The US Automotive Aftermarket Through AI's Lens: A Research Experiment

Prefer a pdf? Click the button below and complete a brief form.

By ChatGPT 4o: A question mark comprised of various automotive parts

Intro

Having spent the better part of a decade immersed in the automotive aftermarket, I've gained a real appreciation for just how intricate and surprisingly resilient this industry is. Think about it – complex distribution, customers ranging from DIYers to massive fleets, constant tech shifts – yet it keeps adapting and growing despite major shake-ups.

So, when I started messing around with AI for strategic research, where better to point it than an industry I know inside and out? 

My hypothesis felt straightforward, but also a bit ambitious: Can today's AI research tools really provide accurate, deep analysis of a complex industry like this without needing specialized training? And maybe more interestingly, how would different AI systems stack up against each other on the same task?

In this case study, I'll walk you through what I found – both about the state of the aftermarket itself and the real-world capabilities (and limitations) of using AI for this kind of research. You'll see comparisons between different tools, learn how the tools responded differently to the same standardized prompt, and hopefully get a clearer picture of how executives in this space can leverage these approaches for smarter decision-making.

Whether you're a seasoned aftermarket pro wondering what AI brings to the table, or a leader looking for faster, better ways to get industry intel, I hope this experiment offers some practical takeaways and maybe sparks some new ideas about how we approach industry analysis as technology keeps moving…and accelerating.

Research Experiment Overview

Hypothesis

I entered this experiment with skepticism: I expected the automotive aftermarket's complexity would trip up even the most advanced AI research tools, resulting in incomplete or uninformed takeaways. The industry's intricate channel structure and specialized terminology seemed like the perfect challenge to expose the limitations of generative AI research capabilities. Afterall, there’s a reason that Northwood University offers an Automotive Aftermarket Management program! 

I was particularly interested in seeing how these tools would handle industry-specific concepts like:

• Multi-tiered distribution networks

• Industry terminology 

• The relationship between VIO, VMT, and repair rates

• Channel-specific competitive dynamics

I expected Perplexity to perform best based on my previous experiences with it on other topics, while I was less familiar with the deep research capabilities from ChatGPT and Gemini. Manus AI was a last minute entry. The AI world is moving so fast new tools and features are being released almost every week. 

Methodology

I designed a straightforward approach:

1. Create a single, identical prompt for 4 leading AI research tools

2. Submit the prompt to ChatGPT (OpenAI), Gemini (Google), Perplexity, and Manus AI

3. Analyze each response for accuracy, depth, organization, and understanding of industry nuance

4. Fact-check key assertions where possible

5. Conduct cross-analysis using a different AI system to compare the outputs

The prompt asked for distribution channel analysis, competitive landscape assessment, industry drivers, trends, AI adoption patterns, success stories, and pain points. While not exhaustive, it covered the key dimensions any industry executive would want to understand. My thinking was let’s see how well it can do without a detailed prompt, but enough to point it in the right direction. Here is the exact prompt:

“Conduct a deep analysis of the Automotive Aftermarket in the US. Compare the last 5 years to the next 5-10 years. What are the different channels within the industry? How does product flow from manufacturer to point of sale? Who are the players? Who's winning? Who's losing? What are the major drivers to the industry? What are the major trends? What are leading indicators? How are companies using AI, specifically generative AI, if at all? What are the success stories and failures? What are the largest pain points for the industry? Each channel and its major players?”

It's worth noting that organizations like MEMA and Auto Care Association serve as critical intelligence hubs for the industry, publishing research and analysis that shapes understanding of market dynamics. Their annual release of the Factbook was something I looked forward to every year as we set out to start planning for the next year. AI won’t replace what they do right now, but it could enrich and optimize their work and workflows. Imagine getting the Factbook or similar content more than once a year! 

Key Findings

The results completely contradicted my expectations:

1. ChatGPT emerged as the clear winner, delivering a surprisingly comprehensive and mostly accurate analysis that captured the industry's complexity in ways that would typically take an industry newcomer years to develop.

2. Gemini offered a decent high-level overview but missed entire segments (notably remanufacturing) and lacked depth in several areas resulting in no actionable insights that are not obvious to even industry outsiders

3. Perplexity, despite my high expectations, brought up the rear of the pack. It included several significant hallucinations that couldn't be verified even when using other AI tools to fact-check. Like Gemini its actionable insights were lackluster. 

4. All 4 systems could deliver information in minutes that would typically take hours to days of research to compile.

5. The quality gap between systems was much wider than expected, suggesting we're still in an early phase where choosing the right AI tool matters significantly.

In less than 10 minutes, ChatGPT accurately captured industry metrics and relationships that typically require extensive exposure to conferences, tradeshows, and supplier meetings to fully understand. This level of performance was genuinely surprising given the industry's complexity.

