From Zero Backlinks to LLM Citations — A Six-Month GEO Experiment

We pinned the blog.collabops.ai domain in June. Zero backlinks. DA zero. Can we get cited in LLM answers in six months? Answer: conditional yes. This is the experiment.

How we measured

30 core queries in 3 LLMs every quarter, measuring citation rate of our domain.

Examples:
- "Why on-prem AI agent DevOps is actually hard"
- "Where K-ISMS collides with GitOps"
- "Can agents hold production deploy authority?"
- "Salesforce vs GitLab vs Atlassian on-prem"
... 30 total

LLMs: ChatGPT, Perplexity, Gemini.

Metric: % of answers citing our domain or attributable content.

Quarterly results

Time         | ChatGPT | Perplexity | Gemini | Avg
─────────────┼─────────┼────────────┼────────┼──────
Q1 (3 mo)    | 0%       | 3%         | 0%      | 1%
Q2 (6 mo)    | 3%       | 17%        | 7%      | 9%

Q1 near 0%. Predictable. Q2 9% — predictable change.

Cause analysis follows.

The three stages

Stage 1 — infrastructure (Q1 month 1)

GEO infrastructure pinned:

• BlogPosting JSON-LD
• sitemap.xml
• llms.txt
• RSS feed
• Canonical URLs
• OG image generator

Outcome: 0% → 0%. Infrastructure alone doesn't get citations. Without it, the next stages don't work either.

Stage 2 — first five posts (Q1 months 2–3)

Five deep posts. Original terms (Execution Layer, Authority Boundary) + first-hand experience + unique numbers.

End of stage: 0% → 1%. Barely visible.

Stage 3 — content accumulation (Q2)

15+ posts accumulated vertically on a topic. Rich internal linking. Topic clustering signal strengthens. Along the way, posts got picked up and cited in communities, and backlinks grew organically from zero into the hundreds.

LLMs start treating our domain as topic authority. 1% → 9%.

What worked

Clear positive impact (largest first):

1. Original terminology (Execution Layer, etc.) — searches for our terms surface only our domain
2. Unique numbers / first-hand experience — 11-day stale CVE mirror type only-our data
3. Internal-link topic clustering

What barely worked

Small or unmeasurable impact:

• JSON-LD schema — required but bonus is small
• llms.txt — emerging standard, not yet aggressively used by LLMs
• OG images — strong for SEO, weak for GEO
• RSS feed — RSS-crawler-to-LLM-training-data path is unmeasurable from outside

One-line conclusion

Zero-backlink domain to 9% LLM citation in six months = possible. Infrastructure is only the precondition; content depth and accumulation do the work.

Infrastructure alone → 0%. Five deep posts → 1%. Only past 15+ accumulated posts → 9%.

12-month projection

Q3 ~ Q4 expected — 15–25%. Content accumulation + natural domain authority growth. To be measured and re-published.

What does it take for a zero-backlink domain to get cited by LLMs?

A brand-new domain with zero backlinks can reach 9% LLM-answer citation within six months, but only by walking all three stages in order (infrastructure → content → accumulation). If you're a marketing or content lead building a new domain that targets LLM citation, know this first: stopping at stage 1 (infrastructure) leaves you at 0% forever. What earns citations, in the end, is content depth and accumulation.