Time to read: 8 minutes

We interviewed Alex Nassar, a student at Saint Ignatius High School who reached out to us for help with his AP Research project. Alex interviewed 42 HVAC companies from Northeast Ohio and found that they used AI on their websites and felt like it helped their business. However, this isn’t an SEO performance study, and these results aren’t based on analytics. They reflect the business owners’ perceptions of how AI affected their websites’ goals.

If we don’t read past the headline and we stop there, the takeaway is basically: HVAC companies used AI, and it worked. Watch our interview with Alex above, or continue reading to learn more about what he found and our thoughts.

Study Highlights:

  • 62% of the 42 companies used AI somewhere on their site
  • 96% of that 62% said AI positively affected their goals

But it’s really important to go past the headline and get into the details to make sure we don’t come away with the wrong conclusion, which can really hurt us strategically down the line.

💡 Key Takeaway: This is where things get really tricky and where we really need to lean into proper attribution.

Perception Is Not Attribution

The perception from the side of the business owners was that it felt like it helped. Customers mentioned the site looks more professional, and so they were making the connection between using AI extensively and the good results. The bottom line: if you really want to know what’s working, we have to look at hard numbers.

✅ What actually tells you AI is working:

  • Organic traffic
  • Conversion rate
  • Search rankings
  • Revenue

And we have to line that up with the actions we’ve taken and the changes we’ve made to our website. In this case, most of the positive attribution was based on self-reporting. It was not based on these metrics. So already it’s a little bit flaky.

Are We Sure It Was AI?

The other thing we have to ask is whether AI actually caused these positive results. If they weren’t monitoring carefully, there’s no way to know.

❌ What the self-reporting doesn’t rule out:

  • Fresh Google reviews that came in
  • Seasonality
  • Word of mouth
  • A competitor slipping
  • Other sources entirely, even social media

These bumps can come from a lot of different directions, and that’s why full awareness is so important before we draw any conclusions.

Group B Refined With AI – They Didn’t Build With It

Here’s the thing. Let’s just assume the AI actions taken on these websites were the thing that improved business. It’s not like they completely designed and wrote their sites with AI alone.

Group B – the largest group of participants – made up 52.4% of the study.

They refined with AI. They didn’t build with it.

Bar chart of the three study groups. Group A, built and refined with AI, 9.5%. Group B, refined only and did not build, 52.4%. Group C, used no AI at all, 38.1%. How the 42 companies actually split.

That’s really important to know. They used it for things like explaining things more clearly and simply to customers or adding FAQs to the bottom of an already written page. This was existing content, and the AI was used to enhance it – which is a good use (and one that can be taken far further with a qualified AI search agency).

Mount AI: What Overreliance Looks Like

We’ve seen a large number of businesses design and write with AI, which we would consider an overreliance on it, and it simply doesn’t work. In fact, the crashes are so common it’s got a name: Mount AI.

⚠️ The Mount AI pattern: What you’ll often see happen with a site that is over-relying on this tool is an initial gain – followed by a wall and then a fall.

It’s also worth noting where a study like this one lands on that curve.

Line chart labeled rankings and traffic over time, rising to a peak then falling sharply. A marker partway up the climb reads "the study’s interviews land here, results still look great." A second marker at the peak reads "the wall, then the fall, months after anyone asked The Mount AI shape – and where the study’s interviews land on it.

Two Case Studies From the Wild

Momcozy: They were growing fairly organically until they decided to aggressively scale up with AI. They inevitably hit a wall, and you can see the tremendous fall that happened after that – which unfortunately did undo a lot of the natural growth they had beforehand.

A prospect who came to us: They decided to give these AI tools a spin and write their content with them. The same thing happened. They hit a wall, traffic fell, and they haven’t been able to recover since, despite a lot of different efforts to do so.

Two organic-positions charts from Ahrefs. Left, Momcozy: years of organic growth, a steep AI-driven climb, then a sharp fall. Right, another site: strong for years, then a drop off a cliff it never climbed back from. Two real sites that climbed Mount AI.