The next section provides a detailed comparison of how each tool performed, with specific examples that highlight the differences in their research capabilities.

Putting the AI Tools Head-to-Head: Who Won the Aftermarket Challenge?

What Were We Even Judging On?

To make this a fair fight, we needed some ground rules – criteria that actually matter when you’re trying to use this stuff for real business strategy:

1. Accuracy: Did it get the facts right? Simple enough, but crucial. Were market size numbers ($391B in '23, heading to $472B by '27, according to the data) in the ballpark?

2. Depth: Did it just skim the surface, or did it dig into the gritty details? Think distribution channels, the DIY vs. DIFM split, the surprising importance of remanufacturing – stuff that actually drives the industry.

3. Industry Understanding: Did the AI get the aftermarket? Did it understand concepts like Right to Repair, the impact of ADAS, or why everyone’s scrambling for technicians?

4. Organization: Was the output a jumbled mess or something you could actually read and use without needing a decoder ring?

5. Actionable Insights: Okay, great analysis, but so what? Did it offer any smart ideas, warnings, or opportunities a real business leader could run with?

6. (Bonus) Speed: While we didn't formally score it, let's be real – speed matters. All the tools were way faster than a human intern, but some felt slicker than others.

The Main Event: How Did They Stack Up?

We ran the same core research prompt through ChatGPT, Gemini, Perplexity, and Manus AI. Their responses are hyperlinked in each tool’s title. Each tool was evaluated on a scale of 1 to 5 with 5 being Excellent and 1 being Poor. Here’s the lowdown based on the scores (see the image/table) :

A bar chart comparing AI tools across various criteria

• ChatGPT (Score: 25/30): The Surprising Champ
  
  

  • The Good: Honestly, it blew the others away. High marks across the board, especially in depth and understanding the industry's nuances. It connected dots between trends (like e-commerce squeezing margins) and provided insights that felt genuinely useful. It was the only one to mention the well-known industry term, “sweet spot.” It wasn't perfect, but it felt like a solid starting point for real strategic thinking. Perhaps key to its performance, ChatGPT asked clarifying questions. 

ChatGPT follows up on the prompt. 

ChatGPT nails the Sweet Spot and links it to a specific channel.

The Catch: Like any LLM, you gotta check its work. ChatGPT provides sources for a lot of the facts, figures, and trends, but that doesn’t mean everything is good as is. For instance, it assumed Worldpac is still part of Advanced Auto. Its AI section presents reasonable sounding applications and stories. It did offer sources, but those sources didn’t provide any citations or references for their work. 

ChatGPT was the only tool to go into detail on Reman.

ChatGPT wasn’t perfect. Some suspect sources were relied upon for a key section.

• Manus AI (Score: 21/30): The Organized Assistant
  
  
  

• The Good: This one felt different – more like an agent following steps, which tracks with Caiwei Chen’s MIT review. It scored well on accuracy and organization, presenting information clearly. The actionable insights were pretty solid too. Plus, seeing its "work" is a nice touch for transparency. 

• The Catch: It didn't quite match ChatGPT's depth or industry grasp in this test. Manus was also the slowest coming in closer to 15 minutes. Under the hood is Claude Sonnet 3.5 (maybe 3.7 by now) and a host of other open source tools. Claude is one of the most expensive models out there, so it needs to justify the price. Unlike the other tools, it only lists the references at the end. They are not hyperlinked and you don't know what was used where. I do suspect that the result can be finetuned to ChatGPT levels or perhaps even beyond, but that would take a few interactions. 

Manus AI’s output which includes a file that puts everything together as well as separate files for each of the major sections. There’s also the option to turn the result into a website, similar to Claude’s artifacts.

• Gemini (Score: 18/30): The High-Level Skimmer
  
  
  

• The Good: It gave a decent fly-by of the market, hitting the big numbers. If you needed a quick, basic summary, it wasn't terrible. 

• The Catch: Depth? Not so much, even after sharing its research plan, which you can edit or let it fly. It completely missed major spaces like remanufacturing. Insights felt generic, like regurgitated headlines. A sharp industry outsider could reproduce the results without much effort. 

Gemini’s history is difficult to access beyond a certain point. I could find the original deep research prompt, but it wasn’t clickable to expand the details. Luckily, I saved it ahead of time! Gemini provided a product segmentation that would only be useful for industry outsiders and its market segmentation was very limited in scope as well.

• Perplexity (Score: 15/30): The Disappointment
  
  
  
  

• The Good: Um... it's supposed to be good at research? Don’t write Perplexity off due to this one performance. I suspect there are relatively easy tactics to mitigate, if not eliminate, hallucinations. 

