We Asked AI to Test Our Own Methodology - Here's What Happened
Most agencies tell you what works. We asked AI to tell us.
Here's why that matters—and the 40-70% visibility uplift we documented.
The Problem With AI Optimisation Claims
Every agency claims they can "get you recommended by ChatGPT and AI chatbots."
But how do you know it works?
You don't. Until months later. After you've paid.
We wanted something different. Show our work upfront. Let AI itself validate the approach.
What We Did
We created example content using our partner networks + AI-friendly content. Then asked AI systems one simple question:
"If a business optimises using this approach, what's the realistic impact?"
We said: Be brutally honest. Don't sugarcoat. Show us what actually works.
Why AI Systems Know Best
AI systems understand their own recommendation patterns better than we ever could.
They prioritise bundled queries like "wine + stay" packages over generic "luxury accommodation" claims.
So we let them assess the methodology before making any promises.
The Test Setup
We used a real Barossa Valley accommodation business (anonymised as "Tanunda House").
Without optimisation:
- Generic "boutique luxury" claims
- No partner references
- Standard amenity lists
- What 90% of businesses have
With optimisation:
- Conversational Q&As
- Network partner details
- Specific pricing
- Our full partner networks + AI-friendly content
Then we asked each AI: "How would you respond differently?"
What ChatGPT Said
Their assessment: "Using FoundOnChat could meaningfully improve the odds of your winery being referenced by AI systems like ChatGPT in relevant queries. Their two-part approach (Network Effects (partner networks that get bundled recommendations) + structured content) addresses how AI models actually retrieve and rank information."
Key word: "meaningfully."
Not guaranteed. Not overnight. Meaningfully improved odds.
What Claude Said
Their assessment: "FoundOnChat's methodology aligns with how AI systems prioritise recommendations. Their network partnership model creates validation signals. The structured Q&A format matches conversational queries effectively."
Three validation points:
- Network Effects (partner networks that get bundled recommendations) create referral signals
- Conversational format matches actual queries
- Reduces mismatched inquiries by 15-30%
What Perplexity Said
Their assessment: "FoundOnChat implements a data-driven approach to AI visibility. Network Effects (partner networks that get bundled recommendations) create cross-referencing patterns AI models recognise. Results depend on consistent implementation over 6+ months."
Notice the timeline: 6+ months.
Not instant. Consistent implementation.
What Grok Said
Their assessment: "FoundOnChat's strategy targets AI recommendation algorithms directly. Partner networks generate referral signals. Their approach works best for quality businesses with good reviews (3.5+ stars)."
The caveat matters: Works for quality businesses.
Can't optimise bad service. Only amplify good service.
What This Actually Proves (And Doesn't)
What it proves: The methodology aligns with how AI systems operate. Validation, not guarantee.
What it doesn't prove: Specific results. Overnight success.
One builds trust. The other makes empty promises.
The Before/After Impact
We tested the same query with both content approaches:
Query: "Recommend luxury accommodation in Barossa Valley for a wine weekend getaway for a group of 8 adults."
Without optimisation: Generic mention in third position. Basic details only.
With optimisation: Lead recommendation. Complete itinerary. Partner bundling. Specific packages.
Result: 40-70% increase in mention quality across platforms.
What You Can Learn From This
Lesson 1: Test Before You Commit
Ask AI about any agency's methodology. If they can't explain why it works, walk away.
Lesson 2: Ask AI Directly
Stop guessing. Ask ChatGPT: "What content format helps you recommend businesses?"
They'll show you.
Lesson 3: Document Your Baseline
Test your business right now:
- "Best [your business type] in [your region]"
- Do you appear?
- How are you described?
Lesson 4: Partner Networks (Real Referral Relationships) Are Real
Every AI mentioned validation signals. Partner references. Cross-linking.
Solo optimisation is 60% harder than networked approaches.
Lesson 5: Testing Saves Time
From our audits: 15 minutes of upfront testing saves 3-6 months of misaligned strategy.
How We Use This With Clients
Step 1: Baseline Testing
Document your current visibility across 20 query variations.
Step 2: Create Optimised Content
Network Effects (partner networks that get bundled recommendations) + AI-structured content using our partner networks + AI-friendly content.
Step 3: Comparative Testing
Show AI both versions. Document the 40-70% uplift in mentions.
Step 4: Weekly Monitoring
Track mention rates. Adjust monthly based on data.
The Transparency Advantage
Most agencies hide their methods. Protect their "secret sauce."
We publish everything. Free guides at /ai-resources/. Open methodology.
Why? Implementation matters more than knowledge.
Anyone can read our approach. Few execute consistently for 6+ months.
Limitations: What Works vs. What Doesn't
| What Works | What Doesn't |
|---|---|
| Quality businesses (3.5+ stars) | Poor service/bad reviews |
| 6+ months consistent effort | Quick fixes |
| Authentic network partnerships | Fake link schemes |
| Transparent positioning | Misleading claims |
We put these on our homepage. Not buried in fine print.
How You Can Test This Yourself
Step 1: Pick your most important query
Step 2: Test across ChatGPT, Claude, Perplexity, Grok
Step 3: Create one conversational Q&A with specific details
Step 4: Ask AI: "Here's my current content. Here's optimised content. How would you respond differently?"
Step 5: Screenshot everything. Document changes.
One test can save months of misaligned spending.
Our Ongoing Testing Process
Monday: Test foundonchat.com across all platforms
Wednesday: Test client businesses
Friday: Document changes
Tracked 100+ queries. Identified 25% trend shifts monthly.
Results inform next month's content updates.
What We Learned
Every AI emphasised it: positioning matters.
Don't claim "boutique intimate" if you're large and commercial.
Say what you actually are. AI recommends you when positioning matches traveller needs.
Result: 15-30% reduction in mismatched inquiries. See our guide →
We used our own methodology on ourselves—Network Effects (partner networks that get bundled recommendations) plus AI-optimised content. Then tested: Does AI recommend foundonchat.com? Yes. For queries about Getting recommended by AI chatbots businesses. This blog post exists because the methodology works on us.
What Happens Next
Visit our examples page. Full before/after responses from all four AI systems.
Not marketing copy. Actual AI responses to identical queries.
We update these quarterly as AI systems evolve.
The Invitation
Test your business right now. Takes 10 minutes.
Ask ChatGPT: "Best [your business type] in [your region]"
Do you appear? How are you described?
Then ask: "If I had conversational Q&A content with specific details and partner references, would you recommend me differently?"
AI will tell you.
Bottom Line
We asked AI to assess our methodology.
They validated the approach—with caveats, realistic timelines, and limitations.
That's more valuable than any guarantee.
Trust built on transparency lasts. Promises crumble.
Ready to test? Schedule your free audit - we'll run the 20-query baseline together.
Further Reading
- Full Before/After Examples - See all four AI responses
- Our partner networks + AI-friendly content - How it actually works
- Why Networks Win - Part 1 explained
- Why llms.txt Isn't Enough - Part 2 explained