GENERATIVE ENGINE OPTIMIZATION

CBI vs Pefindo vs CRIF

Share of voice, mention rank, dan query-level evidence
9ROUTER MODEL RUN COMPLETE
LLM Tested
3
ChatGPT · Claude · Gemini
Query Set
6
Indonesian credit bureau intent
Responses Analyzed
18
One response per LLM/query
Leading Brand, Avg Rank
Pefindo · #1.1
Across completed model runs

Share of Voice by LLM

SOV = queries mentioning brand ÷ all tested queries for selected LLM. Brand-specific comparison queries inflate mentions; use neutral query set for directional competitor benchmark.

Average Mention Rank

Rank derived from first order brand appears in response. Lower is better. This is response ordering, not Google ranking or verified market share.

Model Summary

LLMBrandMention SOVAvg RankCitations

Analysis

Query Mention Matrix

One row per query and LLM. Rank is extracted from answer order. “—” means brand not mentioned.
LLMQueryCBIPefindoCRIF
Data source: 9router batch run. Models: ChatGPT `cx/gpt-5.6-sol`, Claude `cc/claude-sonnet-5`, Gemini `ag/gemini-3.8-flash`. Citation counts only detect URLs returned in model output; models were not run with verified live web-search connectors. Refresh needs re-running query probe.