Marketers Using AI Publish 42% More Content [+ New Research Report]
In this article, we’re looking specifically at AI’s impact on content creation. How common is AI content use? What tools do people use? How often do people edit their outputs?
Download and read the full PDF report here (no email required): The State of AI in Content Marketing.
It includes extra research and analysis on AI content performance and costs, expert commentary from a dozen content and AI experts, and the full research methodology.
Download here.
- AI use for content marketing is extremely prevalent: 87% of respondents use AI to help create content.
- AI use allows companies to publish 42% more content each month: the median monthly publishing frequency using AI was 17 articles, compared to 12 for those not using AI.
- ChatGPT is the most common AI model for content creation, used by 44% of respondents, followed by Gemini (15%) and Claude (10%). In total, 94 distinct AI tools were referenced.
- 97% of companies edit and review AI content. Only 4% of respondents publish “pure” AI-generated content. 80% of respondents manually review AI content for accuracy.
- Blog posts are the most common type of AI-generated content (87%). Brainstorming (76%), outlining (73%), and content updates (67%) are the top AI-assisted tasks.
AI use is extremely common in content marketing.
Out of 879 responses, 769 (87%) reported using AI to create or help create content. Just 111 reported that they did not use AI (13%).
If this seems surprising, it’s worth considering that AI is now a native part of Google Docs, Google Sheets, Gmail, Slack, LinkedIn, you name it. It’s hard to avoid.
AI use was most prevalent at small (10–49 employees) and enterprise (1,000+ employees) companies, and least common at micro (1–9 employees) companies:
AI use was most common in marketing teams of 2–10 people, but even at large marketing teams of over 100 people, 89% of respondents said that they used AI:
What the experts think
“We’re finding that organizations of all sizes are extremely interested in leveraging AI in their marketing initiatives. In fact, smaller organizations have the benefit of being more nimble and fast-paced with their integrations”.
“At the enterprise level, there is a higher level of concern around data sharing within LLMs. Several clients of ours have explicitly forbid sharing of company information with LLMs due to concerns of how it may eventually be used.”
AI allows companies to significantly increase their content creation.
The median publishing frequency for companies using AI was 17 articles per month. For companies not using AI, the median was 12 articles per month, or 42% lower:
Company size can also impact publishing frequency (bigger companies generally publish more than smaller companies), so it’s helpful to look at the difference in publishing frequency by company size.
The data suggest that AI is having the biggest impact on micro (1–9 employee) and small (10–49 employee) companies, while enterprise companies actually publish more without AI workflows.
Given the amount of legal review and compliance that happens in the biggest companies, this isn’t necessarily surprising:
Here’s how publication frequency was distributed among the responses. Although extreme publication frequencies were reported even for non-AI users, the extreme values were much more common among AI users, with multiple companies using AI to help publish 100, 150, or even 250 pieces of content per month:
| Minimum | First quartile | Median | Third quartile | Maximum | |
|---|---|---|---|---|---|
| No AI | 0 | 4.5 | 12 | 40 | 2500 |
| AI | 0 | 7 | 17 | 35 | 10000 |
Here’s a histogram showing the distribution of publishing frequencies, with the 1% most extreme outlier values excluded for legibility:
What the experts think
“Content writers can easily double or triple the amount of marketing content they’re creating by using AI tools. AI-assisted writing is particularly helpful for multi-channel campaigns that require creating different content formats for different channels.”
“AI isn’t replacing human creativity in content marketing—it’s amplifying it by allowing teams to do more. The most successful organizations aren’t choosing between AI and human talent, but are combining both to create more content that resonates with their audience. This hybrid approach is what we’ve been using at Campfire Labs to find a competitive advantage.”
“In an AI-dominant world, the scale of content will grow exponentially. From a cost standpoint, search engines will have to be more judicious about how they allocate spend on each site’s crawl resources—a.k.a. crawl budget. The expectation that large sites are better than small sites will have to change.
Don’t just write content because you can. Don’t create a programmatic scaled SEO effort if there’s no real user reason to have scaled pages.”
Generative AI is used primarily for writing blog posts.
Our research found that blog posts were the most common content type created by AI (87% of respondents), followed by website copy and landing pages (64%), and social media posts (63%).
Longer, more complex content formats like whitepapers (15%) and ebooks (17%) were the least common content formats created by AI. 
What the experts think
“Generating blog content is—for the vast majority of companies—the least challenging content type on their domain. Hardly any branding exercise takes place on blog content. Usually, no one is concerned about maintaining a brand’s tone on the blog. Instead, they’re most often focused on publishing content “to rank”. It’s content that requires minimal expert input (or, at least, used to). Pretty much, keywords and a title and you can get going. It’s content that keeps on giving: It can be promoted on social, newsletters etc. And it’s content that, if it ranks, it will keep on bringing new visitors and potential customers.
“On the other hand, website copy, social posts or newsletter copy require you to be you. You can’t de-prioritize branding here, and getting it right with AI is often a challenge.”
Unsurprisingly then, most AI-assisted tasks relate to blog post creation.
When asked how AI was used in different parts of the content marketing and SEO workflow, 76% of respondents said that they used AI to brainstorm topic ideas. 73% reported using AI to create content outlines, and 67% used it for improving their content.
At the other end of the spectrum, creating images (19%), localizing or translating content (22%), and keyword clustering (23%) were the least popular use cases.
