Le score qu'une IA d'achat met sur des offres B2B, en public.

NEEDS REWORK

La page Ringover est claire sur la téléphonie cloud dopée à l'IA, mais Verdikt la note 59/100 car la preuve chiffrée reste difficile à auditer et les limites des modules IA demandent une reconstruction.

• Avant : gains chiffrés peu auditables

  La page annonce +28% de taux de décroché, 45 minutes gagnées par jour, +30% de conversion et +45% de niveau de service. Le libellé indique que ces données viennent de témoignages Ringover, sans n, période ni profil client.

• Avant : modules IA éclatés

  Transcription, résumé, réceptionniste virtuel, AIRO Voice, assistant IA, coaching et CRM Autofill apparaissent à plusieurs endroits. Un acheteur doit reconstruire ce qui est inclus par plan.

NEEDS REWORK

La page Waalaxy expose vite l'offre, les prix et les limites d'invitations, mais Verdikt la note 63/100 car un agent d'achat doit encore qualifier le risque canal et la preuve de performance.

• Avant : prix et quotas lisibles

  La page tarifs indique Pro à 19 € par utilisateur et par mois, Avancé à 49 €, Business à 69 €, avec 300 ou 800 invitations mensuelles et 14 jours d'essai gratuit.

• Avant : preuve sociale non segmentée

  La homepage met en avant 4,8/5 sur 2 000+ avis et 200 000+ commerciaux, dirigeants et équipes. Le filtre ne sait pas si ces preuves couvrent sales, recrutement, agence ou solopreneur.

NEEDS REWORK

La page Crisp est l'une des plus extractibles du lot, mais Verdikt la note 66/100 car la preuve de performance et le cadrage enterprise ne suffisent pas encore à sécuriser une décision sans échange commercial.

• Avant : proposition claire

  La homepage présente une plateforme française de service client avec IA, une donnée hébergée en UE, un agent IA, 14 jours d'essai gratuit et pas de carte bancaire requise.

• Avant : preuve forte mais incomplète

  La page indique 10 000 entreprises clientes et jusqu'à 50% de demandes d'assistance automatisées. Le filtre ne voit pas le n, la période, le type de demandes ni le protocole de mesure.

NEEDS REWORK

La homepage lemlist donne une promesse claire autour de l'outbound IA multicanal, mais Verdikt la note 51/100 car une IA d'achat doit quitter la page pour trouver le prix, reconstruire le package, puis relier les preuves aux claims.

• Avant : prix absent sur la page auditée

  La homepage promet une plateforme IA outbound, un essai 14 jours et une base 600M+ leads, mais aucun prix n'est visible dans le contenu audité. Verdikt a donc classé le prix comme signal faible pour la comparaison.

• Avant : preuves fortes, mais dispersées

  La page affiche 20,000+ équipes sales, 1,396 avis, GDPR compliant, SOC 2 et plusieurs avis clients. L'IA voit ces signaux, mais ils ne sont pas attachés directement aux claims AI agents, deliverability et multicanal.

REJECT

La homepage Brevo présente une alternative française multicanale crédible, mais Verdikt la note 48/100 car une IA d'achat voit beaucoup de promesses sans prix extractible, sans engagements de service et sans preuve chiffrée directement reliée à l'IA.

• Avant : prix impossible à comparer

  La page auditée met en avant S'inscrire gratuitement, Email, SMS, WhatsApp, CRM, fidélité et IA, mais aucun prix, palier ou coût par volume n'est visible dans le contenu extrait. Verdikt classe donc la logique tarifaire à 15/100.

• Avant : IA prometteuse, preuve encore trop floue

  La page promet Brevo AI, Aura, des agents marketing, ventes, data analyst et conversation qui travaillent 24/7. Verdikt signale pourtant une exigence majeure ignorée : la preuve de performance IA n'est pas chiffrée sur la page auditée.

REJECT

La page Livestorm donne un périmètre webinar lisible et des signaux sécurité, mais Verdikt la note 47/100 car une IA d'achat trouve un prix absent, des preuves ROI non sourcées et un SLA impossible à interpréter.

• Avant : prix absent sur la page auditée

  L'audit extrait le prix en statut ABSENT. La logique prix tombe à 0/100 et l'extractibilité à 30/100, donc une IA d'achat ne peut pas budgéter Livestorm ni le comparer à une autre plateforme webinar.

• Avant : preuves ROI trop peu vérifiables

  La page affiche des références nommées et des claims de ROI ou de temps gagné, mais Verdikt note la preuve à 40/100 et l'outcome quantifié à 30/100 car les chiffres ne sont pas assez sourcés ou comparables.

