Key findings at a glance
Executive Summary
AI shopping assistants recommended the merchant that actually sells a part only 37.4% of the time. In StrikeTru’s 2026 audit of 998 product pages across 10 auto parts merchants, an AI assistant was asked where a consumer would buy each exact product. Nearly two-thirds of the time it pointed shoppers elsewhere, usually to Amazon, eBay, RockAuto or CarParts.com, even when the audited merchant stocked the part.
The cause is not a lack of products or basic SEO. It is product data that AI systems cannot easily read, trust or compare. Across the audit we found the same fixable gaps:
- Fitment hidden behind vehicle-selector dropdowns instead of stated in readable text (every fitment dimension scored below 2.0 out of 5)
- Almost no FAQ or comparison content for an AI to quote when a shopper asks “will this fit?” or “which should I buy?”
- Missing Product schema at three of ten merchants, and incomplete schema at most others
- Thin numeric specifications, reviews and video, plus bot protection that blocked some pages from automated readers entirely
Auto parts is one of the hardest categories to get right. Fitment complexity, huge SKU counts and the split between OEM and aftermarket data make it unforgiving. That also makes it a strong test case: the gaps found here show up, in milder form, in most ecommerce categories.
Where does your catalog land?
Our Automotive Catalog & Fitment Assessment applies the same framework used in this report to your own products: fitment, product data, schema and AI-search readiness.
About This Research
What we measured
StrikeTru audited 998 product pages (about 100 per site) at 10 auto parts merchant websites. Each product was scored on a 1–5 scale across four dimensions, then tested in an AI shopping assistant.
| Dimension | What it covers |
|---|---|
| Product data completeness | 24 fields, including titles, descriptions, features, benefits, specifications, weights and dimensions, attributes, images, video, reviews, ratings and pricing |
| AI readiness | FAQ availability, comparison content, buying guidance, semantic richness, natural-language coverage, product schema, knowledge-graph and vector-search readiness |
| Automotive fitment readiness | Year/Make/Model (YMM) coverage, compatibility depth, fitment notes and explanation quality, OEM cross-references, exploded views, repair jobs and symptom-to-part relationships |
| Technical schema | JSON-LD structured-data audit scored out of 100 for valid, complete Product schema |
| AI discoverability test | An AI assistant was asked: “Where would a consumer buy this exact product?” We recorded whether the audited merchant appeared among the top 5 recommended retailers, and at what rank |
Study details
| AI assistant(s) queried | Deep Seek V4 Flash / version "0424" / Using API Key |
| Query format | "Where are the top 5 places consumers go to buy the {product_name}? " "Then check your own list and report exactly where {brand} appears." |
| Audit window | July 7 - 18 |
| Merchant selection | Merchants were selected based on their relevance to the aftermarket and truck-parts segments, breadth and complexity of product catalogs, and the potential impact of product discoverability on online customer research and purchase journeys. |
| Product sampling | Category stratified samples. Products were chosen from multiple categories. |
| Scoring approach | Automated crawl + AI assisted scoring with human review of sample responses |
| Detailed-text analysis | Of the 998 pages, 968 had complete auditor notes; percentages quoted from auditor notes use that base |
| Anonymization | Merchants are labeled A–J; product examples are described generically |
Limitations
- AI answers vary with time, wording, location and personalization. Results describe the conditions of this test, not a permanent ranking.
- Ten merchants is a focused sample of the auto parts category, not a statistically representative view of all ecommerce.
- The findings show associations between product data quality and AI visibility; they do not prove that any single factor causes a recommendation.
- Sixteen pages returned a bot-verification screen instead of content; their scores reflect what an automated reader could see.
How AI Is Changing Product Discovery
What is AI product discoverability?
AI product discoverability is how likely AI search engines, shopping assistants and agents are to find, understand and recommend a merchant’s product for a relevant shopper question. It depends on product data that machines can read, verify and compare, not only on keyword rankings.
For two decades, online product discovery followed one path: search, browse several sites, compare pages, read reviews, buy. Success depended on SEO, paid ads and a good storefront.
