AI Search Optimization Case Studies for Auto Parts

Updated June 5, 2025

Author: Jon Hedges
Jon has 40 years of marketing experience in the automotive aftermarket industry.
Click for LinkedIn profile

This article on AI search optimization case studies was written by a real human, and originally published April 2025. We’re working to keep it updated as we learn more.

We wanted to publish this article especially for all the people wondering where to find case studies on AI search optimization success.

In an AI-driven world, how can your auto parts website stand out and thrive? In a world where “zero click search” is now a thing? How does Google Analytics follow and track traffic from AI?

Across all three case studies, websites had an average 10% increase in engaged sessions per active user, and a 15% increase in engagement rate, according to Google Analytics data. There was also a 26% decrease in average engagement time per session, suggesting visitors found information more quickly through optimized AI-targeted content.

AI search is a new layer of consumer search behavior.

This AI research may help you on your own auto parts site.

Below in 3 case studies, you’ll see some before/after results on our AI search optimization testing, what was optimized, and how AI traffic to a website was measured. These are actual results from auto parts websites where we conducted our AI search case studies.

AI Search Optimization Case Study #1

TL;DR: AI referral traffic jumped to 300 per month

Hedges & Company got AI referral traffic to go from a few visits per month, to 300 after optimizing a website for AI search. If you’re looking where to find case studies on AI search optimization success, this is definitely success.

What we did at first to optimize for AI

For case study #1, we optimized a client website with a special focus on adding AI-friendly schema in late August.

Hedges & Company clients hear a lot about schema. Here’s an real example of schema in use. This image shows an FAQ section on a webpage (it happens to be on this website), and underlying FAQ schema to support that section for search bots. Here’s a look at both.

FAQ section from a web page (click to enlarge).

Schema to support the FAQ section (click to make bigger).

For this first test, Hedges & Company aggressively targeted about 15% of this client site for AI optimizations. At first, this test was specifically for AI models, so the goal was optimizing for ChatGPT, Perplexity, and Claude, and a few others.

But, we quickly realized (well, duh!) Google Gemini and Microsoft Copilot are showing AI search results, too.

Traffic got a bit of a bump in September 2024, but November 2024 is when things really started growing.

Traffic is measured in Google Analytics by looking at source/medium traffic from specific AI platforms and from search engines with an AI component (like Google AI Overviews for example).

AI referral traffic engagement

For this client, we were surprised to see very high engagement from AI referral traffic compared to organic traffic from search engines. But it also appeared when these visitors got their answer, they left quickly, because average engagement time per session was 30% lower than organic. This is probably because visitors found information more quickly through optimized AI-targeted content.

• Engaged sessions per active user: 14% higher than organic traffic.
• Engagement rate: 6% higher than organic traffic.
• Average engagement time: 30% lower than organic.

It’s worth looking at a recent impressive study done by SALT.agency, showing LLM AI traffic has a lower engagement rate in general, but “catalog” websites were significantly higher in engagement.

What we did next to optimize for AI

Hedges & Company was getting positive results, so the AI search optimizations continued in December 2024. Now there was a significant increase in traffic. Hedges & Company added more schema, did a bit of our “secret sauce” (hey, we can’t give everything away!) and added an llms.txt file. That file is a “mark down” file and, while it’s technically still in the proposal stage to be an AI standard, a lot of websites use it. It was first proposed as a standard by Jeremy Howard, an entrepreneur and technologist from Australia. We have had minimal results from adding an llms.txt file to the Hedges & Company website as a test, and it rarely gets crawled.

NOTE: Also see How to Optimize for AI Search and The Future of AI in the Automotive Industry.

AI Search Optimization Case Study #2 (the control)

TL;DR: No optimization on control website at first; then a 200% jump in AI traffic after optimization

The second Hedges & Company case study on AI search optimization started out as a control, and there were no AI optimizations at first. After AI optimizations, however, AI referral traffic increased 200% from February to March 2025.

