If you’re a marketer, you may have seen the same LinkedIn posts we have – about some guy who used AI to generate 15,000 SEO-optimized landing pages and got a huge spike in traffic.

And you probably thought the same thing as us: I know AI can produce content faster than I can, and website traffic sounds great … but my audience won’t tolerate content that sucks. I don’t want to damage my brand by pushing out garbage.

So you need to find the middle ground. Use AI to develop efficient, SEO-optimized content … but maintain human authenticity to deliver value.

At Section, we’re finetuning that balance. Humans (Greg, Taylor, Chase, and Azalea) write our content, like this newsletter, but we leverage AI to repurpose that content in strategic ways.

Today, we’re sharing what we’ve learned about using AI for content without compromising quality, and how we used it to get ROI in our SEO efforts.

Why use AI in the first place?

SEO-optimized content has a clear business benefit: it gets eyes on your assets without any ad spend. But it’s also cumbersome. It used to involve an experienced SEO expert ($$$) and a bunch of SEO tools (more $$$), but AI is changing that.

SemRush’s research shows that 39% of marketers saw increased organic traffic after publishing AI generated content, and 64% said it performs the same or better than human-written content.

We were skeptical because we have an ultra-smart, discerning audience. So in 2024, we experimented. We used Claude to optimize (and in some cases, generate) our SEO content, with the following results:

• We brought in over 10,000 new organic site visits
• We ranked in the top 10 positions for each keyword
• And each optimization took 30 minutes or less

The key to our success: Using AI to repurpose great content, not create it from scratch.

We feed Claude content that our team has researched, written, and edited to reference, then prompt Claude to repurpose that content into new assets.

Step 1: Choosing the right reference material

Providing resources in your brand voice gives AI something to mimic, and reduces the likelihood that it will make stuff up. Both of these things will significantly cut down your editing time.

For example, here’s how we created this guide to rank for the keyword “What is Prompting?”.

1. We uploaded the workshop slides for our Intro to Prompt Writing course and a blog post we’d previously written on 8 secrets to great AI prompting

2. We told Claude to draft the guide using those resources, and to optimize the content to rank for the keyword “What is Prompting?”. Here’s a prompt for that:

“I would like to draft a blog post of [X] words to rank for the keyword [KEYWORD]. Please reference the attached resources for both content and style and draft this blog post, focusing heavily on SEO best practices to ensure the piece will rank highly for [KEYWORD]. Include sections on [X], [Y], AND [Z].”

Our advice: Be direct about how you want the reference material to be used. If you want this SEO piece to be a true content repurpose, make sure you’re direct about AI only pulling content from the resources you uploaded. And scan what it gives you back to ensure it adhered to your requested word count (sometimes it cuts it short) and outline.

Step 2: Using AI to optimize your content

Never take AI’s first offer. Claude will give you a pretty good first draft, but it needs additional prompting to optimize for SEO / keyword ranking.

Here’s how to optimize your draft for SEO

1. Ask Claude to assess its draft to determine how optimized it was for your keyword. Then ask it to provide suggestions for further optimization. Here’s a prompt for that:

“Reference the draft you just wrote and assess how optimized it is for the keyword [KEYWORD]. Then provide suggestions for how the piece can be improved to rank more highly for the keyword [KEYWORD].”

2. Ask Claude to integrate its suggestions into a more optimized second draft with the prompt:

“Based on the suggestions above, rework your original draft to include as many of these suggestions as possible to make this piece more optimized for the keyword [KEYWORD]”

Our advice: At this point, some optimizations still have to be done by a human – like internal linking and adding in visual assets.

Step 3: Have AI take care of the little lingering details

With SEO, the little things add up (and they take time).

Once you’ve finalized the content, here’s how to add the finishing details:

1. Ask Claude to generate a slug (read: URL), title tag, and meta description that is descriptive and optimized for your target keyword. Here’s a prompt for that:

“To accompany this content, please generate a URL, title tag, and meta description for this content that is optimized for the keyword [KEYWORD].”

2. Add images to your blog post to boost SEO. If you don’t have any relevant images to add, ask ChatGPT to make one through its DALL-E integration. Just upload your completed blog post and then choose a section or a theme that you’d like it to create a graphic for. Here’s a prompt for that:

“Review the attached blog post and create an image for the section titled [SECTION TITLE]. This should be an [IMAGE TYPE (e.g. flow chart, table, illustration)] focused on [IMAGE SUGGESTIONS (e.g. my tips on how to prompt)].”

3. You have to have alt text if you’re going to add an image to an SEO piece, so ask AI to generate some to go along with the graphic it created (or you had on hand). Here’s a prompt for that:

“Based on the attached blog post and the image you just created / the attached image, draft alt text to accompany this image that will help me rank highly for the keyword [KEYWORD].”

Our advice / warning: DALL-E can’t spell to save its life, so you may want to ask it to avoid text in your prompt or be prepared to iterate with it. You’ll also get better results if you share a color palette, an image style that you prefer, and images your own team has created for reference.

The bottom line on using AI in content generation

Our tests proved the ROI of using AI for high-effort, low-risk SEO content: We ranked in high positions for the keywords we were targeting and spent way less time on production.

But these results aren’t going to translate to every piece of content you write. We don’t use AI to draft these Strategy Briefs, and we won’t because giving you authenticate, valuable insights matters to us – and we can’t replicate our human ability to determine what’s valuable through prompting.

If you’re wondering where else AI can slot into your content efforts, try out a matrix like this one:

And when it comes to SEO, SemRush’s research – coupled with Google’s public stance on AI-generated content – demonstrates that AI-generated content isn’t penalized, low-quality content is. So it still matters if what you generated is valuable, even if AI says it’s optimized.

Keep a human in the loop – that’s led to good results for 52% of SemRush respondents, and for us.