All things are new in the New Year—including how you think about writing marketing content.
The marketing world (and indeed the educational, tech, and publishing worlds) are all abuzz with how the most recent AI content generators are disrupting any field that requires humans to put pen to paper.
There are plenty of ways LLM (large language model) AI chatbots like ChatGPT can be extremely useful to marketers. Although they cannot synthesize the nuances of brand identity and advanced CRO strategy, these tools can be useful for brainstorming.
What is ChatGPT good for?
Use ChatGPT to generate ideas for:
- Email subject lines
- Headings for landing pages
- Intro or closing paragraphs for an email
- A foundation for product descriptions
- Meta descriptions
- Rewording a problematic sentence or phrase
- Alt text to describe images
With the rise of AI availability, marketing automation, and no-code, many marketers find ChatGPT capable of supporting them in day-to-day technical tasks like:
- Custom code snippets
- Generating a formula in SQL
- Error handling
However, if you’re stuck looking for ways to change tones or just get outside your head when brainstorming, we think it can be useful, if at times formulaic.
How does this affect content marketing (and how will this affect how we spend our budget)?
Using ChatGPT for long-form content marketing, especially anything that will be crawled by search engines, is a gamble without an editorial process in place. While there’s certainly a chance your content marketing budget may get freed up somewhat by some efficiencies in brainstorming or rewording, you will still want to use skilled marketers, copywriters, and editors who are all in close alignment with your SEO team for your web pages and blog posts.
Why shouldn’t we use ChatGPT for blog posts and other web pages?
Yes, you could ask ChatGPT to spit out 500 words on your topic of choice and even tell it which keywords to use, what tone to employ, and what demographic you’re writing for.
But the problem is three-fold:
- Planning and goals for a page are still paramount—these tools can’t help you with that. You need to know how a piece fits into a content strategy, including what keywords you’re targeting, your audience and their search intent, and the broader search competitive landscape.
- After it’s generated, the content will need to be fact-checked and edited for brand accuracy, tone, potential bias, calls to action, and everything that makes your copy effective and on-brand. Never copy and paste a large swath of text that has been AI-generated.
- Most importantly, you can bet that Google will continue to monitor pages for quality of content and rank pages accordingly. Google prioritizes E-E-A-T (experience, expertise, authority, and trustworthiness). So, if you’re simply generating content with an LLM (large language model) like GPT3, then you are paraphrasing existing content, which is inherently duplicative in nature and may not be seen as valuable by Google.
Will Google be dinging web pages that aren’t human-written?
It’s unclear whether Google can identify text written by large language models (like GPT3/ChatGPT) at this time, but what is clear is that Google values content that serves the reader, not content written to please search engines. The Helpful Content Update announced in August 2022 aims to prioritize people-first content in searches, so whether your content is human or AI-written, it has to be carefully laid out and perfected to perform well.
Tread carefully if you’re thinking about using AI text generators. Google’s spam policies do address spammy automatically-generated content and “will take action if content is ‘generated through automated processes without regard for quality or user experience.’” In a November tweet, Danny Sullivan (Google search liaison) clarified, “It’s unlikely some AI content is going to feel written by people without some degree of human review.”
One thing we can count on is a continued game of cat-and-mouse as AI creators seek to create undetectable copy, while search engines work to reward those pages that most effectively meet searchers’ intent.
Here’s the takeaway: If you want to ensure that search engines don’t flag your pages as spam, now or in the future, do not incorporate AI-generated copy on your site without a robust content strategy and editorial process in place.
What are the drawbacks of ChatGPT?
On its own main page, this tool discloses that it “may occasionally generate incorrect information,” “may produce harmful instructions or biased content,” and has “limited knowledge of the world and events after 2021.” That means this content is backward-looking and does not consider new perspectives. For many use cases—especially YMYL content or fast-developing industries such as crypto or technology—there would absolutely need to be a step for manually adding in new information and context.
In addition to these, no matter how useful and downright slick this tool may be, there’s a matter of expectation and integrity. Humans still expect anything written to be the product of another human. For better or worse, at this moment, posting large catalogs of AI-generated content may be frowned upon at best or, at worst, considered deceptive and thus cause a loss of trust.
Especially when the understanding between the writer and the person needing the content (i.e., copywriters and their employers, students and their teachers, advertisers and their clients) presupposes that you put your own thought, research, and experience-honed skills into your words, doing anything else is a severe breach of an unwritten contract.
What happens when that understanding does not exist? Like customers reading ad copy, emails, or product descriptions?
Here’s where it gets tricky. Teams will need to decide for themselves where the line should be drawn on their own marketing copy, but consider this:
- If the reader knew it was AI-generated, would they care? Whenever AI-generated content is user-facing, you should disclose its use.
- Will the text be thoroughly edited for tone and brand? (It definitely should be, ideally by a professional editor.)
- Is the person in charge of content an expert who makes AI-generated content better? Or are they using it as a crutch?
- Are your expectations for the volume of output and spending on content going to change? (And if they do, would you be able to pivot back quickly when ChatGPT becomes monetized or other hurdles appear?)
At Intrepid, we will continue to monitor and test chatbots like these and, more broadly, the development of LLMs, determining where they can provide efficiencies and where they should be avoided in favor of manual, skilled writing.
Is your company in a downturn? Cutting SEO budget may cost more in the long run than just preserving it. Instead, plan smartly. Use these SEO best practices to know how to cut back carefully.
You’ve likely heard the buzz around tools like ChatGPT, GPT-4, and other LLMs (large language models). But what exactly are LLMs, and how can you use these AI-tools for SEO and content development? Let’s look closer at this intersection of
Sign up for the Intrepid Digital newsletter for the latest digital marketing news, tips, and more