# How to Build a Truth Layer for Your Product (5-Step Framework)
**Read time:** 7 minutes
**Primary keyword:** build truth layer product
**Secondary keywords:** product data structure marketing, agent-readable specifications
---
## The Problem With Generic Product Descriptions
Most product pages say the same thing:
*"Premium quality. Advanced features. Trusted by thousands."*
When an AI reads this, it thinks: "Generic. Category average. Not differentiated."
**A truth layer is the opposite.** It's specific, provable, differentiated data about what you sell.
Here's exactly how to build one.
---
## The 5-Step Framework
### Step 1: Document Everything About Your Product
Start with a spreadsheet or Google Doc. Write down everything:
**Basic specs:**
- Dimensions, weight, materials
- Colors/options available
- Price
- Shipping/delivery time
- Warranty/guarantee
**Functional specs:**
- What it does (be specific)
- How it works (mechanism, not emotional language)
- Who it's for (be narrow, not broad)
- Who it's *not* for
- Performance metrics (speed, accuracy, durability, cost savings)
**Comparative:**
- How you differ from nearest competitor
- What you do that they don't
- What they do that you don't (be honest)
- Price comparison
**Evidence:**
- Test results, benchmarks
- Customer data (if you have it)
- Third-party validation
- Certifications
- Case studies with numbers
**Example (running shoes):**
Don't write: *"Our shoes are built for runners who want comfort and performance."*
Write:
```
Target: Overpronators with mild-to-moderate arch collapse
Design: Dual-density midsole (Shore A 65 core, 50 sidewalls)
Materials: Engineered EVA foam + carbon fiber reinforced nylon plate
Weight: 8.7 oz (size 10 US men's)
Testing: 200-runner trial, 8-week protocol
Results: 83% reported reduced impact pain vs. baseline
Frequency response: Tested for impact absorption at 3 m/s deceleration
Price: $189
Available sizes: 5–15 US (women's), 7–16 US (men's)
Durability: Tested for 300 miles before 10% material degradation
Warranty: 1 year manufacturer defect. 30-day wear trial guarantee.
```
The second version gives an AI everything it needs to understand, differentiate, and match to customer intent.
---
### Step 2: Get Opinionated
Your truth layer needs opinions.
Not emotional opinions ("we love running"). Strategic opinions.
**Examples:**
*"We don't believe in endless customization. Fewer options means better default experience. We ship with 7 core settings, not 50."*
*"We only serve teams under 50 people. Enterprise features dilute our product philosophy."*
*"We charge per company, not per user. Easier budgeting, no surprises."*
These aren't nice-to-have. They're essential to your truth layer because:
1. They help AI agents understand who you're for
2. They differentiate you (not everyone has these opinions)
3. They survive compression
**What not to do:**
- "We believe in quality." (Every company says this)
- "We're customer-first." (Generic)
- "We use cutting-edge AI." (Means nothing without specifics)
**What to do:**
- "We use [specific approach] because [specific constraint/reason]. This means [specific tradeoff]."
---
### Step 3: Structure Your Data
Now that you have specific, opinionated information, structure it so an AI can read it.
**Three formats:**
**Format 1: FAQ Pages**
```
Q: Is this shoe for overpronators?
A: Yes, specifically mild-to-moderate. If your arch collapses more than 15° on the pronation scale, we'd recommend our stability-plus model instead.
Q: How much weight will it save vs. competitor shoes?
A: Our shoe: 8.7 oz. Competitor A: 9.2 oz. Competitor B: 8.4 oz. We optimize for stability over absolute weight.
