How Rankings Work
A step-by-step explanation of how a brand moves from AI response to a ranked position on this platform.
Step 1 — Prompt generation
For each query segment (e.g. "home healthcare dubai"), 18 intent-based prompts are generated spanning awareness, comparison, and decision intent. Prompts use real buyer language — no keyword manipulation.
Step 2 — Multi-engine execution
All 18 prompts are run across ChatGPT, Claude, Gemini and Perplexity within the same analysis window. Each response is parsed to extract brand mentions, position, and cited sources.
Step 3 — Appearance rate calculation
Formula
appearance_rate = prompts_brand_appeared_in ÷ total_prompts_run
Example: if Vestacare appeared in 15 of 18 prompts → 83% appearance rate.
Step 4 — Evidence scoring
| Signal | Points |
|---|---|
| Citation frequency data present | +2 |
| Named authority-source cited | +2 |
| Brand appears in 3+ distinct prompts | +2 |
| Structured entity signals detected | +1 |
Maximum possible score: 7. Publication threshold: ≥ 3.
Step 5 — Quality gate
A query page only publishes when both conditions are met:
Condition 1 — Evidence score ≥ 3
At least one strong signal (authority source or repeated appearance) must be present.
Condition 2 — Brand count ≥ 3
At least 3 distinct brands must appear in the results. Single-brand pages are not published.
Step 6 — Slug locking
Once a page is published, its URL (/best-home-healthcare-dubai) is locked permanently. If normalisation rules change in future, existing published pages retain their original URL to preserve all external citations and inbound links.
Step 7 — Versioning
When rankings are updated (new analysis cohort), a new version row is created. The page displays the latest version. Historical data is retained — older rankings are never deleted.
Ranking tiers
| Appearance rate | Label |
|---|---|
| 75% + | High Visibility |
| 50% – 74% | Moderate |
| Below 50% | Low Visibility |