I guess you can say I’m an early adopter when it comes to the use of AI. I wouldn’t consider myself an AI expert by any stretch of the imagination, but I’ve looked for ways to leverage it since ChatGPT first became a thing in 2022.
This isn’t without some resistance, of course. Change and evolution are hard. But I’m also conscious of the many dangers of AI. Beyond the fact that automation could make many advertising functions irrelevant, there are brand risks related to low-quality content and creative slop.
So I’ve ventured into this new world with an open mind, but cautiously. I wouldn’t bury my head in the sand. But I also wouldn’t use it carelessly in ways I may eventually regret.
Now that we’re nearly three years into this, I thought it might be useful to share how I’m using AI today, as both an organic content and Meta advertising companion.
General Approach and Concerns
First, let’s outline my general approach to leveraging AI and my overall concerns. When it comes to using tools like these for your business, it’s important that you first map out the potential risks, rewards, and overall comfort level.
The market is flooded with AI-first tools, and even tools that may have nothing to do with AI but are attempting to cash in on the buzz. You can’t, and shouldn’t, use them all.
With that in mind, here is how I’ve approached using AI tools…
1. AI will not replace me as a content creator.
We are now in the 15th year of my business, and a whole lot has changed during that time. But content creation makes my business go. Blog posts, videos, and podcast episodes attract people to me.
To be clear, I absolutely could use AI to replace me as a content creator. We’ve seen how easy it is since the Internet is flooded with AI-generated websites, videos, and social media interactions. But I don’t want that.
I enjoy this part. I don’t want a tool to create this content for me. Not only is there a risk of losing my unique voice in the process, I value the challenge of content creation. It exercises my brain, and I don’t want to eliminate that.
2. Use AI to increase accuracy and efficiency.
I am going to continue creating my own content, but I can use AI to make me better at it. It’s beneficial if it helps me avoid errors and skip or speed up steps that otherwise limited me.
Examples include article editing, video editing, and brainstorming content ideas to get out of bottlenecks. My content will still undeniably be me with my stamp, voice, and unique style. AI is the tool to help me do it better and more accurately.
3. Use AI to replace tasks I don’t want to do.
I’m not talking about replacing people I’d normally hire. I’m talking specifically about the tasks that I’ve always done myself that I’m either not that good at or don’t particularly enjoy. These would be high priority candidates for using AI.
4. Avoid slop.
Now that the initial novelty of AI has worn off, the cracks are much more evident than they may have been initially. We are flooded with AI-generated slop: Social media posts and comments, articles, images, and videos. It’s slop because it’s easy to create, produced at high volumes, and low quality.
This is where we enter that complicated middle ground of AI. Yes, it’s great for creating content at a higher volume. But the danger is that we end up creating more for the sake of creating more, and the low-quality content becomes a brand risk.
I’m reminded every day of what I don’t want to be on social media. I understand the allure of using AI to create more content and engage more. But so much of this stuff is templated trash that lacks any soul, and it’s obvious.
Because of how easy it is to create content now, it’s on one hand difficult to stand out because it’s all so watered down. But on the other hand, human-guided connection feels so much more valuable now.
And I understand that at some point AI will be better and the non-human driven content will be more difficult to recognize. But for now, human-created content stands out among the slop.
With this as a foundation, here are some specific ways that I’m leveraging AI for both organic content and advertising strategy…
Organic Content Assistance
I’ve run a content-driven business for more than 14 years now. While quite a bit has changed, I’m also very process-oriented. I have a very specific weekly content schedule that involves repeatable steps. It’s a cadence that happens every week, and I’ve found ways to churn out content at a very high rate by sticking to my process.
But there comes a point where you can’t possibly get more efficient without sacrificing quality. At least, that is true without some help. “Help” could be hired or AI-assisted.
Here’s a list of the various ways I’m using AI to help with content and website experience…
1. Website chatbots.
I spotted an opportunity to improve user experience on my website using AI early on. One of the challenges is that I’ve created so much content that it can be overwhelming to visitors looking for help. An AI chatbot trained on my content could be a helpful extension of me.
After going through many iterations of tools, I now get help from the following two chatbots…
JonBot for Members-Only Access: This might be my favorite use of AI. JonBot is a website companion that is trained on more than 700 pages of my content. I’ve refined it continuously during the past two years to make sure that it doesn’t stray away from my own content and philosophies while being sure it’s not giving outdated advice.
JonBot was initially available to all visitors of jonloomer.com, but it simply became too popular. I started running into usage limits and increased costs. It also became evident that people were using it in ways I had not anticipated to help run their own businesses.
As a result, I moved JonBot to a PHC – Elite membership benefit. And that’s been a good decision since members who don’t want to ask our community countless questions can instead bounce ideas off of JonBot.
Content Assistant: When I removed JonBot from jonloomer.com, I wanted to replace it with something scaled down. It would help guide people, but it wouldn’t be the conversational resource like JonBot. If you’re looking for content that can help you with a specific type of problem, this bot will recommend a series of related links.