• The Catch: Oof. Lowest score. Accuracy was a real problem – we're talking hallucinations at 2 levels. First, in the response itself and then again with its sources. For suspect sections I reviewed the source cited only to find it had nothing to do with what it wrote in its analysis. All this in spite of being the tool that cited the most sources! I’m still trying to wrap my head around where it’s going wrong. To make matters worse, depth and understanding were weak. 

Several egregious hallucinations. The Honeywell Forge is an actual thing, but the article cited doesn’t even mention Honeywell. The same goes for every other bullet point in this screenshot. They might be actual examples, but the sources cited do not mention them. It was at this point that credibility was completely lost.

I was really interested in this $50M mistake, but the source doesn’t mention it at all. The source itself is fine.

They cited their sources as you can see, MEMA having the prominent #1 position as it should.

So, What's the Play for Aftermarket Leaders?

This little experiment wasn't just for kicks. It tells us a few things if you're thinking about using AI for your own business intelligence:

1. Right Tool for the Job: Don't assume all AI is created equal. Need deep strategic insights? ChatGPT looks like the stronger bet right now. Need structured data pulled together? An agent like Manus might be interesting. Need a quick summary? Gemini might do, but know its limits. Don’t rule out Perplexity. I have experienced better results and AI is moving so fast that this article will be dated by the time you read it. Don’t disregard having multiple tools or using an aggregator like Poe.

2. Trust, But Verify (Seriously): None of these are magic wands. Treat the output like a draft from a very fast, sometimes overconfident intern. Double-check critical facts, especially if it sounds too good (or weird) to be true. Whether you’re familiar with your research subject or not, don’t let it lull you into a false sense of security. Responses are going to sound convincing and if you understand how LLM’s work, then you should have a good inkling why. If you don’t, then it’s worth investing a little time to learn. 

3. AI + Human = Best Results: This isn't about replacing your experienced team members. It's about augmenting them. Use AI to do the heavy lifting on research, then bring in human expertise to validate, add nuance, and make the final strategic calls.

4. Get Creative: this case study focuses on industry analysis, you can adapt it in countless ways. Pricing, competition, product? Customers, suppliers? Sentiment? Incorporate your data (securely of course) and I bet the results improve even further. 

5. Find a way to stay up to date: It doesn’t have to be you and remember -  AI can even help here. I cannot stress this enough, AI is moving fast. We may hit a plateau at some point in the future, but its current trajectory is exponential. To illustrate, I started this experiment at the very beginning of March 2025 only to have Manus AI come out on March 5th. After testing it myself, Manus absolutely warranted inclusion. Since I started writing my findings, Google has released Gemma 3 and Gemini 2.5 Pro and more. Claude finally added web search. DeepSeek released V3-0324 and Alibaba released Qwen2.5 Omni. Reve AI came out of stealth to hit the top of the image generation leaderboard… only to be outdone a day or two later by OpenAI’s native image generation and Ideogram 3.0. Google even upgraded its deep research. An update to this case study could be in store in the very near future. 

What does GenAI have to say about the US Automotive Aftermarket?

So, after putting these different tools through their paces, what picture did they paint of the US automotive aftermarket? Beyond just the scores and comparisons, what core messages, trends, and warnings emerged? And more importantly, what should leaders in this space actually do about it?

For industry veterans, there’s not much new and certainly nothing major. The tailwinds, headwinds, and recommendations all align with content from MEMA and Auto Care. However, after reading through the various analyses myself some points stuck out to me like sticks in the mud. 

Dealerships: ChatGPT indicated dealerships are gaining traction (https://www.aftermarketmatters.com/national-news/distribution-channels-jockey-for-share-of-difm-growth/). Competition is definitely heating up. While the overall numbers do not reflect any major changes, local markets could experience something different. This is something that all aftermarket players should pay attention to. Independent share could and should be a growing area of attention. EV’s require additional equipment and training costs without a known payback period. Savvy dealership owners and service managers will look to exploit their current resources to offset those investments, making independents an obvious target. 

Consolidation and fallout: There are large changes looming over the industry and not everyone will be able to keep up. Some of these changes, like EV’s and other complex vehicle technologies like ADAS require strategic planning and time to execute. Turnkey solutions, if they exist, will be expensive. This all translates into continued activity in the M&A and bankruptcy space at all levels and channels. A turn in the current momentum for Right to Repair would accelerate the current pace. 

More channel jockeying: E-commerce isn't slowing down in the aftermarket – consumers want convenience, and that's driving changes in how parts actually get where they need to go. It’s forcing everyone to think beyond just stocking shelves in a warehouse. Take drop-shipping – it popped up in the AI research, and it makes sense. Drop-shipping isn’t new, but I think we can all agree it’s on the uptick. For the online players (e-tailers), it's a way to offer a massive catalog without holding tons of inventory, just shipping straight from the manufacturer or a distributor. 

