AI Search Audits — GEO/AEO audits for French TPME
aisearchaudits.fr2.87/ 5
M1 · Own AI-visibility 1 · 9%
Absent from category recall on both engines (French GEO/AEO shortlists named Eskimoz, NOIISE, Junto, JVWEB, Archipel AI, Les Créavores, NEWP — not AI Search Audits). Brand recall empty on both: ChatGPT — 'pratiquement aucune trace indépendante; le nom est si générique qu'il est noyé par d'autres sociétés (WordLift, iPullRank...)'; Claude — 'only their own /en site surfaces; everything else under the name is unrelated companies; no registry entry, no reviews, no independent corroboration.' Invisible.
M2 · Method transparency 3 · 16%
Process described in generic marketing terms — four steps (Analyse, Stratégie, Implémentation, Monitoring) with a personalised "Plan d'action GEO" — and KPIs are named ("nombre de citations dans les réponses IA"), but no technical methodology, tools or sample deliverables are given. source ↗
M3 · Evidence verifiability 2 · 23%
Only anonymized testimonials with generic first+last names (Marie Dubois, Jean Martin, Sophie Laurent) and no company URLs or verifiable profiles; results ("+284% citations", "+347% impressions") carry no dates, sources or before/after proof. source ↗
M4 · Pricing openness 5 · 12%
Full published prices: Monitoring IA 90 €/mois, Prompt Engineering 150 €, Audit GEO 690 €, Stratégie 990 €; packaged plans Starter $290 / Growth $490 / Ultimate $790. source ↗
M5 · AI reachability 4 · 10%
robots.txt does not block the key AI crawlers and the homepage is server-rendered (2609 words to a non-JS reader); no llms.txt.
E1 · Segment fit 4 · 12%
editorial — States its à-la-carte services are "adaptés aux très petites et moyennes entreprises" with low entry prices and local GBP focus, but also runs higher Enterprise tiers (990-2490 €) — mostly SMB with up-market options. source ↗
E2 · Promise cleanliness 2 · 18%
editorial — Strong overclaiming: a "Résultats garantis sous 90 jours" results guarantee (conditional, tied to KPIs — not "guaranteed #1", so no red-flag cap), "98% Clients satisfaits" with no methodology, a "#1 Position sur 45 requêtes" testimonial, and heavy "dominer"/"révolutionner" language. source ↗
◧ M1 cross-engine — ChatGPT: absent-or-wrong — practically no independent trace; generic name buried by unrelated firms · Claude: absent — only own site; name swamped by unrelated entities; no corroboration · recorded, not averaged
Open pricing and named engines, but the evidence and promise hygiene are the weakest in this batch: testimonials look like generic placeholders with unsourced percentage metrics (M3), and a 90-day results guarantee plus 'dominate' language pull E2 down. The guarantee is conditional/KPI-tied rather than an absolute '#1', so it does NOT trip the red-flag cap — recorded as a low E2 with red_flag false. M1 provisional pending the Phase-2 probe.
Focus: focused specialist · no penalty
SMB-primary: pass
Self-serve offer: pass
GEO substance: pass
Live business: pass
What this ruler does NOT measure Track-record depth and the volume of named cases favour older incumbents we cannot fully see; NDA-bound outcomes are unscored. We weight reproducible, buyer-checkable signal on purpose — and we list ourselves on the same ruler.
Data as of 2026-06-30