A lot has happened in the year since we published the first part of our AI content study. Download our free white paper now with the results of our “AI Content Study Part 2: AI Texts and Google Rankings: How well does AI content rank?” Download it now. In it, our eo AI task force evaluates the ranking ability of AI texts. You will also find instructions on how to optimise your AI-generated texts.
Table of Contents:
Foreword
Short facts about the eo AI content study part II
Study description
Evaluation of the texts by eo AI task force
Evaluation of the texts by experts
Evaluation by Google – seo performance
5 steps to ranking texts
Reality check: How much time does AI really save?
Conclusion & outlook
This whitepaper highlights the potential and limitations of AI-powered content in terms of SEO. You will learn how AI content ranks on Google, what challenges it poses, and what steps are necessary to make it rankable. We also discuss specific technical, strategic, and creative aspects that are crucial for creating high-quality content.
Study description: What, When, How, Who? Procedure and more
During the test period from 1 May to 31 August 2024, the eo AI Task Force generated 118 website texts using the chatbots ChatGPT-4o (OpenAI) and Claude 3.5 Sonnet (Anthropic). The texts created by the chatbots were evaluated, assessed and compared with content written by experienced copywriters at eology GmbH:
Note
The term ‘human-created’ requires a definition that provides clarity. Not every person has the ability (knowledge of the German language, editorial training, etc.) and talent (affinity and aptitude for writing) to create high-quality texts. In the case of the eo AI content study, SEO knowledge is an additional factor.
For this reason, the AI texts in the eo AI study are compared with texts written by trained SEO editors and evaluated according to the quality standards that eology Content Creation follows for high-quality, value-added content with high ranking potential.
The primary goal of the eo AI Content Study Part II is to evaluate the ranking abilityof AI texts. Google evaluates content according to various criteria to ensure that users receive the most relevant and high-quality results. A key benchmark is the so-called E-E-A-T model (Experience, Expertise, Authoritativeness, Trustworthiness). These criteria are used to evaluate the quality of content and the trustworthiness of sources. According to this model, website content must convey expertise, experience, authority and trustworthiness. If this is not the case, websites rank lower or are even penalised. This can affect the entire domain and not just individual pages with poor content.
It is particularly critical when E-E-A-T is not fulfilled in so-called Your Money Your Life content. These are pages that deal with topics such as health, finance and legal information. The topics for which texts were created using AI for the study range from soft topics such as ‘celebrating children’s birthdays’ to medical topics such as the disease scarlet fever or legal topics such as protection against dismissal during pregnancy.
The most important point of the eo AI Content Study Part II was to examine whether AI texts can meet Google’s E-E-A-T model. This examination was carried out in three steps:
Evaluation of the degree to which the target group orientation criteria of the SEO quality standard were met by the experts of the eo AI task force
Evaluation of E-E-A-T by experts from the fields of law, medicine and finance
Evaluation of ranking ability via the test domain die-familie.org
A brief summary of the study results!
AI texts in detail: Answers to the most pressing questions
Can AI texts rank on Google?
Yes. The study shows that AI-generated texts are fundamentally rankable. Content from Claude 3.5 Sonnet in particular achieved strong top 30 rankings. However, AI texts are no substitute for quality checks – without human oversight, SEO and content errors can impair performance.
Do AI texts meet the E-E-A-T criteria?
When it comes to informational topics, AI texts achieve high fulfillment rates of over 80%. However, in sensitive areas such as medicine, law, or finance (YMYL), weaknesses in data accuracy, depth, and trustworthiness become apparent.
Are AI texts as good as human SEO texts?
AI texts provide solid basic content, but often remain superficial, contain redundancies, or include incorrect facts. Human expertise is crucial for ensuring that content is technically accurate, emotionally appropriate, and strategically well-prepared.
How much time does AI really save?
Depending on the type of page, the time savings range from 37% to 53%. The greatest savings are achieved in research and initial drafting. However, editing, SEO optimization, and fact checking remain human tasks.
What are the biggest risks with AI content?
The biggest risks are hallucinations, incorrect figures, and a lack of emotional intelligence. Content errors can have serious consequences, especially when it comes to YMYL topics. AI content should therefore never be published without being checked.
Would you like to talk to our experts about AI visibility?
As this is an empirical study based on a limited text corpus with the aim of assessing SEO performance potential, other methods, test scenarios and use cases may yield different results. The eo AI Task Force welcomes constructive feedback, professional exchange and discussion.
In this whitepaper, we highlight the different SEO challenges faced by site owners and e-commerce decision makers, and show possible solutions that can enable sustainable optimization for long-term success. Download all the info! ... Continue readingSearch Engine Optimization for more Web Traffic
Large companies and corporations face far greater challenges than SMEs when they want to become active in search marketing. In our new whitepaper, we describe the potential of this marketing discipline and how it can be exploited. ... Continue readingSearch Marketing for Groups & large Corporations