Generative AI adds a new path. Shoppers now ask full questions, such as “What’s the best brake pad set for towing under $150?”, and get a synthesized answer with recommended products and places to buy. The AI reads descriptions, specifications, attributes, reviews, structured data and compatibility information, then decides what to say. Often the shopper sees a recommendation before visiting any merchant’s site.
The shift is already measurable
- Traffic: Adobe reported that traffic from generative AI sources to US retail websites rose 1,300% year over year during the 2024 holiday season, and 39% of surveyed US consumers had already used generative AI for shopping. [1]
- Checkout inside AI: In September 2025, OpenAI launched Instant Checkout in ChatGPT with the open Agentic Commerce Protocol, co-developed with Stripe. [2]
- Agentic commerce standards: In January 2026, Google announced the Universal Commerce Protocol, built with Shopify, Etsy, Wayfair and Target, to power checkout for eligible listings in AI Mode and the Gemini app. [3]
Four AI surfaces merchants now need to serve
| Surface | What it does for the shopper |
|---|---|
| AI search summaries | Search engines synthesize answers, comparisons and recommendations above or instead of the traditional list of links |
| AI shopping assistants | Tools such as ChatGPT let shoppers describe needs in natural language, ask follow-ups and compare products across brands |
| Conversational commerce | AI guides the purchase by asking clarifying questions (vehicle, budget, use case) and narrowing options |
| Agentic commerce | AI agents act for the shopper: comparing prices, checking stock, confirming compatibility and increasingly completing the purchase |
What AI needs from a product page
| Content AI can explain and quote | Data AI can compare and verify |
|---|---|
| Product-specific descriptions and benefits | Structured specifications and standardized attributes |
| FAQs and how-to content | Weights, dimensions and measurements |
| Comparison and buying guidance | Fitment / compatibility stated as text |
| Reviews, images and SKU-specific video | Product schema: brand, identifiers, price, availability |
When this information is missing, AI systems either overlook the product or fill the gap with a competitor’s page that does provide it. Many small and mid-sized merchants still manage product information in spreadsheets, disconnected systems or directly in the storefront. The fix is content and data work, not a platform migration.
What We Found: Nine Findings From the Auto Parts Audit
Finding 1: Most merchants are invisible to AI assistants for their own products
Across all 998 products, the audited merchant appeared among the AI’s top 5 recommended places to buy only 37.4% of the time. In the other 62.6% of cases, the AI recommended different retailers, almost always a large marketplace (Amazon, eBay) or a parts aggregator (RockAuto, CarParts.com), even when the audited merchant stocked the part.
View data table
| Merchant | AI discoverability rate | Avg. rank when it appeared |
|---|---|---|
| Merchant C | 69.0% | #2.4 |
| Merchant A | 60.0% | #4.4 |
| Merchant I | 57.1% | #3.0 |
| Merchant G | 43.0% | #3.2 |
| Merchant J | 37.0% | #3.0 |
| Merchant B | 32.0% | #4.2 |
| Merchant E | 32.0% | #4.5 |
| Merchant D | 18.0% | #4.7 |
| Merchant H | 14.0% | #4.6 |
| Merchant F | 12.2% | #5.0 |
Technical readiness and AI outcomes don’t always move together. Merchant F had one of the best schema scores in the audit (95/100) but the lowest AI discoverability (12.2%). Merchant C had no detectable Product schema yet led the field (69.0%). Structured data appears necessary but not sufficient: brand authority, content depth and third-party signals such as reviews, links and marketplace presence also shape whether an AI recommends a merchant directly.
Key takeaway
Being indexed is not the same as being recommended. AI assistants default to marketplaces unless a merchant’s own product data gives them a clear, trustworthy reason not to.
Finding 2: Descriptions say what a part is, not why to buy it
Descriptions (3.45/5) and Features (3.38/5) were the strongest scores in the audit, but Benefits scored only 2.70/5. Descriptions usually restate what a part is (a brand paragraph or a spec list) rather than explain what it does for the buyer, which is the framing an AI needs to justify a recommendation.