```
**Format 2: Structured Lists**
```
Shoe Specifications:
- Midsole material: Dual-density EVA (core Shore A 65±2, sidewalls 50±2)
- Plate: Carbon fiber reinforced nylon
- Weight: 8.7 oz (size 10 US men's)
- Price: $189
- Ideal for: Overpronators (arch collapse 8–15°)
- Testing data: [link to testing methodology]
```
**Format 3: JSON-LD Schema Markup (for your website)**
```json
{
"@context": "https://schema.org/",
"@type": "Product",
"name": "ProStability Running Shoe",
"description": "Dual-density midsole stability shoe designed for mild-to-moderate overpronators",
"brand": "YourBrand",
"offers": {
"@type": "Offer",
"price": "189.00",
"priceCurrency": "USD"
},
"aggregateRating": {
"@type": "AggregateRating",
"ratingValue": "4.6",
"ratingCount": "287"
}
}
```
**Use all three.** They serve different purposes:
- FAQ = humans + agents can read
- Structured lists = clear, easy to scan
- JSON-LD = machine-readable (helps Google and agents parse data)
---
### Step 4: Test for "Compression Survival"
Here's the acid test: Can your positioning survive being compressed into one sentence?
**Compression test:**
If an AI had to summarize you in one sentence for a customer who asked "What's a good running shoe for overpronators?", what would it say?
**Bad (doesn't survive compression):**
"Premium running shoes with advanced cushioning."
(This could describe 100 brands. Not differentiated.)
**Good (survives compression):**
"Stability shoe with dual-density midsole designed for mild-to-moderate overpronators. 83% pain reduction in testing. $189."
(Specific enough that an AI understands exactly what this is and who it's for.)
**How to test:**
1. Go to ChatGPT or Claude
2. Ask: "What's a good [your category] for [customer intent]?"
3. See if you show up
4. If not, your positioning isn't compressed-resistant
---
### Step 5: Implement & Monitor
**On your website:**
- Add schema markup to product pages (use Yoast, RankMath, or manual JSON-LD)
- Rewrite product descriptions using specificity (not emotions)
- Add FAQ pages with agent-readable Q&A
- Create spec sheets with structured data
**Monitor:**
- Re-test compression survival every 30 days
- Track whether AI agents recommend you
- Check your ARI score monthly
- Refine based on what customers actually ask about
---
## Common Mistakes
**Mistake 1: Still Using Emotional Language**
❌ "Premium quality with exceptional performance"
✅ "Tested on 200 overpronators. 83% reported reduced pain vs. baseline."
**Mistake 2: Being Too Broad**
❌ "For anyone who runs"
✅ "For mild-to-moderate overpronators who prioritize stability over weight"
**Mistake 3: Hiding Specs in Images**
❌ Product image with tiny text showing specs
✅ Specs on the page itself, readable by machines and humans
**Mistake 4: Making Claims Without Evidence**
❌ "Most accurate on the market"
✅ "94% accuracy on email classification. Tested on 10K emails. Compared against [competitors]."
**Mistake 5: Not Taking Opinions**
❌ "We try to serve everyone"
✅ "We serve teams under 50 people. We don't scale to enterprise."
---
## Quick-Win Implementation
Don't rebuild everything at once. Start with one product page.
**Week 1:**
- Document exact specs
- Identify your core opinion (who is this for/not for?)
- Rewrite the product description using specificity
**Week 2:**
- Add FAQ schema to the page
- Add JSON-LD Product schema
- Test compression survival
**Week 3:**
- Get feedback from customers (does this match how they understand the product?)
- Refine copy based on feedback
- Move to next product page
---
## Examples by Category
### SaaS
❌ "Intuitive software that saves time"
✅ "Email parsing tool using Claude API. 94% accuracy. Reduces triage by 60%. $29/month, no per-user fees."
### E-Commerce
❌ "High-quality products"
✅ "Organic cotton, sustainably dyed, made in Portugal. Certified B-Corp. Price: $47."
### Services
❌ "Expert consulting with proven results"
✅ "Specializes in companies doing $2–10M ARR. Average client sees 30% growth in first 12 months. $5,000 engagement minimum."
---
## Next Steps
1. **Pick one product page** to build the truth layer for
2. **Follow the 5 steps** above
3. **Test compression survival** (ask an AI if it would recommend you)
4. **Refine based on results**
5. **Scale to other products**
Want help? [Get the ARI report] to see exactly where your truth layer is weak.
Or [book a Truth Layer Audit] if you want us to do it.
---
**The bottom line:** An AI agent doesn't have time for your tagline. It needs specific, provable, opinionated data. That's your truth layer. That's how you survive in the interpretation economy.