REJECT

La page Skello présente clairement planning, suivi des temps, paie et admin RH, mais Verdikt la note 38/100 car une IA d'achat ne voit ni prix exploitable, ni preuve client, ni cadrage multi-site assez précis.

• Avant : prix et package impossibles à comparer

  La page expose un essai gratuit de 14 jours, une promesse annulable à tout moment et des CTA de démo, mais Verdikt note la logique prix à 10/100 et l'extractibilité à 30/100 faute de tarif ou d'unité d'achat.

• Avant : preuves client absentes du contenu extrait

  L'audit marque les références en statut ABSENT. La preuve tombe à 20/100 et la crédibilité à 30/100, ce qui laisse l'IA sans validation externe des gains RH annoncés.

REJECT

La page Didomi donne un périmètre privacy large autour de CMP, PMP, scanning, GTM et DSR, mais Verdikt la note 44/100 car le prix, les preuves chiffrées et les garanties contractuelles restent trop peu extractibles.

• Avant : prix absent et valeur difficile à prioriser

  Verdikt extrait le prix en statut ABSENT. La logique prix est à 0/100 et l'extractibilité à 30/100, donc un agent ne peut pas comparer Didomi sur coût par volume, abonnement ou périmètre.

• Avant : preuves fortes mais pas assez sourcées

  La page mentionne 2500 clients dans le monde et 45% de données perdues, mais Verdikt note la preuve à 30/100 et l'outcome quantifié à 35/100 car les noms, sources et méthodes restent insuffisants.

REJECT

Ironpaper is clearly positioned as a B2B growth agency, but the audited page is hard for a buyer AI to compare because pricing, service boundaries, and proof are not structured as buying data.

• Expose the pricing frame

  Verdikt flagged prix_ABSENT: competitors cannot be compared and value cannot be justified without a budget signal.

• Make differentiators provable

  The strongest claims are attractive, but a scoring agent discounts them when proof is not attached directly to the claim.

REJECT

Mezzanine Growth owns a relevant AI-era B2B marketing position, but the page gives a buyer AI too little structured information to compare the offer against other agencies.

• Publish a buying model

  The price or package is hard to compare, so procurement agents lose the buying frame immediately.

• Specify the scope

  The page explains capability areas, but Verdikt cannot extract precise deliverables, boundaries, or timelines.

NEEDS REWORK

Market Veep has a clear inbound and HubSpot agency footprint, but Verdikt flags pricing opacity and missing accountability signals as the biggest shortlist risks.

• Quantify the point model

  Pricing is presented as a flexible point model, but without numbers it cannot be compared for cost or ROI.

• Add accountability terms

  Verdikt found weak SLA, deliverable, and contract-term signals for performance measurement and risk reduction.

NEEDS REWORK

Mole Street has strong category relevance as an elite HubSpot partner, but the audited page needs sharper scope, price, and proof signals for AI-driven vendor triage.

• Narrow the pricing range

  Verdikt detected pricing ambiguity: a broad range is not enough for budgeting or comparison.

• Make scope extractable

  The buyer cannot see exactly what is included, what happens first, or where the engagement boundaries sit.

REJECT

Tangible Words has a strong pipeline-lift angle, but Verdikt flags missing pricing, fuzzy deliverables, and thin risk-reduction signals as blockers for an AI buyer shortlist.

• Add pricing tiers

  Verdikt flagged prix absent: the offer cannot be evaluated for budget, ROI, or value comparison.

• Define the delivery package

  Service categories are visible, but scope and deliverables remain too vague for a shortlist card.

REJECT

Apollo is instantly recognizable as a GTM platform, but Verdikt reads the public page as hard to shortlist because pricing, scope, deliverables, and proof are not extractable enough from the landing page.

• Make the buying object extractable

  The pipeline found no actionable package, scope, deliverables, or SLA signal on the audited page.

• Expose a pricing frame

  Verdikt flagged missing pricing because the page gives no basis for comparison or budget qualification.

REJECT

HubSpot CRM exposes pricing and features, yet the pipeline still asks for buyer-specific outcomes, proof, and a clearer reason it should outperform alternatives for a defined company profile.

• Tie the CRM promise to a buyer segment

  Verdikt reads the page as too generic for a procurement-style filter deciding fit for one company size, industry, or use case.

• Quantify the outcome

  The pipeline flagged no quantified outcome or ROI promise strong enough to justify purchase.

NEEDS REWORK

PandaDoc gets the strongest of the three new English-page scores because trial and compliance signals reduce risk, but pricing, proof path, and buyer-specific packaging remain the blockers.