One air-intake sensor listing had a description that was “extremely weak, consisting only of the product title repeated.” Generic brand boilerplate (for example, “meets or exceeds OEM standards”) stood in for product-specific copy on dozens of pages. An AI asked “why should I buy this part?” has nothing to draw on, so it uses a competitor’s page instead.
Finding 3: Structured specifications thin out fast
Technical Specifications averaged 2.97/5, but the more precise the data, the lower the score: Weights & Dimensions 2.18/5, Measurement Attributes 2.00/5. In the auditor notes (base: 968 products):
- 43.8% of product pages had no weight or dimension data at all.
- 79.8% had no downloadable spec sheet, product sheet or installation PDF.
- 30.4% had missing or meaningless image alt text, such as one generic tag reused on 10–20 images.
Finding 4: FAQs are almost entirely absent
FAQ Availability averaged 1.44/5, among the lowest scores in the audit. Nearly every platform had a Q&A module, but it was empty: “No questions have been answered for this product yet.” The tooling exists. What’s missing are the three to five answers a buyer, or an AI answering for one, needs before purchasing.
Finding 5: Fitment data is the category’s biggest blind spot
All eight automotive fitment-readiness dimensions scored below 2.0/5.
Most merchants rely on an interactive “select your vehicle” tool rather than showing compatibility as readable text. That works for a person clicking dropdowns, but an AI system generally can’t operate a dropdown and needs the compatibility list stated on the page. OEM Cross-References scored 1.60/5: even when a part number appears, the OEM reference numbers shoppers and AI systems actually search for are usually missing.
Make your fitment readable to AI
We design fitment data models that keep the vehicle selector for shoppers while publishing YMM compatibility, fitment notes and OEM cross-references as crawlable content, generated from your PIM and ACES data.
Finding 6: Reviews and real product video are largely missing
Reviews averaged 1.54/5 and Ratings 1.76/5. Videos averaged 1.31/5, the lowest score among the 24 data-completeness fields, and the real gap is larger: on 49.9% of products the only “video” detected was a tag-manager script that shared a video-like URL pattern, not a video at all.
Finding 7: Attributes are inconsistent, and content is built for one market
Attributes averaged 2.68/5. Pages typically listed 3–5 basics (brand, condition, part number) but lacked material, finish, connector-type and compatibility attributes. 81.9% of products had content only in English with no hreflang, and 41.3% showed prices only in USD with no currency option.
Finding 8: Schema markup is either solid or completely absent
Seven of ten merchants had working Product schema, scoring 66.7–95 out of 100. Three merchants (C, E and H) had no detectable Product schema on any audited page. Among merchants with schema, these fields were most often missing:
| Missing schema field | Occurrences across audited products |
|---|---|
| additionalProperty | 564 |
| shippingDetails | 551 |
| offers.priceValidUntil | 367 |
| itemCondition | 234 |
| sku / mpn / gtin | 199 |
| offers.url | 199 |
| brand | 198 |
Finding 9: Some pages were unreadable to automated visitors
On 15 of 100 audited products at Merchant C and 1 of 100 at Merchant D, the audit tool received a bot-verification screen instead of the product page. This is an access problem, not a content problem, and it likely affects AI crawlers and shopping agents the same way. A merchant can have excellent product content and still be invisible if automated systems can’t get past the front door.
