Time to read: 8 minutes
More and more businesses are doing SEO audits with ChatGPT these days. And today, I’m going to tell you why that is not as good as a human audit.
I don’t mean to take all your AI toys away. I just did another video saying that when you write content, you shouldn’t just use AI to do it. It should be 75% human written, 25% AI. And today, I’m going to take your SEO auditing away.
But in all seriousness, this is actually a really difficult thing for clients or prospects to see and understand. So, I’m going to walk you through why doing an SEO audit with AI is severely lacking compared to a human audit.
If you follow an AI audit, you’re going to diverge from a truly impactful strategy.
At Face Value, a ChatGPT SEO Audit Looks Decent
So, I jumped into ChatGPT, and I gave it a website. This is the kind of output you’re going to get. And you can see why it’s so difficult to tell that this is lacking just by looking at it, because a lot of this sounds right. And in some areas, it might even be right.
Here are a few of the “biggest SEO opportunities” it gave me:
- Build out the service and location architecture. How can you argue with that? It’s a great recommendation, and they even give you some details to it.
- Turn the Our Work page into an SEO asset. We know that these project-based pages can be really helpful with conversions. So, that’s a good recommendation as well.
- Create stronger individual service pages. Sounds great.
Where the AI audit and our strategy agree: location pages, project pages, and internal linking
So, you can see that if I was just a business owner and I’m not using these AI tools every day, if I’m not doing SEO every day, this might look just fine. So, why would I sign on with an agency? This is giving me everything I need.
Close Enough to Sound Right, Wrong Enough to Cost You
Well, here’s the deal. Once we move past kind of the surface-level recommendations and start going deeper, that’s when the cracks start to appear, and we actually start deviating from a strategy that’s actually going to make us money.
Blog Content Without Localization
One of the recommendations this audit made was for blog content, and it gave me this topic: “How much does a roof replacement cost?” Okay, and that’s where it stopped. So, that’s very general. There’s no localization to that.
If I go ahead and search that, I get an AI Overview answer. And you can see right here it says between, let’s say, $6,000 and $20,000.
But the thing is, my business was local to Indianapolis, Indiana. When I search that with the location in the search, you can see that we get a different result. There are a lot of differences.
| Search | What happens |
|---|---|
| How much does a roof replacement cost? | General AI Overview answer. We’re not likely to get cited as an Indianapolis business. |
| How much does a roof replacement cost in Indianapolis? | A different result. We’re much more likely to get cited, and there’s even an Indy-based business getting cited there. |
Localize it, or AI eats it.
So when we’re writing blog content, we need to be talking about the area that we are experts in and talking about the scenarios that are specific to that location.
Flagging CallRail Tracking as a Problem
You can also see in the output of this ChatGPT audit that it called out the fact that we’re using a lot of different phone numbers on the homepage. Now, that can be a problem, but the issue is we did that for a reason.
We’re using a service called CallRail, and that uses different phone numbers for tracking, so you can tell where your leads are coming from. That’s actually a great thing. Attribution is only getting harder, and this is one of the ways that we can tell where leads are coming from.
The “too many numbers” were CallRail tracking numbers, on purpose.
Three of the Biggest Levers Weren’t on the List
Past the recommendations that drift, there’s a bigger problem: the things the AI audit never mentioned at all.
Link Building
ChatGPT was totally blind to link building. Now in this scenario, the client is in a very competitive space, which is roofing. And as far as we can tell from authority scoring like Ahrefs’ Domain Rating, their authority is much lower than their competitors’. So link building is a big part of this SEO strategy.
Source: Ahrefs Domain Rating, 2026. The client’s 25 is inflated by spam links.
Map Pins in Priority Areas
One of the most aggressive and impactful things that we could be doing for this business is securing map pins in other priority areas where their visibility isn’t as good, and creating Google Business Profiles around those addresses. This was another item that ChatGPT didn’t even mention.
Source: Local Falcon, “roof replacement,” 2026. Share of top-3 map results by distance from the pin.
So while you are updating and optimizing your Our Work or portfolio pages, you wouldn’t be doing something like this, which is going to give you a much higher ROI.
AI-Written Content
And finally, ironically, ChatGPT did not notice that almost the entire website was written with AI, which is a huge reason this website is not performing at all.
They’re doing the thing that I recommended against in another video. They’re just giving ChatGPT a topic, it’s writing the entire article, and they go ahead and publish that on their website.
That’s not setting you up to perform in search. It’s not setting you up to convert.
You’re not getting your unique expertise and know-how onto the page, and therefore your site is not differentiated. It represents the average of what’s already out there.
Source: Google Search Console, 90 days. All 10 blog posts at kickoff read as AI-written; only 2 showed any search impressions.
Why You Want a Human SEO Audit
So, the big takeaway is that AI auditing for SEO can look like it’s on the right track, but actually it takes you to the wrong destination. And the more you look into these audits and the more you follow them, the more off track you’re going to get.
The two audits start together, then split apart.
It called out location pages, project pages, and it even mentioned internal links. All of that stuff is great, but you have to know the full context and the full picture in order to be able to pick the right initiatives to spend your time on early on, and prioritize appropriately to impact the bottom line as quickly as possible and as much as possible.
If you follow an AI audit, you’re going to diverge from a truly impactful strategy. And that’s why you want a human being doing it for you. They’re going to notice things like this:
| Area | ChatGPT’s audit | A human strategist |
|---|---|---|
| Blog content | Generic cost post | We need to be localizing our blog posts |
| Phone numbers | “Too many numbers” | No need to worry, it’s CallRail tracking |
| Link building | Not mentioned | We definitely need a link building campaign |
| Map pins | Not mentioned | We need to be securing map pins, because this is a local client |
| AI-written copy | Not mentioned | A rewrite, consolidation, and retiring ineffective pages |
Four Questions to Ask Any AI Recommendation
The differences are hard to see, and we have to catch them anyway. Run each recommendation through these before it goes into a strategy.
- What didn’t it mention?
- Would it need client context?
- Is it local enough?
- Is the “problem” on purpose?
What didn’t it mention? Would it need client context? Is it local enough? Is the “problem” on purpose?
Partner With an Agency That Knows the Right Way to Use AI
I know AI makes things easier, but it’s really important to partner with an agency that knows the right way to use AI. It’s not a one-size-fits-all tool. You have to have the right internal philosophy for using it to get the best results out of it.
