After going through Semrush’s 148-page AI report on content marketing, I came across several insights that caught my attention and answered some lingering questions from previous product campaigns. Given its length, I’ll break down the key data into digestible sections.
You can access the full report here.
Before we begin, I’d like to ask readers to consider the following questions:
- How many people use large language model tools at work to improve efficiency?
- Do you think AI can help improve your work efficiency? If not, what are your concerns?
- Of all the content you’re exposed to, how much do you think is AI generated without your knowledge?
Data Source for the Report
The data is sourced from a survey of over 2,600 small and medium-sized businesses.
Here’s a demographic analysis of the respondents:
- Company Size: Most respondents are from small businesses:
- Micro businesses (1-9 employees): 47%
- Small businesses (10-49 employees): 30% *This means 76% of respondents are from companies with fewer than 50 employees.
- Medium businesses (50-249 employees): 16%
- Large businesses (250+ employees): 7%
- Agencies: 1%
- Job Roles: Over half of the respondents hold leadership positions:
- Business owners/founders: 55%
- Marketing managers: 11%
- CEOs: 8%
Are You Using AI?
First, 67% of respondents use AI tools for SEO and content marketing, while the remaining 33% do not.
The motivations or considerations for individuals or businesses choosing not to adopt AI tools in their content marketing and SEO strategies are as follows:
35% of people are simply unaware that AI tools are an option. What about the 65% who know but haven’t started using them yet?
- For 37% of respondents, it comes down to a lack of training or understanding of how to use these tools correctly.
- Another 31% are more concerned about the originality of the end result, while 30% express similar worries about quality.
From these survey results, we can see that a significant portion of people’s feedback on current AI relates to not knowing how to use it correctly. There are also concerns about the risk of plagiarism in generated content, while another group chooses not to trust AI-generated content. The vast majority of respondents focus on questioning the quality of AI-generated content, considering it unreliable or risky.
Are those who don’t use AI satisfied with their current situation?
For businesses not using AI, the answer is surprisingly negative.
- Less than half (49%) of businesses consider their content marketing efforts to be effective or very effective.
- A significant number (43%) of businesses rate their content marketing efforts as just average, while 65% consider them ineffective.
We see similar results when respondents reflect on their effectiveness in attracting organic traffic (SEO).
- Only 18% believe their work is very effective, while the majority (43%) consider their results to be just average.
However, this raises even more questions for me.
If the marketing effect is average, why do these businesses that don’t use AI still refuse to adopt AI?
The following two sets of data are even more thought-provoking:
Interestingly, 40% of users who claim not to use AI can complete a long blog post in just one hour. And 48% of users spend only 5 hours per week on content creation.
These data might represent the following perceptions of these users and businesses:
- They may not consider content marketing important.
- They might be writing hastily just to “complete the task” rather than truly investing in valuable content creation.
- They may be investing most of their marketing resources in other channels, such as paid advertising, while neglecting content marketing.
Another set of data can corroborate my point:Users who use AI (before using AI): 49% spend over 3 hours writing a long article, 35% spend 2-3 hours.
-
Difference in quality awareness: This comparison clearly shows that the user group choosing to use AI already placed more emphasis on content quality and was willing to invest more time even before adopting AI.
-
Significant efficiency improvement: After adopting AI, 36% of users complete long articles within 1 hour, and 35% within 2-3 hours. This contrasts sharply with 49% needing over 3 hours before adopting AI, demonstrating AI’s enormous potential in improving efficiency.
From the data, we can see that those users who valued content quality and were willing to invest time even before using AI tools are the real target user group for AI tools. These users understand the value of quality content, have mature content marketing strategies, and are seeking ways to improve efficiency while maintaining or enhancing quality. This insight has significant implications for the market strategy of AI tools.
Rather than trying to persuade users who don’t value content marketing to adopt AI, it’s better to focus on those who already recognize the importance of content. For this group of users, the key is not to tell them “they should use AI,” but to demonstrate how AI can help them better achieve their existing goals.
