future article, require an introduction and conclusions, write down the context, how the topic should be disclosed. You can even specify the type of page or site on which you are preparing the material. Key phrases and additional LSI words are required. If a direct entry is necessary, then this is also indicated. It would not be superfluous to clarify which quotation marks, the length of the dash to use in the material, which list format suits you best, how many items there should be in it. Also write down the need to add HTML markup, tables, images, if relevant.
- Be sure to check the generated text. Here you should pay attention to its compliance with the topic, expertise, reliability of the information provided. In some cases, you will need the help of a person who is really well versed in the subject and will be able to understand whether the material created by the neural network is true.
When working with neural networks, collecting information for preparing prompts, preparing semantics, information about competitors and performing a huge amount of related work, mobile proxies from the MobileProxy.Space service will become an indispensable assistant. This is a tool that will ensure reliable protection of your privacy when working on the Internet, prevent all kinds of hacker attacks and any other unauthorized access by replacing the real IP-address and geolocation of your device with its own technical parameters. It also provides the ability to work in multi-threaded mode with social networks, message boards, bookmakers, use programs that automate routine and similar work. And all this without the risk of running into a ban or any other restrictions from the system. You will also be able to bypass any regional restrictions by choosing the most suitable GEO for yourself.
You can study all the features and functionality of mobile proxies at the link https://mobileproxy.space/en/user.html?buyproxy. On the site, we also suggest paying attention to the current tariffs and the fact that the price of use directly depends on the duration of the purchase period. The most advantageous solution in terms of cost is to purchase mobile proxies for a year. In any case, you get a quick and easy renewal, convenient operation, professional round-the-clock assistance from technical support specialists.