But what about the traditional Warehouse Distributors (WDs)? Drop-shipping could be their ticket to staying relevant in the short to medium term, letting them serve those e-tailer accounts and maybe even facilitate more direct shipments to installers who just need the part now. Alongside this, you've got the whole 'online-to-offline' (o2o) play becoming table stakes. We're talking about BOPIS (Buy Online, Pick-up In Store) and BORIS (Buy Online, Return In Store). It’s how the brick-and-mortar guys – retailers and even some WDs – fight back, mixing online ease with their physical presence. It does mean rethinking logistics, maybe even more direct manufacturer shipments to cover local online orders, but it’s all part of adapting to how people buy parts today

Tailwinds Driving Growth:

The fundamental drivers supporting the aftermarket remain strong:

• Aging Vehicle Fleet: Cars are staying on the road longer than ever. The increasing average age of vehicles continues to be a primary engine for growth, translating directly into more repair and maintenance opportunities over a vehicle's lifetime.

• Growing VIO: The sheer number of vehicles in operation (VIO) continues to climb, expanding the overall size of the addressable market.

• Resilient VMT: Vehicle miles traveled (VMT) have shown resilience, recovering post-pandemic and contributing to wear and tear, thus driving demand for parts and services.

Headwinds Creating Challenges:

Despite the positive fundamentals, several significant challenges require strategic attention:

• Increasing Tech Complexity: The rapid integration of ADAS, complex electronics, and the gradual shift towards EVs are fundamentally changing repair requirements. This necessitates significant investment in new tools, training, and potentially shifts profit pools (e.g., towards dealerships initially for complex EV repairs).

• Persistent Technician Shortage: The struggle to find and retain qualified technicians remains a critical bottleneck across the industry, impacting service capacity, potentially driving shop consolidation, and increasing labor costs.

• Economic Uncertainty & Cost Pressures: Inflation, while potentially easing, continues to pressure consumer budgets and business operating costs. Intense price competition, particularly from large e-commerce platforms, is squeezing margins for traditional players.

• Supply Chain Risks: While improving from peak disruptions, global supply chains remain vulnerable, requiring ongoing focus on resilience, visibility, and diversified sourcing strategies.

• E-commerce & Consolidation: The relentless growth of e-commerce continues to disrupt traditional distribution models. Simultaneously, consolidation continues across distributors, retailers, and service providers as players seek scale and efficiency advantages.

The Wildcard: Right to Repair & Data Access:

Lurking beneath the surface is the ongoing battle over vehicle data access and Right to Repair legislation. Currently a somewhat neutral factor with incremental state-level wins, any significant federal action or shift in manufacturer strategies regarding telematics and diagnostic data could dramatically alter the competitive landscape, particularly for independent service providers. Who controls the data increasingly controls the customer relationship and repair opportunities.

Actionable Recommendations:

Navigating this landscape requires targeted strategies:

• For Parts Manufacturers:

• Prioritize R&D for high-tech components (ADAS, EV, advanced electronics) where expertise commands value.

• Enhance digital capabilities (B2B e-commerce, data syndication) to effectively support all channel partners.

• Evaluate remanufacturing opportunities for complex parts to address cost and sustainability concerns.

• For Distributors:

• Leverage data analytics for superior demand forecasting, inventory optimization, and logistics efficiency.

• Become indispensable partners by providing robust technical support and training for shop customers navigating new technologies.

• Strategically utilize private label programs to compete effectively while managing margins.

• For Retailers (Online & Brick-and-Mortar):

• Build seamless omnichannel experiences bridging online convenience with in-store services (pickup, returns, advice).

• Utilize customer data ethically for personalized marketing and service reminders.

• Differentiate physical stores through expert advice, value-added services (tool rental), and strong community engagement, especially for DIYers.

• For Service Providers (Dealers & Independents):

• Make continuous investment in technician training and diagnostic equipment for complex vehicle systems an absolute priority.

• Focus relentlessly on customer experience – convenience, transparency, and trust are key differentiators.

• Consider strategic specialization (specific technologies, vehicle types) or partnerships to access needed capabilities and maintain competitiveness.

The Bottom Line:

In the coming decade, the aftermarket industry will look somewhat different – more digital, possibly more consolidated in some areas, and dealing with a car parc gradually shifting to electric and connected vehicles. However, the core demand driver remains: millions of vehicles on the road that need parts and service to stay running. As long as Americans rely on personal and commercial vehicles, the aftermarket will remain vital. The companies that thrive will be those that embrace change, leverage technology, and stay attuned to customer needs, while those slow to adapt may see their share slip. If the last five years demonstrated the aftermarket’s resilience (handling economic swings, a pandemic, supply hiccups, and technological change), the next five to ten will showcase its ingenuity and adaptability in the face of new disruptions and trends. The road ahead is filled with both challenges to navigate and growth to capture – a journey the U.S. aftermarket appears well-equipped to undertake.