The Eight Gaps Behind the 37% Headline
| # | Gap | Score | Why it blocks AI discoverability | First fix |
|---|---|---|---|---|
| 1 | No FAQ content | 1.44/5 | Nothing on the page answers “will this fit?” or “how does it compare to OEM?”, so the AI quotes a competitor | Pre-populate 3–5 FAQs on top SKUs |
| 2 | No comparison or buying guidance | 1.13 · 2.32/5 | The AI can’t frame “which should I buy?” from your content | Add “how to choose” and “vs OEM” sections |
| 3 | Missing Product schema | 3 of 10 at 0/100 | Price, availability, brand and identifiers can’t be reliably extracted | Implement complete Product JSON-LD |
| 4 | Fitment not machine-readable | All 8 < 2.0/5 | In auto parts, fitment is the product; dropdown-only data is effectively invisible | Publish YMM tables and OEM cross-refs as text |
| 5 | No real product video | 1.31/5 | Misses a growing multimodal signal of legitimacy and usage | SKU-specific video for top sellers |
| 6 | Sparse reviews and ratings | 1.54 · 1.76/5 | Less social proof for an AI weighing which retailer to recommend | Post-purchase review requests |
| 7 | Shallow numeric attributes | 2.00–2.68/5 | AI can summarize but can’t compare on measurable specs | Standardize dimensions and key specs |
| 8 | Bot protection blocking access | 16 pages blocked | A page automated readers can’t load is equivalent to no page | Review bot rules; allow verified crawlers |
Close these gaps faster
Governed enrichment fills missing attributes, writes benefit-led descriptions and FAQs, and publishes complete schema from one source of truth, without letting AI invent fitment or specifications.
The AI Discoverability Maturity Model
Use this five-level model to benchmark where a catalog stands today and what to do next.
| Level | Typical characteristics | Business impact | Primary focus |
|---|---|---|---|
| 1. Search Invisible | Thin descriptions, missing specs, few attributes, little structured data, duplicate manufacturer content | Hard to rank or cite; heavy reliance on paid ads | Build a complete, consistent catalog |
| 2. SEO Ready | Optimized titles and meta, better descriptions, basic categories, some reviews, limited specs | Search visibility improves; AI still struggles with detailed questions | Enrich structured data; standardize attributes |
| 3. Structured Catalog | Comprehensive attributes and specs, clear variants, consistent taxonomy, quality assets, product relationships | AI has far more context; operations become more efficient | Add AI-oriented content and richer schema |
| 4. AI Discoverable | Rich descriptions, FAQs, comparison content, fitment as text, video, complete schema, buying guides, consistency across channels | Much more likely to appear in AI recommendations | Optimize for AI shopping with governance |
| 5. Agent Ready | Real-time availability and pricing via APIs, product knowledge graphs, automated syndication, continuous governance | Products can be evaluated and bought by autonomous agents | Keep adapting to new AI commerce protocols |
Where the audited merchants land
No merchant reached Level 4 or 5. Four merchants sit at Level 2 (C, E and H had no Product schema; I had a data-completeness average below 2.4/5) and six at Level 3 (A, B, D, F, G, J): working schema and stronger data, but missing FAQs, comparison content and fitment as text. The findings concentrate in the gap between Level 3 and Level 4.
Which level is your catalog at?
Get a prioritized plan to reach Level 4, covering fitment, product data, schema, marketplace and AI-search readiness.
How to Improve AI Discoverability
If you’re at Level 1–2: build the foundation
- Consolidate product data in a PIM instead of editing it page by page in the storefront.
- Standardize a core attribute set (material, dimensions, condition, compatibility).
- Implement complete Product schema. For a merchant at 0/100, this closes a hard technical gap.
If you’re at Level 3: build content for AI, not just for browsing
- Write FAQs into product pages instead of relying on an empty “ask a question” form.
- Publish fitment as text (YMM tables, fitment notes, OEM cross-references) alongside the vehicle selector.
- Add comparison and buying guidance, even a short “how to choose” or “vs OEM” section.
- Produce SKU-specific video for high-volume or high-return categories first.
If you’re approaching Level 4: stay consistent and measure
- Syndicate the same enriched data everywhere: storefront, marketplaces, Google Merchant Center and AI shopping integrations.
- Test AI discoverability monthly with a fixed set of product queries.
- Prepare for Level 5: expose real-time price and availability through APIs as agentic checkout matures.
The auto parts fitment playbook
| Do this | Why it helps AI discoverability |
|---|---|
| Generate fitment content from your PIM and ACES data | One governed source keeps storefront, marketplaces and AI-readable text consistent |
| Show a YMM compatibility table in HTML on each product page | AI systems can read and cite the vehicles a part fits without operating a dropdown |
| List OEM and interchange part numbers | Matches the identifiers shoppers and AI systems actually search for |
| Write fitment notes in plain language (qualifiers, position, exceptions) | Reduces wrong-fit recommendations and returns |
| Keep the vehicle selector as the shopper experience | People keep the convenience; machines get the data |
Seven quick wins for the next 30 days
- Check bot protection. Make sure security rules aren’t serving verification screens to legitimate search and AI crawlers.