What the experts think
“AI is great for ideation, research, and planning. It can drastically improve the time to complete SEO tasks that previously had to be done manually.”
“With many AI tools offering web search and deep research features, it’s much faster and easier to research, outline and write longform content like blog posts and in-depth articles. That’s the beauty of AI-powered writing, AI helps you across every stage of the writing process, from conducting initial research to making final editing.”
“If you’re not using AI at the start and the end of your content creation process, you’re missing out. Brainstorming, coming up with different angles, and then proofreading or structural editing are things that LLMs are pretty good at.
But I am very sceptical with what ‘optimizing content for organic search’ means. If it means providing keywords and asking it to incorporate them in your content, I think you’ll be in for a nasty surprise at some point. The biggest optimization you can get from AI is for logic and readability. Sure, provide relevant queries and keywords at the start. But focus on optimizing the structure, covering gaps your competitors don’t, formatting content appropriately and more.”
ChatGPT was the most commonly used AI tool for content creation, by a significant margin.
44% of respondents reported using ChatGPT, followed by Gemini at 15%, and Claude at 10%.
What the experts think
“While ChatGPT is a popular tool for content creation, the quality of the content it generates will vary depending on the model you’re using and the quality of the prompts you’re providing. Based on my own usage, I find that Claude, Gemini and Perlexity sometimes generate better quality content even when using simple prompts. So if you’re not happy with ChatGPT’s output, I recommend trying those other tools.”
In total, a staggering 94 different tools were mentioned in the survey. These were the 20 most popular tools mentioned:
We’ve built a content tool that uses AI in all the right ways: AI Content Helper.
AI Content Helper marries human creativity with AI efficiency and Ahrefs’ competitor intelligence. It’s designed to help content and SEO teams scale their content production and improve their search rankings—without sacrificing human creativity and skill.
AI Content Helper’s features include:
- AI content creation. Ask a state-of-the-art AI writing copilot to write paragraphs, fill content gaps, and edit and improve your article content.
- Content scoring. Score the search optimization of your content (and your competitors) from 0–100, and get actionable recommendations for improving optimization for your target keyword.
- Topic optimization. Use AI to identify “topic gaps”: missing information to add to your article to help readers answer their questions and Google rank your content.
- Multiple intents. Many SERPs contain multiple intents, so AI Content Helper lets you pick the exact articles you want to compare with.
- Competitor insights. See content scores, word counts, referring domains, and more for your competitors’ articles, along with the full HTML of each competing page—perfect for seeing how the top-ranking articles are written.
Get started with AI Content Helper now or learn more here.
I’ve always wondered whether I’m losing out by using gpt4o or Claude, instead of a “wrapper” that combines multiple APIs into one interface. According to our research, probably not.
77% of the AI tools listed by respondents fall into the “LLM” category: that is, they are specific LLM models or model providers, like Claude from Anthropic or gpt-4o from OpenAI.
Only 23% of the listed tools are considered “wrappers”: software tools that use one or several LLM models with additional features added, like SurferSEO or Jasper.
What the experts think
“With LLM providers like OpenAI and Anthropic in a race to deploy new models and new features, most companies seem well served by using LLM models themselves, with little need for specialised wrappers. When teams build custom workflows on top of APIs, it’s possible to create very sophisticated workflows using just the basic LLM models.”
Google has softened its stance on AI disclosure (they recommend disclosing “when it would be reasonably expected”—so it’s far from mandatory). With that in mind, it’s not surprising that very few companies bother to disclose AI use.
Of those companies that use AI, only 16% choose to disclose the use of AI in their content. The remaining 84% choose not to disclose AI use:
Disclosure was most common at micro (1–9 employees) and enterprise (1,000+ employees) companies. Enterprise companies in particular are likely to have rules and regulations that make disclosure necessary:
It’s also slightly reassuring that the companies that are the heaviest of AI content are also the most likely to disclose its use, with 26% of companies publishing majority “pure” AI content disclosing its use:
Generally speaking, very few companies publish solely “pure” AI content.
Only 4% of respondents reported that over 75% of the content they published was pure AI content, without any editing or additional input. In contrast, 50% of respondents reported that none of the content they published was pure AI content.
What the experts think
“While there’s a temptation to use the outputs of AI directly on your site’s content, marketers need to resist this. We’ve seen from the March 2024 core update just how damaging AI-only content can be. Instead marketers can utilize AI for initial outputs—but have their teams review the content to actually make it useful for users.”
“It’s extremely dangerous to publish strictly AI written content. While Google says they don’t take a stance against AI content, it’s very clear that solely AI written content is detected by Google and the site and content is eventually punished for it. There’s plenty of research and examples of this happening.”
Almost everyone reviews or edits AI content, in some form.
97% of respondents have some kind of review process for AI content. Manual review was the most common method (80%), followed by working with subject matter experts (35%), and working with an editor (30%). 14% of respondents ask AI to review AI content for accuracy.
What the experts think
“Hallucination is a fact of life for LLMs, but the growing use of RAG and grounding—using third-party resources to check and validate LLM output—is starting to eat away at this problem. AI content will never be perfectly accurate, but neither is human-written content.”
Want to learn how AI content performs, or how much companies spend on AI content?
You can download and read our full PDF report here (no email required): The State of AI in Content Marketing.
جهت دانلود و یا توضیحات بیشتر اینجا را کلیک نمایید