No background in copywriting or AI? No problem! Learn how to write compelling ad copy that will launch your campaign out of this world with the help of ChatGPT.
You don’t have to be a marketing genius to write ads on your own. While experience certainly helps, chatbots are paving the way for people with no real background in marketing to draft compelling copy.
Copywriting is a fundamental part of advertising. It’s what gets your product or service in front of your audience and lets them know you’re the best solution to their problem. Persuasive copy inspires viewers to take action and ultimately drives more sales.
But what if you don’t know the first thing about writing ad copy or know where to begin? Worry not!
You’ve got the ultimate writing assistant at your disposal—ChatGPT.
What are the best practices for writing ad copy?
Using chatbots to write ad copy is entertaining, to say the least, but how do you know if it’s adhering to copywriting best practices?
Get your copywriting advice straight from the horse’s mouth. Google’s comprehensive article detailing the best practices for copywriting is a fantastic starting place. This is a foolproof place to start.
You can always ask the language models for copywriting best practices. When I asked ChatGPT this question, the output mimicked the suggestions outlined by Google, including:
- Using clear call-to-actions and keeping things simple
- Testing and adjusting accordingly
- Leveraging emotional language
- Honesty and transparency
- Knowing your audience
Understanding your audience with ChatGPT
The first step in the process isn’t writing. All good writing starts with an idea. When we’re contemplating our ad copy strategy, it’s critical to understand our target audience and their pain points.
Chatbots have a knack for brainstorming about the people who may use a given product or service. One of the easiest ways to start ideation is by copying and pasting the text from the website we’re planning to promote into ChatGPT.
Example prompt: “Who is the target audience based on the following: [copy and paste the text from the landing page].”
In the example above, I copied the text from Perfect Search Media’s Content services page.
ChatGPT has limited short-term memory and recalls past parts of the conversation within the same thread. Review the output and if the language model is wrong, explain the error so it can fix its mistake before proceeding.
After confirming the correct target audience, we can ask about the issues our product or service solves for them.
Example prompt: “Create a bulleted list of the pain points the target audience has this company solves”
Once we identify the problems we’re solving, we’re able to figure out which pain points we’re going to use in our messaging and segment our efforts into separate campaigns as needed.
And with only a couple of prompts, we have a seed list of customer issues relating to our landing page and the target audience’s pain points to theme our ad copy around. Pick and choose the themes that align the most with the conversations you’re having with customers.
Using generative AI for ad copy creation
Once we’ve decided on the themes for our ad groups, we’re ready to begin drafting compelling copy.
Even though we can use chatbots for writing any type of ad, we’ll be discussing Google because the advertising giant accounts for 70 percent of total global ad investments.
For Google search ads, we need up to 15 headlines per campaign. We’re allowed up to 30 characters per headline including spaces.
Example prompt: “Please write me 15 headlines based on [campaign theme], 2-5 words each”
Some of these will be over 30 characters. Pick and choose the best ones, paste them into a spreadsheet, and edit accordingly. When we need more, all we have to type is “continue” and it’ll keep adding to our list. Be selective with headlines and work through several requests until you find enough winning choices.
We’ll also need up to four headlines for Google ads that are up to 90 characters long.
Example prompt: “Please write me 10 descriptions for the ad campaign based on the headlines above, remember to include a call to action, 12-17 words long”
Some will be better than others. A few of the descriptions will require more work to cut down to our 90-character maximum limit. Tell it to keep going until you have a proper list of options to work with. Then choose the best ones.
Analyze customer conversations to improve ad copy
Ironically, landing pages aren’t always the best source of inspiration for ad copy.
However, highlighting insights gleaned from customer conversations is an extremely effective strategy for writing ads.
A hot take from a client can be stronger material than we could ever imagine on our own.
Instead of copying the text from a landing page, try feeding the bots a transcription of your last product demo, customer testimonial, email chain, or sales call.
Overcoming limitations of Generative AI for ad copy creation
AI technology is improving by the day. However, there are still some limitations to generative AI. Before we dive into cranking out copy with ChatGPT we need to go into it knowing the shortcomings.
ChatGPT struggles with counting characters
ChatGPT is wonderful with words but falls short when it comes to counting letters. Asking for a maximum number of characters (including spaces) still doesn’t seem to work quite right, even though this will inevitably improve with time.
Solution: For length restrictions, Specifying the number of words instead of characters helps.
While they’re incredible writing assistants, chatbots are not autonomous. They have a history of confabulating data—aka hallucinating. Consider this your disclaimer: All of the output will require human review.
Solution: Fact-check everything. Make sure it didn’t manifest something irrelevant. Don’t expect to copy and paste the output without revisions.
Limited or inconsistent memory recollection
Even though it has short-term memory, it has a funny habit of forgetting what it was just told. Typing as much learned information back into the platform with each succedent prompt keeps the output in alignment with our goals.
Instead of “Please write me 15 headlines for an ad campaign”, enter “Please write me 15 headlines for an ad campaign based on [campaign theme], 2-5 words each, for [target audience], that goes to this page [pasted text from website].”
Solution: Use highly detailed prompts. The better the prompt, the more useful the output.
The chat history can stunt creativity
Chatbots are not perfect instruction followers. While the chat history usually helps, trying the same prompt without the old conversation can be surprisingly effective in some cases.
Solution: If the answers are becoming too repetitive, ask the same prompt in a new thread.
Some platforms are better than others
Even though they’re similar, there are some instances where Bing Chat outperforms ChatGPT and vice versa. Every AI platform has its strengths and weaknesses.
Solution: Try more than one and see which you prefer.
There are network errors on ChatGPT
ChatGPT broke records for the fastest user growth reaching 100 million users in only two months. And even though there are tons of resources allocated for this software, the ChatGPT server gets bogged down and shows network errors.
Fixing it is usually a case of refreshing your browser. Although, there are some times it’s unavailable to free users for longer periods.
Solution: Refresh your browser or wait and try again another time.
Having a conversation with ChatGPT about the pain points of your target audience can give you loads of insight into the best themes for ad campaigns. Just as analyzing customer conversations can do. But since language models don’t truly understand feelings, the output should be met with scrutiny and requires a thorough review.
With the right prompts, pasting website text into ChatGPT is an extremely effective way to brainstorm new ideas for ad copy. And while the technology is still new and has limitations that will eventually improve, there are plenty of ways to work around the shortcomings that exist.
The most important thing to remember about leveraging language models for crafting copy is the output is only as good as our prompt. It really is a matter of garbage in garbage out.