rethinking AI:

the human connection

Thousands of pieces of content, and they all tell the same story: AI can make or break your business. The deciding factor will be the humans you choose to run it.

man standing on a red road looking at a mountain

about our study

We analyzed the performance of more than 2,000 blogs: 1,000 articles written and planned by humans, 500 articles written and planned by AI, and 500 that were “expert-led, AI-powered.” What that means is that artificial intelligence helped optimize each stage of the content lifecycle and the humans that oversaw the prompts, outputs, and optimizations were experts in their respective fields: SEO, content strategy, UX/UI, copywriting, and marketing.

article composition

As part of our analysis, we teamed up with clients, partners, and sister agencies to analyze data across 8 diverse industries: SaaS, Retail, Experiential Services, Digital Publishing, FinTech/FinServ, Commerical Real Estate, Healthcare, and Professional Services. Blogs were evenly distributed throughout the buyer’s journey and labeled as ToFu (top of the funnel), MoFu (middle of the funnel), and BoFu (bottom of the funnel). However, B2B content was the most predominant — accounting for nearly two-thirds of all articles surveyed.

D2C vs. B2B Content

using AI for content planning

To gauge AI’s role in content planning, we tested three critical functions: 1.) identifying content needs (gap analysis), 2.) conducting SEO research (determining keywords to target based on perceived volume & difficulty), and 3.) classifying & optimizing existing content (pillar post strategy).

identifying content needs (gap analysis)

The gap analysis process was significantly accelerated by employing ChatGPT as a virtual assistant. We used a process in which we fed ChatGPT URLs compiled via Screaming Frog. ChatGPT, leveraging its ability to read and understand URL structure & meta content, analyzed the URLs and compiled summaries of the main themes, content types, and potential content gaps.

These insights helped identify missing topics and content needs much faster than humans alone.

Expert-led, AI-powered processes reduced the time needed for manual content analysis from weeks to a few days, accelerating the overall content planning cycle by an average of 4x.

conducting SEO research using AI

While ChatGPT's knowledge was cut off in 2021, organizations could leverage its historical knowledge and predictive capabilities to predict the SERP landscape with 98% accuracy (when compared to paid tools like MozPro).

Additionally, ChatGPT could identify related keywords and long-tail keyword opportunities, while delivering strategic recommendations on which terms to target for both short-term and long-term gains in mere seconds.

Furthermore, by using AI for data processing and analysis, SEO specialists were able to review, process, and report keyword data 3x faster than before.

content classification & optimization

ChatGPT was leveraged to classify existing blog content into various categories and cluster topics by analyzing just a URL string. It was able to identify the main themes and sub-themes, while making linking recommendations designed to achieve maximum impact.

Not only was the classification process faster (taking only a few seconds compared to weeks with humans), but AI also proved to be extremely effective, with a success rate of 99%.

Moreover, by identifying the main themes and sub-themes in each piece of content, ChatGPT provided recommendations for optimizing the existing content. For example, it suggested ways to turn standalone blog posts into pillar posts (comprehensive guides on a particular topic) and cluster posts (a series of related content that link back to the pillar post).

This approach boosted organizations’ SEO strategies by improving site structures, increasing page views, and reducing bounce rates.

AI-Enhanced Content Creation

Using ChatGPT led to a 5x increase in content production for clients piloting the “expert-led, AI-powered” approach. This was achieved through several key capabilities, listed below in more detail.

Rapid Drafting and Editing

ChatGPT can generate draft content at an impressive speed. By inputting a brief or outline, ChatGPT can produce a full draft in minutes, complete with an introduction, body text, and conclusion. Once the draft is created, editors can step in to refine the content, add style, and ensure it aligns with the brand's voice. This combination of AI and human effort greatly reduced the time taken to create initial drafts, leading to a significant boost in content output.

For example, a blog post about "AI in Education" that might take a human writer several hours to draft can be generated by ChatGPT in less than an hour. This gives content creators more time to focus on finessing the piece rather than getting stuck on the initial draft.

Content Variation and Scaling

ChatGPT's ability to understand context and generate relevant text makes it possible to create a wide variety of content quickly. By slightly altering the input brief, we can generate different versions of the same piece of content, suitable for various platforms or audiences. This allows for a scaling strategy where a single content idea can be spread across multiple channels.

For instance, a brief for a white paper on "Data Privacy Challenges in AI" could be reworked to create a series of blog posts, an infographic, social media posts, or even a script for a video.

AI Content Performance

In the below example, “targeted keywords” refer to all keywords deemed as high value by the sales and marketing teams. As such, branded terms were typically omitted from the survey unless results were localized to a specific market or region — or the branded term was considered a market differentiator.

When compared to human-only results, “expert-led, AI-powered” organizations saw nearly a 6x improvement in the number of targeted keywords indexed by Google — and a 300% improvement among non-branded keywords occupying Position 0-10 position.

Indexed Search Terms

Targeted Keywords indexed

The AI-only content, however, was the worst performer of them all. That’s because ⅔ of AI-only content failed to index by Google. By comparison, 99% of “expert-lead, ai-powered” and human-only content indexed.

AI-Only Indexing

When it comes to organic traffic, organizations leveraging an “expert-led, ai-powered” approach saw organic traffic from new articles improve by 3x — based on the average views/article taken over a three-month window.

man in suit riding a rising red arrow

Resubmitting ai-only generated content

When it comes to AI content indexing, the adage that "first impressions matter" rings particularly true. Given the current digital landscape, search engines like Google have become incredibly proficient at determining high-value content from spammy, low-quality pieces.

Consequently, any content generated solely by AI and considered low-value can take as much as three months to be re-indexed by Google.

This extended timeline can have significant implications for any content strategy, making it critical to ensure the quality and relevance of AI-generated content from the outset.

While AI, such as ChatGPT, can produce structurally sound and grammatically correct text, the nuance, context, and emotional resonance that a human writer can provide are still essential. The human connection helps ensure that the content is engaging, unique, and provides real value to the reader.

The Future of SERP

While no one can predict the future with certainty, it’s hard to argue that AI won’t significantly impact the SERP landscape. We’re already seeing it in our beta-testing with Google and Microsoft. When beta-testing Google’s latest AI-powered search engine, what did we find?

a google search result of "home builders near me"

Totally different local rankings. Totally different SERP placement. Totally new look.

What sets it apart from past iterations? A renewed emphasis on the human connection. Third-party education and review sites are prominently featured, pertinent business hours are listed within the SERP, and human-generated reviews and feedback dominate the landscape.