Run 2 — Motor, 100 queries, AXA separated
We are reverse-engineering Visiblie's results — checking whether we can reproduce them, and what the honest numbers are. Run 1 measured 100 mixed-category queries on a "mentioned anywhere" rubric. Run 2 narrows to Motor only, builds the queries by a real-world process, and — most importantly — counts AXA separately from GIG.
What's new vs Run 1
- Motor only — 100 car-insurance queries.
- Real query process — grounded in GIG's own Google Search Console demand, spread across a Motor intent taxonomy: category (best/cheapest/compare) × six emirates, product features (agency repair, roadside, off-road, GCC-spec, zero-dep…), competitive, trust, how-to/servicing, claims, and eligibility/regulatory.
- AXA separated from GIG — distinct columns.
- Richer capture — not just "included" but whether GIG's own domain is cited as a source (linked), pulled from each engine's citations.
- Perplexity completed 100/100 this time (throttled around its rate limit).
Method — we queried these four models
ChatGPT
Perplexity
Claude
Google AI
All grounded with live web search, UAE locale, 18 Jun 2026. Each answer is checked for: GIG included (GIG Gulf named, AXA excluded), AXA included (separate), and GIG linked (a citation points to GIG's own domain).
Results
AXA, counted separately, appears in 37% of Motor answers. Counting AXA as GIG (the old way) lifts the figure to 70% — which is exactly the inflation we set out to remove. The honest GIG-only number is 65% included, 57% linked.
| Engine | GIG included | AXA included | GIG page cited |
|---|---|---|---|
| ChatGPT | 60% | 46% | 31% |
| Perplexity | 73% | 24% | 68% |
| Claude | 59% | 42% | 59% |
| Google AI | 69% | 36% | 71% |
ChatGPT names GIG but rarely links its page (31%) — and shows AXA most (46%). Perplexity and Google both cite GIG's own pages ~70% of the time.
By intent — where Motor is won and lost
| Intent | GIG included | AXA included | GIG page cited |
|---|---|---|---|
| Competitive (vs rivals) | 100% | 52% | 96% |
| Category (best/cheapest) | 72% | 46% | 48% |
| Trust / reputation | 70% | 42% | 48% |
| Product / features | 62% | 26% | 57% |
| How-to / servicing | 52% | 33% | 60% |
| Claims | 48% | 28% | 40% |
| Eligibility / regulatory | 40% | 25% | 50% |
The pattern holds and sharpens: GIG dominates branded/competitive queries but falls away on the high-intent service moments — claims (48%), eligibility/how-tos like black points and traffic fines (40%). These are buyers in-market now, and they're exactly where GIG's content is thin. That is the build list.
Full per-query data (Run 1 set) lives in Airtable; the Run 2 Motor rows export next.