Marketing should emphasize how AI can optimize existing workflows, and how it can allow creators to focus more on creative and strategic work, rather than simple word processing. Case studies can be used to show how AI tools can deepen research, improve content structure, or increase content interactivity.
In terms of product development, more focus should be placed on features that can enhance content quality, not just speed up writing. For example, providing better research assistance tools, improved SEO suggestions, or smarter content structure optimization.
Customer education should also shift towards more advanced topics, such as how to integrate AI into overall content strategy, how to use AI-generated insights to improve content planning, or how to use AI to better understand and meet target audience needs.
This approach not only attracts more valuable customers but also builds more lasting customer relationships. Because you’re not just selling a tool, but providing a solution that can truly enhance the quality and efficiency of their work.
What are AI users thinking?
From the above data, we can see:
AI tools are widely used in the content marketing field. 58% of users utilize AI for content and topic idea research, which is the most common use. In terms of content creation and optimization, 52% use AI to rewrite and paraphrase text, while 50% use it for writing from scratch.
Regarding content formats, blog posts (58%) and social media posts (55%) are the most popular types of AI-assisted creation. AI is also used for content strategy planning, with 47% of users employing it to develop content marketing strategies.
Notably, AI plays a role in improving content quality, with 29% of users using it to enhance copy readability and 26% using it to optimize tone.
While text content remains the primary application area, AI is gradually expanding into multimedia content creation, including short videos (31%), images (28%), and audio (7%).
More than half of the users understand AI
What surprises me here is that a considerable number of businesses and users deeply understand the principles and mechanisms behind large language models.
Nearly half of the users adopt the practice of multi-step prompts and additional questions, which actually reflects their grasp of the “Chain of Thought” (CoT) concept. This method simulates the human thinking process, guiding AI to reason step by step, thus resulting in more accurate and logical outcomes. This not only improves output quality but also enhances the explainability of AI responses, which is crucial in many business applications.
At the same time, 41% of users know how to make AI play specific roles, demonstrating their profound understanding of System Prompts. System prompts are key to setting AI behavior and roles, and can significantly alter the style and content of AI output. The widespread application of this technique indicates that users have recognized the flexibility of AI and are exploring how to shape AI into specialized tools suitable for specific tasks.
This set of data reveals the dual role of AI tools in content creation: both as a powerful assistant and as a tool that needs supervision and refinement. The vast majority of users (73%) personally review the tone and style of AI-generated content, and 48% conduct fact-checking, clearly indicating that users maintain a cautious attitude towards AI output. This widespread human intervention not only reflects concerns about the quality of AI-generated content but also highlights the crucial role of human judgment in the content creation process.
At the same time, up to 73% of users modify AI-generated content to some extent. This data strongly suggests that there is still huge room for improvement in AI tools. Users expect more precise output that better meets specific needs, pointing the way for further development of AI technology. However, this widespread modification behavior also reveals an important fact: AI is not designed to completely replace human creators but to exist as a tool for enhancing human capabilities.
If you can’t tell the difference, does it matter if it’s real?
As a professional in this field who has personally developed and designed an SEO content generation tool, I often ponder this question: Is AI-generated content truly reliable? There are numerous debates and discussions on the internet about the validity of this topic. I won’t go into detail here, but interested readers can check the reference links at the end.
Regarding the quality of AI-generated content, I’m reminded of a scene from the TV series “Westworld” that I watched years ago in college –A Host is asked “Is she really ‘real’?”, to which she replies,
“If you can’t tell, does it matter?”
In this report, there’s a very interesting section where they designed an AI test. The survey covered over 700 US consumers and cleverly designed a double-blind experiment to evaluate the effectiveness of AI-generated content. The research team carefully prepared content in various formats, with both human-created and AI-generated versions for each format. Participants were asked to choose the version that resonated more with them, without being told the source of the content. The experiment used advanced AI tools like ChatGPT and ContentShake AI, while also inviting several professional writers to participate in content creation. This human-machine collaboration approach not only ensured the comparability of content quality but also reflected the current trends in the content creation industry.