And we know this kind of thing is pretty widespread. If you go to Google Trends and search something like “SEO recovery,” you’ll see this:

+550%

Increase in “SEO recovery” searches over the last five years – most of it from 2025 and eight months into 2026.
Source: Google Trends

: Google Trends line chart of U.S. search interest in "seo recovery" from 2022 to 2026, flat until 2025 then climbing steeply to its peak. A marker at the inflection point reads "the shortcut’s bill comes due – everyone’s Googling the way back. +550% growth in U.S. searches for “seo recovery” over the past five years. Source: Google Trends.

A lot of the businesses and individuals doing these kinds of searches scaled with AI too aggressively, and now they’re looking for help.

The Real Cost Is Time

I want to be clear: I completely understand why business owners would try to do this. It would be great if you could summon a tool and have it take care of your website for you. But it just doesn’t work.

The real problem is that the time you spend building with AI, publishing, hitting the wall, and then trying to figure out what’s going on and recover is time you could have been starting the clock on real organic compounding results.

For a lot of the clients who come to us, we have to spend the initial months of the engagement digging them out of the hole that AI got them into. Then, once we’re back to baseline, we can start growing on top of that.

It’s my advice to not be tempted by the convenience of AI tools – because in the pursuit of saving time, you can actually lose a lot of time.

AI Writes at the Average

The other thing about AI is that it writes at the average. There’s nothing unique or proprietary about it, so it’s really easy for competitors to replicate – they’ve got access to the same tools you do.

And in terms of visibility in search results and AI outputs, if those systems are looking for an answer and you represent the average, they’re much more likely to default to a source with higher authority than you.

💡 Key Takeaway: To give yourself the best chance of visibility, unique expert-led content is the way to go.

Our Ratio: 75% Human Expert, 25% AI

75% Human Expert Content | 25% AI

The ratio we run at Sixth City Marketing.

Our practices are much more in line with those of Group B, where we’re doing enhancements to the content rather than relying on AI tools to purely write it.

But we take it much further than just clarity and FAQs. There’s a lot more you can do in this 25% to improve your content and be way more efficient – but you have to partner with an agency that has the experience and know-how to do it right and take your content well beyond what it would be if it were just plainly written.

What the Right Use of AI Looks Like

This is one of our own clients, a wildlife removal company, who came to us with an AI-written website that was stagnating and not bringing in enough business. We applied our 75:25 ratio and our AI best practices to enhance it and make it more efficient. The results speak for themselves.

Line chart of the client’s organic positions from November 2024 to August 2026. It opens flat, marked "stagnant, AI-written – months of flat," then climbs steadily and holds near the top, marked "rebuilt, and it holds – no cliff, it compounds Stagnation, then durable progress. The opposite of Mount AI. Source: Ahrefs, organic positions (1–10).

Metric Result
Leads (June 2026) 343
ChatGPT referrals +217% year over year
Site clicks Up
Google Business Profile clicks Up
Top revenue keyword ranking Moved from #3 to #1
✅ The Outcome: The client has added hours and a second site to our engagement.

Two Ways to Use AI Tools

Simply put, there are basically two ways to use AI tools.

The High-Risk Way The Right Way
What AI does Writes and designs everything Assists your experts, designers, marketing agency, and you as the SME
The mix AI-led 75% human expert / 25% AI
What tends to happen Initial gain, then a wall – traffic and leads fall. Real organic compounding results

Line chart comparing two paths over twelve months. The high-risk AI build spikes to a peak around month six, marked "false spike – the survey’s whole snapshot," then falls away. SEO done the right way climbs steadily and passes it, marked "still compounding – a year of momentum. Same twelve months, two very different places to stand.

What Google Says About It

If you go to the Google Search Central blog, you’ll see that Google has taken a hard stance on using generative AI to create content.

It’s looking for little to no effort, little to no originality, and therefore little to no value.

So if your site represents that rather than you and your business, it’s not going to help you grow.

Have a Question About the Study?

If you’d like to get in touch with Alex to ask him about his findings, you can reach him at alexnassar2009@gmail.com.

About the Author

Brian Gorman

Brian is the SEO Director at Sixth City Marketing and manages our team of SEO strategists. He has 10+ years of doing SEO for mid-size B2B companies, small local businesses, and even large ecommerce brands. Alongside his experience as an SEO conference speaker, you can find his insights featured on Moz, Hootsuite, and many other notable publications. Follow Brian on LinkedIn