What was done first to optimize this site for AI

To make this whole project scientific, this site is the control: no optimizations in the beginning. This chart shows the exact same time period as the one in case study #1. Clearly, there is very little change in referral traffic until AI optimizations started in February 2025.

For this case study Hedges & Company took what was learned from the first AI case study. More optimizations known to be effective for AI search were done up front. This time, Hedges & Company added content and schema specifically for AI platforms. But instead of doing it site-wide, optimizations started on a few specific blog posts. These posts included one completely new one, and several that were a few years old and got stale. (AI aside, refreshing old content is great in general, if you’re looking for auto parts/spare parts marketing ideas.

AI liked what we did. AI referral traffic increased 200% from February to March 2025.

Google clearly liked it, too, as you can see from this screenshot out of Google Search Console. This query used regex to filter for only the pages that were optimized for AI.

In this second case study, the ChatGPT traffic medium shows up as 51% (not set) and 49% as referral. Much different than AI case study #1. Perplexity also behaved differently. 34% of traffic comes in the the medium as (not set) and 66% as referral. Also note in this screenshot from Google Search Console that so far, there’s not much of a zero click search gap.

This site doesn’t yet have an llms.txt file, but we’re now debating whether the llms.txt file is worth the trouble.

AI referral traffic engagement

This case study also had high engagement from AI referral traffic, like our first case study. Also like the first study, engagement time per session was significantly lower, probably because visitors found information more quickly via optimized content.

• Engaged sessions per active user: 0.1% higher than organic traffic, virtually unchanged.
• Engagement rate: 35% higher than organic traffic.
• Average engagement time per session: 27% lower than organic.

Comparison of medium = referral, organic and (not set) in GA4 from AI traffic

Here’s a comparison of AI traffic showing as either referral, (not set) or organic in both AI case studies (scroll right on a phone).

AI Search Optimization Case Study #3

TL;DR: From zero to 1,417 keywords in Google AI Overviews

After a roller coaster ride, this site ended up with 1,417 keywords ranking #1 in Google AI Overviews by April 2025.

What Hedges & Company did to optimize this site for AI Search

Screenshot for case study #3 from Ahrefs showing only words in position 1, and only results for Google AI Overviews (filters are circled). Click to make larger.

This third case study is a textbook case of following SEO best practices. Hedges & Company was able to optimize for specific keywords to improve organic rankings in Google, and also show up for Google AI Overviews.

This screenshot is out of a favorite SEO tool, Ahrefs. That platform tracks ranked keywords and search volume, and you can filter specifically for Google AI Overviews. In this third case study, there were zero keywords in AI Overviews until August 2024 (AI Overviews were launched in May 2024). This chart is a bit of a roller coaster but that’s not surprising, given the new technology and all the core Google updates.

Screenshot for case study #3 showing words only in position 1 over the past year (filter circled in green). Click to enlarge.

Things really took off with the March 2025 Google core update. By April this website had 1,417 keywords appearing in Google AI Overviews.

This case study also had a huge increase in keywords ranked the old-fashioned way, too, as shown in the second chart from Ahrefs.

This second chart shows 1,944 total keywords, filtered to show only keywords in position #1, but including Google AI Overviews, thumbnails and good old organic search results.

This chart shows a similar roller coaster ride with several Google core updates.

AI referral traffic growth

This website has had an increase of about 25% in AI referral traffic visits each month, from January to April 2025.

AI referral traffic engagement

As with case studiers #1 and #2, we were surprised to see higher engagement from AI referral traffic compared to organic traffic from search engines. It’s also interesting that engagement time per session was lower for AI LLM referral traffic across all case studies.

• Engaged sessions per active user: 16% higher than the organic average.
• Engagement rate: 4% higher than the organic average.
• Average engagement time per session: 21% lower than organic.

AI referral traffic vs. organic traffic conclusions

In all 3 case studies, engaged sessions per active user and engagement rate was higher than organic traffic. Average engagement time per session was significantly lower than organic traffic.