The uniqueness of this experimental design lies in its direct focus on two core issues of AI-generated content: readability and resonance. By eliminating participants’ potential bias towards AI, the study was able to objectively assess the quality and appeal of the content itself.
Here are 6 sets of experimental data extracted from the report:
-
Blog post introduction (indoor cat food):
AI wins: 54% vs 46%
-
Social media ad (Spanish family resort):
AI wins: 70% vs 30%
-
Blog post paragraph (Indoor cat special dietary needs):
AI wins: 60% vs 40%
-
Social media post (skydiving provider selection advice):
AI wins; 65% vs 35%
-
Social media ad (Social media management and planning app):
AI wins: 53% vs 47%
-
Product Description (Simplified video generation app):
AI wins: 65% vs 35%
In all 6 test scenarios, AI-generated content received higher scores, indicating that AI can now generate content comparable to, and in some cases more popular than, human writers.
The fact that survey subjects made their choices without knowing which version was AI-generated suggests that AI-generated content has reached a level where it can “deceive” ordinary readers, making it difficult for them to easily distinguish between human and machine-generated content.
In my view, the effectiveness of AI-generated content is no longer a hypothesis, but a reality being validated by the market. As more and more businesses successfully adopt AI tools, the feasibility of this technology has been proven through practical applications. The voices skeptical of AI are, to some extent, reminiscent of the doubts about mechanized production during the Industrial Revolution. Back then, some believed that machine-made goods “lacked craftsmanship.” Today, we hear similar arguments claiming that AI-generated content “lacks human creativity” or “has no soul.” However, history tells us that such views often result from a lack of understanding or bias towards new technologies.
From a philosophical perspective, distinguishing between AI-generated content and human-created content is challenging. Imagine a scenario where you propose an initial idea, then use AI tools to refine and modify it multiple times, combining it with your feedback to finally complete a piece of content. This process is like the Ship of Theseus paradox. Just as it’s difficult to judge whether a ship that has been continuously repaired and replaced is still the original ship, we similarly struggle to determine whether content that has undergone multiple AI processing and human adjustments can still be viewed as purely human creation or purely AI-generated content.
Summary and Reflection
This report is actually a boost for AI practitioners. In the current environment where content generation is questioned and controversial, we see from the data that the adoption rate of AI tools is continuously rising. This trend not only validates the practical value of AI in content creation but also injects confidence into the industry.
Interestingly, I suddenly realized an important point: the rise in the adoption rate of new tools doesn’t mean the entire market has been educated. On the contrary, it’s those who adopt early that are eliminating the rest. This insight made me feel that rather than spending a lot of time educating the market, it’s better to focus energy on perfecting the tools to achieve exponential efficiency improvements.
This report also reveals a new model of collaboration between AI and humans. Most users review and modify AI-generated content, indicating that AI is more about enhancing human capabilities rather than simply replacing humans.
Another interesting finding is that users who originally valued content quality tend to make better use of AI tools. This might lead to a polarization in the content creation market: those who use AI well will gain a huge advantage, while those who can’t keep up might be eliminated by the market.
Overall, this report not only provides strong data support for AI content generation tools but also points the way for the entire industry. For us AI tool developers, the key is to perfect the tools and achieve significant efficiency improvements, so we can take the lead in this technological revolution.
In human religion and history, people have always chosen to “see to believe.” Like Thomas in the Bible, who needed to see Jesus’ nail marks to believe in his resurrection. Today, we see a similar situation – those who have personally experienced the enormous efficiency improvements brought by AI tools are proving the value of AI through their actions. Those who are still waiting and watching may soon find themselves lagging behind the times. In this rapidly changing era, what’s important is not to convince everyone to believe in the potential of AI, but to let the results of AI tools speak for themselves.