Last updated July 2026
Procurement is changing faster than most manufacturers realize. Gartner predicted in November 2025 that by 2028, 90% of B2B buying will be AI-agent-intermediated, routing more than $15 trillion in spend through machine-to-machine exchanges. That is not a distant forecast. Engineering buyers are already using ChatGPT, Perplexity, and Microsoft Copilot to source components today.
If your catalog pages cannot be read by a software agent, they do not exist in that buying process.
This is an underserved problem. Most AEO content focuses on SaaS, retail, and media. Very little addresses what industrial manufacturers specifically need to do. That is the gap this playbook fills.
The two-part AI procurement audit
Before you fix anything, run these two checks. They take under 30 minutes and reveal the biggest gaps.
Audit 1: Crawler access
Fetch your robots.txt file directly:
https://yourdomain.com/robots.txtLook for any of these:
User-agent: GPTBot
Disallow: /
User-agent: ClaudeBot
Disallow: /
User-agent: PerplexityBot
Disallow: /Also check for a blanket block:
User-agent: *
Disallow: /A blanket block is common on sites that were locked down during the initial ChatGPT scraping panic of 2023. Many manufacturers set it and forgot it. The cost is real. Similarweb data from August 2025 showed that Amazon, which deliberately blocks AI crawlers, received less than 3% of its referral traffic from ChatGPT. Retailers with open crawler access, including Walmart and Etsy, received 10 to 20 percent or more from the same source. The same access dynamic applies to supplier catalogs: if the bot cannot reach the page, the AI cannot cite it.
Also check your CDN or WAF configuration. Cloudflare, Akamai, and similar services often have bot-filtering rules that block non-browser user agents at the network layer, before your robots.txt is even read.
Audit 2: Specification-schema coverage
For each of your top catalog pages, check whether the following are present:
| Audit point | Check method | Pass condition |
|---|---|---|
| Product schema (JSON-LD) | View source or use Google Rich Results Test | @type: "Product" present with name, description, mpn or gtin |
| Specification table visible in HTML | View source | Table or definition list visible in raw HTML, not rendered by JavaScript alone |
| Attributes in PropertyValue format | View source | Each spec row mapped to PropertyValue with propertyID, value, and unitCode |
| FAQPage schema | View source | At least one FAQ block covering application questions |
| Crawler access | robots.txt + CDN settings | No Disallow for GPTBot, ClaudeBot, PerplexityBot |
| Front-loaded summary paragraph | Page inspection | First 60 words name the product, part type, and two to three key specs |
A page that passes all six points is procurement-agent-ready. Most manufacturer pages pass none.
Step 1: Open crawler access
Fix your robots.txt first. Add explicit allow rules for the main AI crawlers:
User-agent: GPTBot
Allow: /
User-agent: ClaudeBot
Allow: /
User-agent: PerplexityBot
Allow: /
User-agent: anthropic-ai
Allow: /Then audit your CDN and WAF. If you use Cloudflare, check the Bot Management settings for rules blocking “other bots” or custom bot categories. If you use Akamai or Fastly, review your bot-traffic policies. Many WAF default settings treat any non-browser user agent as a threat.
One important note: opening access does not mean exposing proprietary content. You can still restrict login-gated configuration tools, customer portals, and pricing pages. The goal is to allow access to your public product catalog pages, where your spec sheets live.
Step 2: Structure your spec sheet for machine reading
The way most manufacturer spec sheets are built, they work well for PDF generation but fail AI retrieval. They rely on formatting, column alignment, and visual hierarchy that AI systems cannot parse from HTML.
A machine-readable spec sheet has a specific structure.
Before (typical manufacturer page)
A product description paragraph written for SEO keyword density. Then an embedded PDF viewer containing the spec table. Then a contact form. No structured data. No HTML-accessible attribute values.
An AI buying agent that visits this page extracts: the product name (maybe), the description (partially), and nothing from the spec table.
After (AI-procurement-ready page)
Front-loaded summary (40 to 60 words, machine-parseable):
The XR-7200 is a brushless DC servo motor rated at 400W continuous, 1200W peak. Operating range: 12 to 48V DC, -40 to +85 degrees C. IP67 rated. Available in 42mm, 57mm, and 86mm frame sizes. Hollow shaft and standard shaft variants in stock.
Structured specification table (HTML Markdown, not PDF):
| Attribute | Value | Unit |
|---|---|---|
| Continuous power | 400 | W |
| Peak power | 1200 | W |
| Voltage range | 12 to 48 | V DC |
| Operating temperature | -40 to +85 | degrees C |
| Protection rating | IP67 | |
| Frame sizes | 42, 57, 86 | mm |
| Shaft options | Hollow, Standard | |
| Weight (42mm) | 320 | g |
JSON-LD schema (in <head> or inline):
{
"@context": "https://schema.org",
"@type": "Product",
"name": "XR-7200 Brushless DC Servo Motor",
"mpn": "XR-7200",
"description": "400W continuous brushless DC servo motor, 12-48V, IP67, operating range -40 to +85 C.",
"brand": { "@type": "Brand", "name": "YourBrand" },
"additionalProperty": [
{
"@type": "PropertyValue",
"propertyID": "continuous_power",
"name": "Continuous Power",
"value": "400",
"unitCode": "W",
"unitText": "Watts"
},
{
"@type": "PropertyValue",
"propertyID": "voltage_range",
"name": "Voltage Range",
"value": "12-48",
"unitCode": "VDC",
"unitText": "V DC"
},
{
"@type": "PropertyValue",
"propertyID": "operating_temperature",
"name": "Operating Temperature",
"value": "-40 to +85",
"unitCode": "CEL",
"unitText": "degrees Celsius"
},
{
"@type": "PropertyValue",
"propertyID": "ip_rating",
"name": "Protection Rating",
"value": "IP67"
}
]
}An AI buying agent that visits the “after” page can extract: product name, part number, all specification attributes with values and units, and the summary paragraph. It can compare your motor to a competitor’s on any attribute a procurement agent asks about.