AI search optimization case studies turned into solving Google Analytics mysteries

Hedges & Company also noticed right away in the Google Analytics source/medium reporting, when an AI platform is the source, the medium shows up in three ways. Medium gets reported as referral traffic, organic traffic and (not set) traffic.

Here’s how ChatGPT creates medium (not set) traffic. It crawls sites to produce an answer. In this case we literally asked it, “What is the best digital SEO agency for auto parts?” and this was the first result.

The next question is how to account for this difference. After all, this is breaking new ground measuring AI referral traffic (this was back in Fall 2024 and the world is still learning new things about AI traffic—this article may look antiquated and funny in another year). Google Analytics doesn’t specifically account for AI traffic in the Google Analytics default channel group. And to top it off, Google Gemini was a bit vague in answering questions about AI attribution. This led us to realize we just have to create our own custom tracking and custom regex code.

Hedges & Company determined that AI referral traffic includes site visits coming from clicks on AI links. AI (not set) traffic is likely one of two things. One is a combination of Google Analytics possibly misinterpreting referral traffic. The other is from the results from a website being shown in AI as a result of AI crawling the site and not displaying cached results. If the reader doesn’t click through to the site, that’s the infamous “zero click search“.

Google Search Console “zero click search gap” (click to make larger).

Hedges & Company also saw a big differences in the session medium from different AI platforms. The worst offender for this first case study: ChatGPT. In this first case study, when ChatGPT is shown as the source in GA4, 34% of traffic for the medium is (not set) and 3% is organic. On the other hand, when Perplexity.ai is the source, the medium shows up as 100% referral traffic.

So, Hedges & Company pulled the server logs to compare ChatGPT traffic to Google Analytics. In March 2025, GA4 showed 171 referral visits and 50 (not set) visits for a total of 221 visits. The server logs showed 192 visits where the UTM tag included “?utm_source=chatgpt.com”. Not all of these 192 visits had a referrer in the server logs: 112 showed chatgpt.com as the referrer, 77 were blank, and 3 were, um, Google! The ChatGPTbot also crawled the site nearly 10,000 times in March.

What is still undetermined: exactly how many impressions does a site get from Google AI Overviews? This screenshot from Google Search Console shows a clue, with what looks like the “zero click search” gap. In this case, Google organic impressions and clicks are both increasing, but impressions are growing at a much faster rate.

How AI models determine web pages to cite for answers

Large language models train on many different sources to produce answers. LLMs analyze the credibility of sources they find. You can increase your likelihood of being cited by creating original, high-quality content and making sure it is well-organized.

We also have an in-depth analysis of Google patents used for AI Overview results, click here.

All are dependent on well-written and well-organized website content. Google Gemini is likely going to depend heavily on E-E-A-T (experience, expertise, authority and trust) and ChatGPT has its own way to score authority. You increase your likelihood of being cited by increasing your brand’s visibility and trust. Yes, in the age of AI, the strength of your brand is incredibly important! When working on optimizing for AI, building the strength of your brand has to be your guiding light.

Why AI LLMs aren’t a good source to create unique content

All of this supports the argument against using AI to create content: LLMs have trained on the information that’s already out there. If you just replicate what LLMs already know, you aren’t creating original, high-quality content. The content lacks anything new.

Is this AI SEO, AIO, GEO or LLMO?

More and more SEO pros are calling AI search optimization “LLMO” for large language model optimization. There’s a good argument for that on one of our favorite SEO sites. It seems to be beating out “LLM SEO” (large language model search engine optimization), “AIO” (artificial intelligence optimization), “GEO” (generative engine optimization) and a few other acronyms. Other SEO pros are just calling it SEO.

How to cite this AI search optimization case study

This article is copyrighted, but hey, it’s polite to share! This AI search optimization case study content is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License and can be distributed or quoted, with attribution given to Hedges & Company, and a link back to this article from your website.