To do this, my Content Assistant is only trained on the titles, links, and descriptions of my content. It is not allowed to give advice. And if you somehow trick it into giving you advice, don’t expect it to be good advice since it’s not actually trained on the text of my content.
2. Blogging Workflow.
I’ve written well over 1,000 blog posts on these pages. I have a system now, but ChatGPT has become my partner to help me through some bottlenecks.
Obvious Spelling and Grammatical Suggestions: When I’m done writing the draft of a blog post, I’ll copy that text and paste it into a ChatGPT project that helps with this task. I don’t want ChatGPT to rewrite my entire article or change my voice. Instead, I’ll first ask it to highlight specific areas of improvement related to obvious spelling and grammatical errors.
By doing it this way, I can go through each recommendation and make the conscious decision of whether the change is better than the original. In some cases, I’ll keep the original to preserve my voice. In others, I’ll see the need for improvement, but make an adjustment different than what was recommended.
Title Suggestions: Coming up with a title is one of my least favorite things. It needs to be SEO-friendly, descriptive, and provocative enough to inspire a click. But I also don’t want click-baity headlines either. I’ll prefer boring and descriptive, but I also understand the value of just enough mystery to get attention.
One of the initial tasks I give ChatGPT when I provide my article draft is to come up with five title ideas. Over time, it’s also come to understand my style and preference here as well.
In most cases, I’ll already have a working title in mind. Sometimes I’ll use one of the suggestions. Other times, I’ll use the suggestions as inspiration to improve my title. Otherwise, I might just stick with the original.
150-Character Metadata Descriptions: This is one of those braindead tasks that I could do myself, but I’d rather not. Since ChatGPT already has the content of my article, it can generate the SEO description for me.
This is one I don’t spend that much time worrying about. It doesn’t need to be perfect. I tend to use it and move on.
URL Slug: I could just publish my blog post without changing the URL slug. But I realize there is some value in what is used here. So I ask ChatGPT for a suggestion when I provide my initial draft.
Focus Key Phrase: I use the Yoast SEO tool, and it asks for a focus key phrase. I’m not an SEO expert, and I’m not relying on ChatGPT to necessarily be an expert either. But this is one of the things I ask for when I provide my initial draft.
Featured Image: I’m not a designer. While I work with a tech partner for major design projects, I’ve never hired someone to produce things like featured images for blog posts. One way around this in the past is that my design partner created a layered template that provided several options for background colors and icons. And I used that for years.
But that eventually becomes repetitive. I’ve worked with ChatGPT to add some variety to my featured images without creating something that is completely inconsistent with what I’ve done in the past.
Granted, this is one of the more frustrating uses of ChatGPT since it’s not always particularly responsive (it took way too long to create the image for this blog post). It will forget the aspect ratio I need or create images that just don’t cut it. It’s taken a lot of refining, but it’s generally been worthwhile.
Social Media Post: Usually after writing a blog post, I’m pretty emotionally exhausted (just look at how long this post is!). While I generally write my own social media posts, I’ll ask ChatGPT for suggestions that I may or may not use.
3. Podcast Workflow.
While The Pubcast has technically been around since 2012, I rededicated myself to it this year with a different format. I publish one 5-10 minute episode every Wednesday.
While I hire someone to edit and publish the episode, I still need to come up with topics, script the episode, and record.
There are two main ways I use AI to help…
Topic Brainstorming: Every week, I publish one blog post, one podcast episode, and three short-form video posts (with text). While there will inevitably be overlap, there also isn’t duplication of content. I want to leverage each unique format to get the most out of a topic.
It can be especially difficult coming up with unique podcast episodes that don’t overlap too much with my blog posts. I use ChatGPT to help me break through that block.
ChatGPT already knows my writing style and the topics I write about since it helps me edit my blog posts. But what I’ll do is create a prompt that includes a reminder of the list of topics of all content I’ve published this year across all formats. And the goal is to come up with a unique angle for a podcast that is consistent with my style.
Here’s the beginning of that prompt…
After some back-and-forth for clarification, ChatGPT returned a list of five ideas. Of those five, I liked one…
As is almost always the case, I didn’t take this outline word-for-word, but it put me on the path to scripting and publishing this episode.
Episode Page Transcription: Once I script and record the episode, I also rely on AI to help with transcription. There’s a section of my website dedicated to my podcast. Each new episode consists of an embedded audio player, links to where you can subscribe, and a transcription.
While I could just paste the raw transcript, I wanted something that’s more readable in paragraph format. So I created a project in ChatGPT where I establish rules on what liberties it can take with slightly rewording the transcript to be readable.
It’s taken quite a bit of refinement and occasional correction, but it usually does a good job now.
4. Email Marketing.
List building and email marketing are huge parts of my business. In fact, I’d say that it’s a partnership between Meta ads and email that makes my business viable.