Step 3: Add FAQPage schema for application questions
Procurement agents do not just ask “find me a 400W motor.” They ask questions like:
- “Which brushless servo motors operate in -40 C environments?”
- “What is the IP rating of the XR-7200?”
- “Is the XR-7200 compatible with a 48V bus?”
FAQPage schema lets your product page answer these directly. Add it for the three to five most common application questions for each component category.
{
"@type": "FAQPage",
"mainEntity": [
{
"@type": "Question",
"name": "What environments is the XR-7200 rated for?",
"acceptedAnswer": {
"@type": "Answer",
"text": "The XR-7200 is rated IP67 and operates from -40 to +85 degrees C, making it suitable for outdoor, washdown, and industrial cold-chain environments."
}
},
{
"@type": "Question",
"name": "Is the XR-7200 compatible with 48V bus systems?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Yes. The XR-7200 accepts input voltage from 12 to 48V DC, covering standard 24V and 48V bus architectures without a step-down converter."
}
}
]
}Step 4: Map your catalog to real buyer queries
Your spec sheet is the answer. But you need to know what the question is.
AI buying agents phrase procurement queries in ways that do not match your part numbers or internal product names. “Sealed motor for food processing line” may describe your XR-7200-IP67 variant, but your page title says nothing of the kind.
Temso covers this end to end: crawler-access audit, schema generation, and ongoing citation monitoring across eight AI engines, starting at $89/mo. For a manufacturer running a catalog of hundreds of component pages, it is the clearest starting point. You can see which catalog pages are being cited by AI systems when buyers ask sourcing questions, and which are invisible.
Profound adds a layer that Temso does not: real prompt-volume data showing what procurement professionals actually type into AI systems when sourcing components. That data lets you write summary paragraphs and FAQ answers in the language real buyers use, rather than in the language of your internal catalog taxonomy.
Semrush adds technical site-audit depth for larger catalog teams, including schema validation at scale and crawlability reporting across thousands of URLs. For teams already using Semrush for traditional SEO, the schema audit integrates into existing workflow.
Surfer’s content editor scores individual product pages against AI citation benchmarks in real time. For manufacturers building out new catalog content, it validates that each new spec sheet meets the structural criteria before publication.
None of these tools replace the core work: structuring your spec data. They help you audit where you stand, generate the schema, and monitor whether AI systems are actually retrieving your pages.
Step 5: Build a comparison-ready catalog structure
AI buying agents do not evaluate components in isolation. They compare. An engineer prompts: “Compare three servo motors under 500W continuous, IP67, operating from -30 C to +85 C.” The AI system retrieves candidate pages, extracts specification attributes, and generates a comparison table in the response.
For your page to appear in that comparison, two things must be true:
- Your page must be crawlable.
- Your specification attributes must be in a format the AI can extract and compare.
According to AirOps Research (April 2026, based on 217,508 retrieved pages across 7,500 commercial prompts), comparison pages containing three or more structured tables earn 25.7% more AI citations than equivalent pages without them. For manufacturer catalog pages, this means a full specification table, a comparison table against the prior model version, and an application-suitability table (recommended use cases and exclusions) all on the same page.
That structure also makes it easier for a procurement engineer to verify the AI-generated comparison against your source data. Trust in your citations depends on your source page being clearly structured and verifiable.
Step 6: Maintain attribute consistency across the catalog
AI systems build a model of your product line from the set of pages they retrieve. Inconsistent attribute naming across your catalog degrades that model.
If your 42mm motor calls the specification “Operating Temp” and your 86mm motor calls it “Thermal Range,” an AI buying agent may not recognize them as the same attribute. When a buyer asks for all motors operating below -30 C, your 86mm variant may be excluded.
Consistency rules to enforce across your catalog:
- Use identical
propertyIDvalues in JSON-LD for the same attribute across all products. - Use identical unit systems (SI unless your market requires otherwise; do not mix mm and inches across the same product line).
- Use identical terminology for protection ratings, certifications, and compliance standards.
- Ensure part numbers appear in both the visible content and the
mpnfield of Product schema.
A glossary page at /glossary for your specification terminology, linked from catalog pages, gives AI systems an additional structured reference for attribute definitions. This helps when buyers phrase queries using informal or cross-industry terminology.
The AI procurement-readiness scorecard
Use this scorecard to prioritize which catalog pages to fix first. Score each page from zero to five points:
| Criterion | 0 points | 1 point |
|---|---|---|
| AI crawler access | GPTBot or equivalent blocked | GPTBot and major crawlers allowed |
| Front-loaded summary paragraph (40 to 60 words) | None | Present, names key specs |
| HTML specification table | PDF only or JavaScript-rendered | HTML-accessible table |
| Product schema with PropertyValue attributes | None | Present and valid |
| FAQPage schema with application questions | None | 2 to 5 questions present |
Pages scoring zero to one are invisible to AI procurement agents. Pages scoring four to five are retrieval-ready. Start with your ten highest-traffic catalog pages and work from there.
What to read next
- AEO tool ranking for the full monitor-to-publish loop: /rankings/aeo-tools
- Temso tool profile (catalog audit and schema generation): /tools/temso
- Profound tool profile (demand intelligence and prompt volumes): /tools/profound
- AEO glossary (schema terms, citation mechanics): /glossary
- How we score AEO tools: /methodology
If you want to see where your catalog stands today, Temso runs a crawler-access and schema audit in minutes at $89/mo, with no specialist SEO knowledge required. Start there before investing in schema generation at scale.