The gap — what we say vs what Visiblie says

Visiblie pitched GIG a monthly retainer (US$3,500–9,500) built on an April 2026 pilot that measured GIG's AI visibility at 12.6% citation share. We rebuilt their measurement from scratch — five engines, AXA-aware, ours to own — and our rigorous Run 3 reports GIG included in 51% of Motor answers. Four-fold apart. This page is the honest reconciliation: most of the gap is not a disagreement about reality, it is two different metrics wearing the same word.

Side by side

The metric trap — 51% is not "4× better than 12.6%"

Visiblie's citation share is a share-of-voice figure: of every brand citation across all answers, what fraction is GIG. Our 51% is an inclusion rate: the share of answers GIG appears in at all. These only equal each other when an answer names a single brand — and AI answers name several.

The rough reconciliation is the important bit. A typical Motor answer names ~5 brands. GIG present in 51% of answers, divided across ~5 brands per answer, is a share of citations of roughly:

So on the evidence we have, the two headline numbers may agree, not conflict. We cannot stand in front of GIG and say Visiblie's 12.6% is wrong — it is most likely a correctly-measured share of a different denominator. The credibility play is that we can decompose theirs and they cannot decompose ours.

Like-for-like — where there is a real gap

The fair comparison is Visiblie's extraction rate (a presence rate, like ours) against our by-intent rate — not their blended share. Visiblie reported extraction by funnel phase; we report it by Motor intent:

This is the divergence that is not a metric artefact. On non-branded category and attribute queries — the ones Visiblie built their whole "growth is upstream" thesis on — they see GIG at 1.6% and 11.3%; we see 56% and 42%. That is a genuine, large, like-for-like gap that has to be explained, not waved away.

What drives the real gap

Four causes, in likely order of weight:

1. AXA attribution. We credit the legacy equity (AXA appears in 46% of Motor answers — and 94% on Claude); Visiblie string-matched "GIG Gulf" and missed it. GIG Gulf is the rebranded AXA Gulf, so their number structurally undercounts.
2. Query construction. Their P2 is deliberately hard, cold non-branded "category formation." Our Motor category set is grounded in GIG's real Search Console demand — warmer, closer to how UAE buyers actually ask.
3. Timing. April vs June. Answer engines re-ground weekly; two months is real movement, especially as Google AI / Gemini surfaces matured.
4. Scope. Their 12.6% blends every line of business and 5 markets; ours is Motor-only, UAE — the line and market where GIG is strongest. Not wrong, just not the same slice.

What only our tool does

And it is ours to read — ~250 lines of open Python, no platform to rent. The point is not that our score is bigger; it is that we can take any number apart and show how it was made.

The questions Visiblie has to answer

Their document never defines the metric it leads with. Before either number can be trusted against the other, three things:

• Is "citation share" a mention or a source link? Our 51% and 43% map to their single 12.6% only by guess until they say which.
• What is the denominator? Brands per answer, answers per prompt, prompts per market — the share is meaningless without it.
• How is AXA handled? If they exclude it, their baseline understates GIG's true position — and the whole "P2 at 1.6%" alarm may be partly an attribution miss.