I'm a CMO who uses a '10-80-10' rule with AI. It's transformed everything from photo shoots to customer feedback.

• Alan Magee, CMO at Orangetheory franchisee Empire Portfolio Group, is going deep on AI.
• AI-generated customer review responses are saving his team around 20 hours a month.
• Empire has also cut back on photo shoots and is using a marketing GPT to spot data trends.

This as-told-to essay is based on a conversation with Alan Magee, chief marketing officer of Empire Portfolio Group, one of the largest franchisees of Orangetheory Fitness, with more than 140 fitness studios across the US. It's been edited for length and clarity.

A lot of marketers don't even know where to start with AI.

Being a smaller, more nimble organization, I think you'll find more people in my type of role are probably using it way more than someone sitting at a big corporate brand.

We have used AI to create visuals and copy. We used MidJourney for our medical clinic brand to create eight-foot window clings that look like realistic people in Central Park.

It takes a lot of prompting, but once you get it there, you can find some real gems that work for it.

A lot of the in-the-box platforms are getting much better now for image generation for things that are low-risk and easy to execute.

I still would not use it to create a TV spot today, but if it's a video inside a studio that's running on the screen, we're using Canva to create that video, and Canva creates it for us with stock imagery that moves and shifts. That works really well and saves us time and money.

Using AI to spot trends

We work with Passionfruit, a staffing fractional agency. They help support marketing teams that are lean. I was part of the pilot group for Passionfruit's product called PIP, which is essentially a personalized marketing GPT.

We've got a wall of data of all of my marketing metrics. Each morning, I can go to this GPT and I'm looking at how many leads I got yesterday across 60 different locations and performing trend analysis. Previously, I was manually punching in Excel data coming from two or three or four different sources and aggregating it together.

I poured it into our walled garden with the Passionfruit PIP product, and then I can ask questions. It does the reporting for me. I cross-check it to make sure it's accurate, and it gives me insights. It saves me a ton of time, and it helps me identify insights and trend analysis that I might have missed by doing it manually. That's the big win. We'll say: Hey, are there yellow or red flags of performance from a location?

I can trust the data a little bit more than putting stuff into ChatGPT because I know how the data is mapped in PIP. I don't know if ChatGPT is going to tell me the truth.

The 10-80-10 rule

I played around trying to dump a ton of unstructured data into ChatGPT, and to get it to give me the same results. Sometimes it does. Other times, it spits out location data that was never put in there.

There's a 10-80-10 rule that's floating around in the marketing world: 10% human input, 80% AI execution, 10% human review.

It helps get to some level of validation in trusting the insights.

I'll ask:

• What sources are you using for this?
• Was it just from Yelp reviews, Reddit, or multiple sources?

Responding to customer reviews

Answer-engine optimization is the number one thing I'm most interested in.

One of the big things that we can control is customer reviews. It used to be that Google reviews helped drive you into the top three of the search bar. We talked to our team about making sure a reviewer mentions a coach's name, and that they're talking about their experience of the studio, or uploading photos.

In generative search, those are small knobs you can turn to help.

My teams used to spend upwards of 20 hours a month responding to Google reviews. About five months ago we switched a new platform called AI Review Response that creates personalized responses in real time, written by rules. They change for everyone, so it's not just the same automated response. It's delivering the same level of output, and we have not seen a drop off.

Showing up authentically on Reddit and YouTube

Orangetheory is very well-positioned from a Reddit perspective. I really didn't give Reddit a lot of focus at any past brands I worked at, but it's definitely come more to the forefront. Reddit and YouTube are two core pieces of contextualized information that feel very valid from people who are posting.

Marketers are going say, "How do I hack that?" I think the users on Reddit read right through that. How you show up as a brand authentically on Reddit is one of the biggest challenges

Will AI replace the CMO or entry-level marketers?

On the question of whether AI is going to eventually take me out of a job at some point, I still believe that marketing is a mix of art and science. AI can help with the science side, but there has to be someone in a seat who is thinking about the art side of it. Thinking about what is happening with the consumer, understanding all the parts and pieces of what's going on with their company or the product, and putting those things together.

AI is definitely impacting entry-level marketers. There is more of an expectation that they should have knowledge of how to use AI.

What I'm looking for is someone who's curious and alert. It's no longer like, "Hey, it's your first day, and here's your manual on how to do every single thing." Those manuals don't really exist anymore. Marketers want people who are going to be able to figure stuff out on their own.

It used to be: this is the one path, and these are the five steps to get you to that outcome. Now, there's maybe five or six different ways to get that outcome, but the expectation is that we need to get there fast.