Methodology
How we measure AI visibility
Citerlly uses sampled buyer-intent prompts across selected AI engines. Results vary by model, prompt, and time — re-scans track movement, not guarantees.
What we scan
- Buyer-intent prompts tailored to the client's category and competitors
- Multiple AI engines on paid tiers (ChatGPT, Claude, Gemini, Perplexity); baseline scans use a subset
- Competitor mention share, cited sources, and per-prompt wins or losses
How each engine retrieves answers
Each AI engine pulls answers differently. Generic GEO playbooks ignore this. Citerlly maps your scan losses to engine-specific fixes — tied to the prompts and sources where you actually lose.
We do not promise rankings in any engine. We measure sampled buyer prompts, identify where each engine recommends competitors instead of you, and re-scan to prove movement.
Perplexity
Always searches the web. Every answer pulls 5–10 live sources ranked for direct answers — similar to Google, optimized for citations.
What we measure: Ship direct Q&A pages for lost prompts + get indexed on domains Perplexity already cites on those prompts.
Typical movement window: 1–2 weeks
ChatGPT
Training data up to a cutoff; GPT-4o can search Bing in real time when search mode is active. Citations appear only when search runs.
What we measure: Target lost prompts with owned pages on sites OpenAI crawls (your domain, Reddit threads, major blogs cited in scan excerpts).
Typical movement window: 2–4 weeks (search) · 6–12 months (training refresh)
Gemini
Deeply integrated with Google Search and Knowledge Graph. YouTube and Google Business Profile carry extra weight.
What we measure: Fix Google-visible gaps first: pages that rank for lost prompt keywords, YouTube if cited in scan, consistent entity signals.
Typical movement window: 1–3 months
Claude
Training cutoff by default; claude.ai can search when enabled. Heavy weight on authoritative, well-structured long-form content.
What we measure: Publish clear entity definition + methodology pages that answer lost prompts with evidence — not keyword fluff.
Typical movement window: 2–4 weeks (search) · 6–12 months (training refresh)
From generic advice to scan-backed plans
| Free GEO playbook | Citerlly deliverable |
|---|---|
| Publish original research and stats | Which lost prompts should cite your data — and which competitors get cited instead? |
| Structured Q&A content | Build one page per lost prompt with a direct answer block + FAQ schema. |
| High-authority backlinks | Which domains from source gaps must list you before engines recommend you? |
| Clear entity definition | Is brand + category + geography consistent on every page AI already cites? |
Scores and reports
The visibility score summarizes how often the brand appears or is recommended across the sampled prompt set. Mention rate, source gaps, and prompt-level diagnosis explain why competitors win. Delta reports compare the latest scan to the previous baseline.
What we do not claim
Based on sampled AI answers across selected engines. Visibility varies by model, prompt, location, and time. Re-scans track change over time; individual mentions are not guaranteed.
Re-scan cadence
Monitoring clients receive scheduled monthly re-scans. Each cycle produces an updated visibility report and a delta summary suitable for client retainers.