Industry | Automotive Aftermarket
Automotive aftermarket has a secret weapon for AI commerce. Most merchants aren't using it.
Fitment data is the most structured product data in any retail vertical. It's also completely invisible to AI agents.
Published February 17, 2026 · 9 min read
TL;DR
- 1.Automotive aftermarket has the richest structured product data of any vertical (ACES/PIES). None of it is accessible to AI shopping agents.
- 2.Three breaks in the chain: ACES XML isn't AI-readable, product pages don't expose fitment in structured data, and no feed submission path exists for fitment.
- 3.The merchant who translates ACES into JSON-LD and exposes fitment via API wins the AI commerce channel for their category.
- 4.Your competitors have the same ACES data. They're not using it either. Someone will.
Ask ChatGPT for a Cold Air Intake. See What Happens.
Go to ChatGPT right now and type: "Best cold air intake for a 2019 Subaru WRX."
You'll get a list. Generic recommendations pulled from review sites, forum posts, and whatever product descriptions happened to rank well on Google. Maybe it mentions Mishimoto or Cobb. Maybe it doesn't. Either way, it's not pulling from any aftermarket specialist's catalog. It's not checking fitment. It's not confirming that the part actually fits a 2019 WRX with the FA20DIT engine.
This is bizarre. Because automotive aftermarket is the one vertical that already has the most structured, most detailed product compatibility data of any retail category on the planet.
The paradox:
ACES and PIES files contain year/make/model/engine fitment data for millions of parts, encoded in a standardized format that's been in use for over two decades. This is exactly what an AI shopping agent needs to make confident, specific recommendations. But almost nobody is exposing it in a format agents can read.
Why Fitment Data Is Perfect for AI Agents
Most product queries are vague. "Good running shoes." "Best laptop under $1,000." Agents have to interpret intent, weigh dozens of attributes, and make subjective recommendations.
Automotive queries are different. They're specific by nature. Year. Make. Model. Engine. Submodel. A customer doesn't ask for "a good turbo." They ask for "a turbo for my 2016 Ford F-150 3.5L EcoBoost." That precision maps directly to structured data.
The ACES/PIES Stack
ACES (Fitment)
Encodes exactly which vehicles a part fits. Uses VCdb (Vehicle Configuration Database) with 83 reference tables covering:
- Year, make, model, submodel
- Engine type, displacement, fuel type
- Transmission, drivetrain
- Bed type, body style, brake system
- Position on vehicle (front/rear, left/right)
PIES (Product Info)
Carries everything else an agent needs to make a recommendation:
- Technical specifications and dimensions
- Marketing descriptions and bullet points
- High-resolution product images
- Pricing and packaging data
- Hazmat flags, weight, country of origin
Together, ACES + PIES give AI agents everything they need: "Does it fit?" + "What is it?" Richer structured data than any other retail vertical.
No other vertical has this. Fashion doesn't have standardized sizing across brands. Electronics compatibility is fragmented. Home improvement has no universal fitment standard.
Automotive aftermarket has a 20-year head start on structured product data. The problem isn't that the data doesn't exist. The problem is that none of it is getting to AI agents.
Where the Data Gets Stuck
There are three breaks in the chain between your ACES/PIES data and an AI agent's recommendation.
Break 1: ACES is machine-readable, but not AI-readable
ACES files are XML. They were designed for EDI-era data exchange between manufacturers, distributors, and retailers. They work great for loading into e-commerce platforms and catalog databases.
But AI agents don't read XML files. They read JSON APIs, JSON-LD structured data, and product feeds. Your ACES data might be perfectly structured internally, but from an AI agent's perspective, it doesn't exist.
Break 2: Product pages don't expose fitment in structured data
Most aftermarket product pages show fitment through JavaScript-rendered dropdowns or compatibility widgets. A human can see it. A crawler can't. Google can't index it. AI agents can't parse it.
Even merchants with good Schema.org markup typically only include basic Product schema -- name, price, image, availability. The fitment data that makes their products uniquely matchable is completely absent from structured data.
Break 3: No feed submission path for fitment
The Agent Commerce Protocol (ACP) defines fields for product title, description, price, images, and identifiers. It doesn't have automotive-specific fitment fields. Google's UCP doesn't either.
There's no standardized way to tell ChatGPT "this cold air intake fits 2015-2021 Subaru WRX with the FA20 engine." You can stuff it into the description field, but that's not structured. It's just text. And agents trust structured data over free text.
