Guide · Marketing
ChatGPT for Marketing Copy: Safe Ways to Use It
ChatGPT for marketing copy works best as a fast first-draft helper, not as a writer you trust without checks. OpenAI released the tool on 30 November 2022, and it is free during a research preview. This guide shows where it helps, where it can go wrong, and how to review a draft before it goes live.
What ChatGPT is, in plain words
ChatGPT is a chatbot from OpenAI. You type a request, and it writes a reply in a chat. It can answer follow-up questions, admit mistakes and refuse some requests.
OpenAI released it on 30 November 2022 as a research preview. The company said it wanted user feedback to learn the tool's strengths and weaknesses. During the preview, use of ChatGPT is free.
That is the key point for marketers. The tool is new, and its maker calls it a preview. Treat it as a draft partner that still needs a human editor.
Sources: OpenAI, ChatGPT: Optimizing Language Models for Dialogue, 30 November 2022
Where it helps
ChatGPT is useful when you need many rough options fast. A person then picks the best and fixes it. Good jobs include:
- Headline and subject line ideas, so you start with ten choices instead of one.
- A first draft of a product description from facts you give it.
- Rewording a long paragraph in shorter, plainer sentences.
- Ad copy variants for an A/B test.
- Outlines for a blog post or a landing page.
In every case, you supply the facts. The tool shapes the words. That split keeps the risk low.
Sources: OpenAI, ChatGPT: Optimizing Language Models for Dialogue, 30 November 2022
Where it can go wrong
OpenAI is open about the limits. Its own notes admit that ChatGPT can write answers that sound believable yet are wrong or make no sense. The company adds that a slight change in how you word a request can change the result. Ask the same thing twice and you may get a different answer.
For marketing, that creates real risks:
- Wrong facts. A draft may give a wrong price, date, feature or statistic, and it will sound sure.
- Made-up claims. It may write a bold promise your product cannot keep.
- Sameness. Many brands using the same tool can end up with similar copy.
- Weak voice. The draft may not sound like your brand unless you edit it.
A wrong claim in an ad or on a web page is your problem, not the tool's. You publish it, so you answer for it.
Sources: OpenAI, ChatGPT: Optimizing Language Models for Dialogue, 30 November 2022
What Google said about AI content
On 8 February 2023, Google published guidance on AI-generated content. Its main point is that Google rewards high-quality, original content, however it is made. The post says that appropriate use of AI or automation is not against its guidelines.
There is a limit. Google says using automation, including AI, mainly to manipulate search rankings is a breach of its spam policies. Its spam policies list spammy automatically generated content, meaning text made without adding enough value and made mainly to rank.
Google also tells writers to focus on people-first content. It suggests asking three questions: who created the content, how it was made, and why it was made. It also says that giving AI an author byline is probably not the best way to tell readers when AI helped.
So the question is not whether a machine helped. The question is whether the page is useful, accurate and written for readers.
Sources: Google Search Central Blog, Google Search's guidance about AI-generated content, 8 February 2023; Google Search Central, Spam policies for Google web search
A five-step review before you publish
Use this check on every AI draft. It takes a few minutes and protects your brand.
- Give it facts first. Paste in your real product details, prices and terms. Do not ask it to supply them.
- Check every fact. Compare each number, date and claim with your own records or an official page.
- Cut what you cannot prove. Remove any promise, ranking or customer claim you have no evidence for.
- Edit for voice. Rewrite lines so they sound like your brand and speak to your reader.
- Add what only you know. Include a real example, a customer detail you can back up, or advice from your team.
Name one person who signs off on each piece. If no one owns the final check, mistakes slip through.
If an agency offers to use AI for your copy, ask how it checks facts and who signs off. Ask the same of any freelancer. Our guide to brand voice can help you judge whether the final text still sounds like you.
Sources: Google Search Central Blog, Google Search's guidance about AI-generated content, 8 February 2023
A simple rule to keep
Use ChatGPT for speed, and use people for truth. Let it make the first pass and the extra options. Then have a person check the facts, fix the voice and own the result.
The tool is only a few months old. Rules and advice will change as more is learned. For now, a human check before every publish is the safest habit.
Sources