This article was originally published in July 2024. Since then, major improvements have been made to LLMs that make them more viable for additional content use cases. Some updates have been made to adjust for accuracy as of May 8, 2025.
When we first published this blog post in July 2024, the marketing community seemed to have reached a rare moment of consensus on ChatGPT: it produced mediocre (at best) content.
But even then, most of us recognized LLMs as amazing tools, and extremely helpful when it comes to finalizing quality content. The trick has always been to lean into the actual strengths of LLMs - plus adapting your workflows every few months to leverage the latest technology from LLM providers like OpenAI, Anthropic, and Google, among others.
Humans vs AI for writing: why not both?
ChatGPT can piece together competent sentences, and Claude can write some solid prose, too. Most would say that the highest quality content — the kind that people want to share with friends — requires a novel point of view, the nuance of human experience, through a lens of introspection that only the writer themselves can bring to the page.
Are humans the only ones equipped for this task? Will AI be able to do it some day? Maybe that day is already here? We say: it’s less about how artisanal your content is, and more about how you’re using AI to do your best work faster.
To do that, marketers have to keep pace with AI advancements to make sure they’re working as smart as possible with the machines.
In this post, we’ll look at use cases, prompts, and related tools we’ve been using to get the most out of tools like ChatGPT and Claude. Our goal has always been to make LLMs an agent of quality and velocity rather than “scaled content abuse.”
ChatGPT is a solid editorial assistant
ChatGPT can take any sort of structured data and present it in novel ways. It’s particularly good at things like:
- Structuring an argument (”Help make this braindump more convincing”)
- Condensing language (”Write this in half the words and be concise”)
- Brainstorming sentence structure (”Mix up the sentence structure of these 3 sentences” or “substitute the word “x” for something else so it’s not repeated across sentences.”)
- Ideating on subheadings for blogs (”Write a subheading for this section of text for a first-page ranking article on bumblebees”)
When you give LLMs strong context to work with, everything gets better. The richer, more original, and uniquely yours that context is? The better output we’ve experienced from LLMs.
More editorial use cases for ChatGPT
The quality of ChatGPT’s output is fairly dependent on the quality of its input. When you provide ChatGPT with a complete piece of (well-written) content to update, you’re providing it with the context it needs to do solid work for you.
Here are a few tasks we offload to ChatGPT with a little human-editing in the mix:
- Optimizing page titles for target keywords. With blog copy and a target keyword, you have everything ChatGPT needs to give you some solid title options.
- Inserting headline tags for structure and user experience. “Section should have 1-3 paragraphs and each paragraph should have 2-4 sentences.”)
- Condensing content within a post. Cutting and condensing is a time-intensive practice but ChatGPT is pretty good at doing a quick, effective job.
- Adapting transcripts from video + audio. Webinars, podcasts, video tutorials are great source material for written blog posts. When provided with a brief summary, ChatGPT can shape long transcripts into structured blog post drafts.
Again, this is all playing to its strengths of data analysis and sequencing. ChatGPT can produce clean, revised copy in this way because we’re providing original data, composed by a human.
Head of marketing Megan Dorcey suggested using GPT as a little research intern back in July 2024:
Example: a prompt for condensing content
Scenario
Let’s say a content strategist submits a case study draft that runs around 3,800 words, but the ideal length is closer to 2,200. One section – labeled “Customer Journey” – is particularly overstuffed. It’s 900 words long and overloaded with nested bullet points, tangents, and repetitive phrasing.
You can paste that section into ChatGPT and prompt it with something like…
Prompt
I'll provide you with case study copy. Condense the copy into a bullet-pointed list. Each bullet point should identify one key stage in the customer journey and include 2 sentences: the first sentence describes the stage, and the second sentence explains its importance within the overall customer experience.
Using page performance data to identify AI optimizations
The optimization tactics you assign to ChatGPT will vary according to the needs of a given page. Those decisions are determined by various performance metrics.
Pages with declining traffic
You can quickly identify pages with declining traffic in the ércule app. Often times these pages are suffering from content decay and just need a little love. Prompting ChatGPT or Claude for ideas on how to bring the page back up in rank for its target keyword or topic can inspire plenty of ideas for an impactful page update.
Lead-generating pages that could be doing more for you
Most marketers know their top lead-generating pages (and if you don’t we should talk.) In the ércule app, you can create a Collection of pages that are your top performers to keep a close eye on - especially lead-generating pages that are doing well in organic search.
To maintain ranking for any of these important pages, you can use ChatGPT to help you generate ideas for what new content to add to the blog posts, like FAQs for the bottom of the post. Or even better: ask ChatGPT Deep Research to perform a CRO analysis for you and implement its suggestions. (We’ve tried it, and it’s not half bad!)
Pages with great engagement metrics but subpar search rank and traffic
We call these pages “wallflowers” in the ércule app. You'll likely need to optimize metadata in order to gain traction for a specific search query. This would be a great instance for bringing in ChatGPT. Let it suggest new titles, headings, and meta descriptions.
It could even help you to batch out this work across your blog. Use the ércule app to find and export a list of every one of these "wallflower" pages on your website. Then bring in ChatGPT to revise them all.
Moving from bespoke AI optimizations to scalable content systems
For many marketers, ourselves included, prompting ChatGPT or Claude directly through its chat interface is a quick way to accomplish a single content task—but there’s so much more you can do with AI since we originally published this article in July 2024.
ChatGPT, Claude, and Gemini all have free or low cost research tools that can browse the web for you and perform incredibly in depth, cited, multi-step research.
And what we’re most excited about: AI automation platforms like AirOps and Gumloop are making it accessible for AI-enabled marketing teams to build agentic workflows that touch every layer of the content stack.
These days, most of our AI time is spent with the latter: building content systems that scale your expertise without scaling your team size.
If you want to hear more about what we’re up to, sign up for our newsletter and/or reach out for a chat.