The Merchant Who Figures This Out First Wins
Structured fitment data gives AI agents confidence to recommend. That's the game.
When a customer asks ChatGPT "what cold air intake should I get for my 2019 WRX?" the agent has two choices:
Without structured fitment
"Here are some popular cold air intakes for the WRX. Please verify compatibility before purchasing."
Generic. Hedged. No purchase confidence.
With structured fitment
"The Mishimoto MMAI-WRX-15 fits your 2019 Subaru WRX with the FA20DIT engine. Includes heat shield. 10 HP gain at the wheels. In stock. Would you like to add it to cart?"
Specific. Confident. Purchasable.
"Fits 2015-2021 Subaru WRX, engine code FA20" beats "WRX cold air intake" every time. The first is a match. The second is a guess.
Now scale that across 50,000 parts. Every part with structured fitment data becomes a specific, confident recommendation for every matching vehicle query. That's an enormous advantage when AI agents are choosing which 2-3 products to show.
The compounding advantage
The first merchant to expose structured fitment data doesn't just win today's queries. They accumulate performance data, conversion history, and agent trust scores that make them harder to displace over time. First mover advantage in agentic commerce is real.
What to Do About It
Four steps. None of them require rebuilding your platform.
Get your ACES/PIES data in order
If you're a distributor, you already have it. Your suppliers (manufacturers) provide ACES and PIES files. Distributors like Turn 14, Meyer Distributing, and LKQ ship updated ACES data regularly.
If you're a manufacturer, you're creating it. Make sure your ACES files are current, your VCdb references are up to date, and your PIES data includes complete product attributes. The Auto Care Association VCdb specification [1] is the reference standard.
Build a translation layer: ACES XML to JSON-LD + JSON feeds
This is the critical step. Take your ACES XML and transform it into formats AI agents can consume:
- JSON-LD on product pages: Embed fitment data in Schema.org Product markup using additionalProperty fields for year, make, model, engine compatibility
- JSON product feeds: Include fitment arrays in your ACP feed [2] with structured vehicle application data, not just text descriptions
- Denormalized records: One product with 200 fitment applications becomes 200 queryable records, not one product with a dropdown
Expose fitment through an API
Build a fitment verification endpoint that AI agents can query in real time:
GET /api/fitment?part=MMAI-WRX-15&year=2019&make=Subaru&model=WRX
{
"fits": true,
"confidence": "exact_match",
"engine": "FA20DIT",
"position": "engine_bay",
"notes": "Direct bolt-on. No modifications required.",
"related_parts": ["MMHOSE-WRX-15", "MMSK-WRX-15"]
}Sub-100ms response time. Real-time inventory check included. This is what turns a "maybe" recommendation into a "would you like to add it to cart?"
Run the audit
Before you build anything, see where you stand. The AI Commerce Readiness Audit checks your product pages for structured data, Schema.org markup, and AI agent infrastructure. 60 seconds. Free. Every aftermarket merchant we've tested so far has scored zero on infrastructure readiness.
Your Competitors Have the Same Data. They're Not Using It Either.
Every aftermarket merchant and distributor has access to the same ACES data. The manufacturers ship it to everyone. It's table stakes for catalog operations.
But right now, nobody is translating it for AI agents. Nobody is exposing fitment as structured data on product pages. Nobody is building real-time fitment verification APIs.
Someone will.
The aftermarket merchant who does it first gets specific, confident recommendations from ChatGPT, Gemini, and Perplexity for every vehicle application in their catalog. 50,000 parts times dozens of fitment applications each. That's millions of specific, purchasable matches that competitors can't touch until they do the same work.
The data already exists. The question is who translates it first.
See what AI agents see when they look at your parts catalog
60 seconds. Free. No signup required. Every aftermarket merchant we've tested scores zero on infrastructure.
Run Your Free AuditRelated Articles
Sub-100ms or Invisible: The Speed AI Agents Require
Your API responds in 500ms. Your competitor's responds in 80ms. Guess who gets the sale?
The Hidden Advantage: Why Your Messy Data Is Actually an Opportunity
Your competitors have prettier websites. You might have better data. That matters now.
The Agentic Commerce Checklist for Q1 2026
A practical action plan for merchants who can't wait for their platform to figure this out.
Your ACES data is a weapon. Use it.
We help aftermarket merchants translate ACES/PIES into AI-readable formats and expose fitment through real-time APIs. Your competitors have the same data. We help you get there first.