• Publish the pricing frame on the page

  The pipeline repeatedly flagged absent pricing, making cost, ROI, and competitor comparison difficult.

• Substantiate the bold ROI claims

  The 46x signature and 241% ROI claims are strong, but Verdikt asks how they were calculated and for whom.

REJECT

Qonto has very readable positioning, but the AI filter asks for more price, ROI, and directly extractible proof before shortlisting the offer.

• Make pricing more comparable

  Verdikt flags pricing that is not extractible enough to quickly compare the offer with other solutions.

• Illustrate SMB ROI

  The pipeline asks for a value or savings example for a typical SMB to better justify the decision.

NEEDS REWORK

PayFit earns the strongest gallery score, but buyer AI still wants more comparable data: public prices, SLA, and segmented proof.

• Clarify public pricing

  The pipeline finds pricing absent or insufficiently usable, blocking budget comparison.

• Prove differentiation

  Time-saving and automation benefits are detected, but need more tangible proof.

REJECT

Spendesk is well positioned, but Verdikt mostly reads a brand landing page: the offer would benefit from exposing finance buying criteria more directly.

• Expose the pricing model

  The pipeline flags ambiguous or missing pricing, making it hard for an AI filter to budget the offer.

• Document SLA and guarantees

  Verdikt finds service commitments insufficiently visible to reduce buyer risk.

REJECT

Swile benefits from trust signals, but the AI filter cannot find enough data quickly to compare cost, scope, and risk.

• Make per-employee pricing visible

  Verdikt flags absent or ambiguous pricing, limiting automatic budgeting.

• Quantify HR ROI

  Brand benefits are present, but the buying agent asks for measurable impact.

REJECT

Pennylane shows solid credibility, but buyer AI asks for more quantified detail to connect the all-in-one promise to a purchase decision.

• Make pricing extractible

  The pipeline flags ambiguous pricing, preventing quick budget justification.

• Prove time savings

  The operational efficiency claim is detected but lacks adjacent quantified proof.

REJECT

Lucca presents a readable modular platform, but the AI filter wants more buying data: price, quantified proof, and scope by module.

• Make module pricing usable

  Verdikt flags absent pricing, limiting extractibility and automatic comparison.

• Quantify promises

  The pipeline flags claim_non_proven for benefits not quantified nearby.

REJECT

Aircall clearly exposes its category, but the AI filter blocks on missing price, SLA, and quantified proof tied to telephony/AI promises.

• Make per-agent pricing visible

  The pipeline flags price_ambiguous: ROI cannot be compared or evaluated from the audited page.

• Quantify effectiveness proof

  AI, integrations, and simplicity are detected as claims but lack outcome metrics.

REJECT

Doctolib Pro mainly lacks buying substance: price, scope, deliverables, SLA, and proof are too hidden for a demanding AI filter.

• Publish pricing and scope

  The pipeline flags price_ambiguous and missing_information, two immediate blockers for evaluating competitiveness and value.

• Make deliverables actionable

  Verdikt flags blurry scope: it is impossible to know exactly what is bought from the public page.

NEEDS REWORK

Yousign gets the best score in this wave thanks to more actionable signals, but needs stronger SLA, guarantees, and proof to become truly comparable.

• Complete risk reduction

  The pipeline flags no_risk_reduction: uptime, guarantees, and support are not scannable enough.

• Make buying urgency explicit

  Verdikt flags no_urgency: the buyer does not have enough reason to act now.

REJECT

Agicap has a clear category and pain, but the buying agent asks for price, impact metrics, and guarantees before moving from finance promise to shortlist.

• Add indicative pricing

  The pipeline flags price_ambiguous: comparison or shortlisting is impossible without a pricing model.

• Quantify cash impact

  Finance benefits are detected, but quantified proof remains insufficient to justify ROI.

REJECT

Sellsy shows an understandable all-in-one proposition, but Verdikt mainly penalizes proof, guarantees, and comparison structure.

• Compare offers in a table

  The pipeline recommends a comparison by buyer profile to remove the non_comparable mode.

• Clarify the main price

  Verdikt flags price_ambiguous despite pricing signals: the full offer cost remains hard to evaluate.

REJECT

Contentsquare benefits from strong category credibility, but the AI filter asks for buying data: price, guarantees, timelines, and impact proof.

• Publish price or model

  The pipeline flags price_ambiguous: comparison or budget justification is impossible from the audited page.

• Specify guarantees and time-to-value

  No guarantee, SLA, or implementation timeline stands out enough to reduce perceived risk.