My email marketing strategy is rather complex, combining automated sequences and manually delivered broadcasts. There are a couple of areas where AI has been particularly helpful in this area…
The Loop: My primary lead magnet is a weekly email subscription called The Loop, which summarizes all of my recent content from the past week. There are actually a couple of ways I’ve used ChatGPT to help with this.
First, I realized earlier this year that this subscription needed to be my highest priority lead magnet. The problem was that I had positioned it very generically as “Weekly Recaps.” So the first thing I did was had a long back-and-forth with ChatGPT to come up with a new name and logo. What I went with was one of MANY possibilities that ChatGPT ran by me.
But I also use it to help with assembling that weekly broadcast. The structure is simple with a numbered list of titles, links and paragraph descriptions. I take all of those from the emails I already sent during the week for anyone who subscribed to getting those specific alerts. I give ChatGPT this list, and ask for a quick review for spelling and grammatical errors as well as suggestions for an email subject, preview message, and intro.
Once again, I’m not taking what ChatGPT gives me blindly, but it helps me move quickly.
Sales Copy Brainstorming: I was never trained in sales. I have a Philosophy degree at a liberal arts college, and I’d consider sales to be my least favorite part of the job. But it’s also required, of course.
Historically, that has meant a very soft sales approach. And that’s generally worked for me. But there are times when I know I can and should be better and do more to ask for the sale.
That came into play a couple of times this year when I had a long back-and-forth with ChatGPT about a new promotion. We discussed my goals, the pain points I wanted to address, and the specific product features I wanted to feature.
It gave me a great starting point for something that I otherwise don’t have a lot of confidence doing. And it also did well!
An important point in it “doing well,” though, is that I am always involved in this process. I’ve said it a few times, but I do not automatically accept what ChatGPT gives me. Sometimes it’s a starting point. Sometimes we start over completely. Sometimes I throw it in the trash and create something entirely on my own, the old fashioned way.
The main thing is that using AI helps keep the process moving forward.
Meta Advertising Assistance
I realize that some Meta advertisers are embracing AI entirely, using it to automate and iterate at scale. I’m not there yet. I feel strongly that human involvement is still needed for now.
But here are examples of how AI is involved in my Meta advertising now…
1. Campaign Planning.
Thanks to Andromeda, we’re in a new era of campaign planning. For years, advertisers focused their time and energy on campaign construction, targeting strategies, and other levers while the ads themselves were (likely incorrectly) seen as a secondary factor. That’s changed now with Meta requesting creative diversification to get the most out of your advertising.
That means moving beyond the unofficial six-ad limit and creating ads catered to different customer personas, pain points, product solutions, and more. It means developing ads using different creative styles and formats, leveraging static images, videos, and carousels while optimizing for different placements and aspect ratios.
All of this can be overwhelming, especially if campaign planning doesn’t come naturally to you. I’ve used ChatGPT to work through this initial planning to help isolate the specific people and pain points my ads should be speaking to and the various angles I should take.
2. Brainstorming Primary Text.
Once I come up with a plan, I work with ChatGPT to execute. I’ll often generate the initial primary text and headline for each ad, and then I’ll have ChatGPT generate text variation suggestions that are consistent with my style and voice.
3. Images or Creative Inspiration.
As discussed earlier, I’m not a designer and the creative side can be a significant bottleneck, especially once we start looking at creating ads in bulk to satisfy Andromeda. But AI can be helpful.
I’ll go back and forth with ChatGPT on a creative concept, and it will generate an initial mockup. I will either work with it to get it refined or take it as inspiration to develop something new in Photoshop.
4. Background Generation.
Meta has native generative tools that can use your original image as inspiration for completely new images or images with AI-generated backgrounds.
Personally, I’ve yet to use any of Meta’s completely AI-generated images, since they’re the dictionary definition of slop…
But I absolutely do select acceptable AI-generated backgrounds…
These provide an easy way to create additional variations.
5. Advantage+ Creative.
There are two elements of Advantage+ Creative, depending on the level of adventure you desire.
Account Level: If you want to throw caution to the wind, you can opt into testing new AI generation features that will impact fewer than 5% of your ad impressions. This is found within your Advertising Settings, and there are currently 11 such features that may be tested.
I leave this on.
Ad Level: And of course, there are the various Advantage+ Creative features that you can turn on or off at the ad level.
While I don’t leave all of these on by default, I need a very good reason to turn one off. Most are relatively harmless, if not helpful.
Efficiency and Scale while Limiting Risk
As I said at the top, I’m not an AI expert. But I also don’t bury my head in the sand and hope it goes away. I want to use it in ways that make me better.
Overall, I’d say that the best way to define my use of AI with content and advertising is using it to increase my efficiency and scale while limiting risk. I don’t want to accept AI-generated images and text blindly. I remain an important driver of the decisions made and directions taken.
We’re also at a point right now where skipping the human element could be dangerous, if not careless. But these tools will continue to improve, as will our trust in them.
Your Turn
How do you currently leverage AI for content and advertising?
Let me know in the comments below!