- Fix schema basics. Add or complete Product schema with brand, sku/mpn/gtin, offers and itemCondition.
- Answer the obvious questions. Add 3–5 FAQs to your top 50–100 SKUs.
- Expose fitment. Render YMM compatibility as a text table on your best-selling parts.
- Replace boilerplate. Rewrite brand-paragraph descriptions on top sellers with benefit-led copy.
- Repair alt text. Give each product image unique, descriptive alt text.
- Start measuring. Ask AI assistants where to buy 20 of your products every month and track how often you’re named.
The bottom line
Every gap in this report traces back to one root cause: product data built for a person scrolling a page, not for a machine answering a question. None of the ten merchants needed a platform migration to move up the maturity model. They needed to treat product data as a strategic asset rather than a by-product of the storefront.
Frequently Asked Questions
What is AI product discoverability?
AI product discoverability is how likely AI search engines, shopping assistants and agents are to find, understand and recommend your product for a relevant question. It depends on complete, structured, machine-readable product data such as specifications, fitment, FAQs and schema, as well as trust signals like reviews.
Why don’t AI assistants recommend my store for parts I sell?
In our audit, AI assistants named the merchant only 37.4% of the time. They tend to default to marketplaces when a merchant’s own page lacks readable fitment, FAQs, comparison content, complete schema or reviews, or when bot protection blocks automated access.
Does Product schema guarantee AI visibility?
No. Schema helps machines extract price, availability, brand and identifiers, but it isn’t sufficient on its own. One merchant scored 95/100 on schema yet had the lowest AI discoverability rate (12.2%).
How should auto parts merchants show fitment to AI?
Publish fitment as readable text on the product page: a Year/Make/Model compatibility table, fitment notes and OEM cross-reference numbers, ideally generated from PIM and ACES data. Keep the vehicle selector for shoppers, but don’t rely on it alone.
What is the fastest way to improve AI discoverability?
Start with access and basics: make sure bot rules don’t block legitimate crawlers, complete your Product schema, and add three to five FAQs and a readable fitment table to your best-selling SKUs. Then measure monthly.
What is agentic commerce, and why does it matter for auto parts?
Agentic commerce is when AI agents research, compare and buy on a shopper’s behalf. For auto parts, an agent must confirm compatibility before buying, so machine-readable fitment becomes essential.
How can I test my own catalog?
Ask AI assistants where to buy a sample of your products and record how often you’re named, or request StrikeTru’s Automotive Catalog & Fitment Assessment, which uses the framework in this report.
Find out where your catalog lands
Get a clear, prioritized view of your fitment, product data, schema and AI-search readiness, based on the same framework used in this report.
How to cite this report
StrikeTru. The 2026 Auto Parts AI Discoverability Report. striketru.com/solutions/automotive/ai-visibility-research-report/Charts and statistics may be reused with attribution and a link to this report.
About StrikeTru
StrikeTru helps auto parts manufacturers, distributors and retailers turn fragmented product data into a foundation for commerce today and AI-driven discovery next, covering PIM/PXM implementation, fitment and ACES/PIES data management, product data enrichment, and B2B and B2C commerce on Akeneo, Shopify, BigCommerce, Adobe Commerce and Miva. StrikeTru also offers TruPIM, its own PIM for growing businesses.
Sources
- Adobe. “Adobe Analytics: Traffic to U.S. retail websites from Generative AI sources jumps 1,200 percent.” March 17, 2025. blog.adobe.com
- OpenAI. “Buy it in ChatGPT: Instant Checkout and the Agentic Commerce Protocol.” September 29, 2025. openai.com
- Search Engine Land. “Google launches Universal Commerce Protocol for agent-led shopping.” January 11, 2026. searchengineland.com