Background
The client is a B2B SaaS company offering a customer data platform (CDP). In late 2025, they discovered a fatal problem: when prospects asked ChatGPT "which CDP is best for mid-sized e-commerce", their mention rate was only 3%, while two competitors together captured 78% of the recommendation slots.
In other words: they had become invisible in the AI era.
Month 1: Diagnostic & Baseline
We didn't rush to write content. We did three things first:
- AI visibility audit: Asked 20 core questions on 4 major AI platforms (ChatGPT, Claude, Gemini, Perplexity) and tracked mention rate, sentiment, and rank position.
- Competitor reverse-engineering: Analyzed the two high-mention competitors' content structure, Schema deployment, and citation source strategy.
- Internal asset inventory: Catalogued the client's existing whitepapers, case studies, product docs, and technical blog.
Diagnosis: the client had content "volume" but severe "structure" deficiency. AI couldn't extract key points, didn't understand scenarios, and couldn't find citation sources.
Month 2: Semantic Asset Reconstruction
We restructured 47 blog posts and 12 whitepapers:
- Added TL;DR, Key Conclusions, and Applicable Scenarios sections to every article.
- Added comparison tables, data visualizations, and FAQ modules.
- Deployed Article + FAQPage Schema for every article.
- Unified the brand's "official definition" across all 5 platforms.
Month 3: Schema & Technical Foundation
Deployed full structured data:
{
"@context": "https://schema.org",
"@type": "SoftwareApplication",
"name": "Client Product",
"applicationCategory": "BusinessApplication",
"offers": { "@type": "Offer", "price": "0", "priceCurrency": "USD" },
"aggregateRating": { "@type": "AggregateRating", "ratingValue": "4.8" }
}
Also fixed 200+ technical issues: missing canonical tags, mixed H-tags, missing image alts, broken breadcrumb structure.
Months 4-5: Citation Source Building
The hardest and most critical phase. We helped the client:
- Publish 8 in-depth case studies on Reddit, G2, and Product Hunt (all signed by real users).
- Invite 5 industry KOLs to publish reviews on X, LinkedIn, and YouTube.
- Get 3 industry media outlets to publish feature articles.
- Clean up the brand's Wikipedia entry and its citations.
"Citation sources are not ad slots — they are the bridge that earns AI's trust."
Month 6: Results
6 months later, on the same 20 questions, same 4 platforms:
| Metric | Start | End | Change |
|---|---|---|---|
| Mention rate | 3% | 52% | 17.3× |
| Average recommendation rank | #4.2 | #1.6 | +2.6 |
| Positive sentiment | 28% | 81% | +53pp |
| Monthly demos from AI referrals | 0 | 47 | New channel |
The more important number: CAC for each AI-referred demo was 62% lower than Google Ads.
Three Lessons
- Structure before volume: Make content "AI-readable" first, then worry about quantity.
- Citation sources are the moat: Competitors can't copy them, users don't see them, but AI quietly trusts them.
- Monitor & iterate is the norm: GEO is not a one-time project. It is long-term infrastructure, like SEO.
What This Means for You
If your brand's AI mention rate is below 20%, don't panic, but act now. The GEO window is much shorter than the SEO window — in 3-5 years, when every brand is "on board", first-mover advantage will decide market structure.
Book a free diagnostic → 5 minutes, your brand's AI visibility report.