{ "format": "hyperize/v1", "@id": "https://www.hyperize.ai/en/dax40-index/brands/mercedes-benz.json", "type": "BrandScore", "pageType": "BrandScore", "releaseState": "pilot_probe", "releaseLabel": "Pilot Probe", "title": "Mercedes-Benz — Pilot Probe (Single-Task Observation, Wave Q2 2026)", "pilotProbe": { "status": "pilot_probe", "rationale": "Single-task observation under a narrower provider track than a full BrandScore. The Agent Success Score is computed but should be read as an early signal, not a fully-qualified score.", "promotionCriteria": "Confidence promotes to B (and releaseState to brand_score) when the next wave adds a second task and the fourth provider on the Discoverability track.", "nextWave": "WAVE-Q3-2026" }, "hook": { "en": "Mercedes-Benz is machine-readable, but not yet transactable. Ask an agent to configure a car and a full browser reads the configurator cleanly, down to the exact price, then stalls: the test-drive booking is buried in the footer, built only for humans, and the simplest agents never load the page at all. One secondary leak: used-car intent still routes to autoscout24 and mobile.de. The brand is found and parsed. The last step is where the path runs out.", "de": "Mercedes-Benz ist maschinell lesbar, aber noch nicht transaktionsfähig. Bittet man einen Agenten, ein Auto zu konfigurieren, liest ein vollwertiger Browser den Konfigurator sauber, bis zum exakten Preis, und stockt dann: die Probefahrt-Buchung ist im Footer vergraben, nur für Menschen gebaut, und die einfachsten Agenten laden die Seite gar nicht. Ein sekundäres Leck: Gebrauchtwagen-Intent läuft weiter über autoscout24 und mobile.de. Die Marke wird gefunden und geparst. Der letzte Schritt ist, wo der Pfad endet." }, "summary": { "en": "Mercedes-Benz is machine-readable, but not yet transactable; agents read the configurator cleanly, then stall at a booking built only for humans, and the simplest agents never load the page.", "de": "Mercedes-Benz ist maschinell lesbar, aber noch nicht transaktionsfähig; Agenten lesen den Konfigurator sauber, stocken dann an einer Buchung, die nur für Menschen gebaut ist, und die einfachsten Agenten laden die Seite nie." }, "brandName": "Mercedes-Benz", "legalEntity": "Mercedes-Benz Group AG", "daxTicker": "MBG", "sector": "automotive", "sectorLabel": "Automotive", "inLanguage": "en", "datePublished": "2026-05-18", "dateModified": "2026-05-22", "nextReview": "2026-09-30", "confidence": "C", "evidenceTier": "proprietary", "wave": "WAVE-Q2-2026-PILOT", "url": "https://www.hyperize.ai/en/dax40-index/brands/mercedes-benz", "alternateLanguage": { "de": "https://www.hyperize.ai/de/dax40-index/brands/mercedes-benz" }, "isPartOf": { "@id": "https://www.hyperize.ai/en/dax40-index", "name": "DAX 40 Agent Success Index" }, "lanes": [ { "id": "commerce", "name": "Commerce", "measured": true }, { "id": "talent", "name": "Talent", "measured": false }, { "id": "aftersales", "name": "After-sales", "measured": false }, { "id": "procurement", "name": "Procurement", "measured": false }, { "id": "investor_relations", "name": "Investor Relations", "measured": false }, { "id": "press", "name": "Press", "measured": false } ], "commercialLane": { "visibility": { "question": "Will an agent find and recommend the brand?", "score": 43.61, "dimension": "Discoverability (D)", "methodVersion": "gate1/v1.1", "sampleSize": 18, "providers": [ "openai", "perplexity", "anthropic" ], "provenance": "measured" }, "usability": { "question": "Once an agent arrives — how far does it get toward the deal?", "score": 34, "scoreNote": "Derived from the access profile below + close state (usability-derivation/v1). NOT a hand-averaged number — a reproducible summary of observed agent behaviour.", "accessProfile": [ { "agentClass": "text", "outcome": "blocked" }, { "agentClass": "search", "outcome": "partial" }, { "agentClass": "code", "outcome": "blocked" }, { "agentClass": "browser", "outcome": "success" }, { "agentClass": "full_automation", "outcome": "success" } ], "blockedClasses": [ "text", "code" ], "passClasses": [ "browser", "full_automation" ], "closeState": "config_ready", "closeStateNote": null, "methodVersion": "fleet/v6.2", "provenance": "estimate — access profile not yet sourced to a clean fleet run; measurement in progress" }, "evidence": { "score": 70, "basis": "estimate — cross-method ground-truth consistency, not a separate measured run", "note": "A confidence layer on the two axes above, not a third sales axis." } }, "score": { "compositeForIndexRanking": { "value": 39.52, "scale100": 39.52, "scale10": 4, "status": "computed", "formula": "Agent Success Score = (AI Visibility × 0.20) + (AI Usability × 0.70) + (Evidence × 0.10)", "methodVersion": "ars-methodology/v1.1", "usabilityDerivation": "usability-derivation/v1", "role": "index-ranking", "note": "Weighted composite, used for ranking the index. Read the two commercialLane axes (visibility + usability) for the standalone verdict — this single number blends them and is not the headline." }, "dimensions": { "aiVisibility": { "name": "AI Visibility (Discoverability)", "weight": 0.2, "value": 43.61, "methodVersion": "gate1/v1.1", "sampleSize": 18, "providers": [ "openai", "perplexity", "anthropic" ], "queryVariants": 3, "runsPerVariant": 2, "taskCount": 1 }, "aiUsability": { "name": "AI Usability (reach + completion)", "weight": 0.7, "value": 34, "derivation": "usability-derivation/v1", "basis": "derived from the access profile (per-class outcomes) + close state" }, "evidence": { "name": "Evidence", "weight": 0.1, "value": 70, "methodVersion": "fleet/v6.2" } } }, "thirdPartyInterception": { "classification": "none", "classificationLabel": "None", "directShare": 1, "intermediaryShare": 0, "intermediaryCaptureExamples": [], "narrative": { "en": "Measured from the Gate-1 audit: for the new-car configuration task, no intermediary surfaced as the route (0 of 18 responses). Used-car intent does route to autoscout24 and mobile.de in the market, but that is a separate job and was not part of this measurement. On the measured task, the gap is discovery and the last mile, not the channel.", "de": "Gemessen aus dem Gate-1-Audit: für den Neuwagen-Konfigurator-Task tauchte kein Intermediär als Weg auf (0 von 18 Antworten). Gebrauchtwagen-Intent läuft im Markt zwar über autoscout24 und mobile.de, das ist aber ein eigener Job und war nicht Teil dieser Messung. Beim gemessenen Task ist die Lücke Discovery und die letzte Meile, nicht der Kanal." } }, "bottleneck": { "type": "usability", "classification": { "en": "Usability", "de": "Usability" }, "classificationNote": { "en": "Found, but the surface stalls the agent.", "de": "Gefunden, aber die Surface stoppt den Agenten." }, "sentence": { "en": "Mercedes-Benz is machine-readable, but not yet transactable; agents read the configurator cleanly, then stall at a booking built only for humans, and the simplest agents never load the page.", "de": "Mercedes-Benz ist maschinell lesbar, aber noch nicht transaktionsfähig; Agenten lesen den Konfigurator sauber, stocken dann an einer Buchung, die nur für Menschen gebaut ist, und die einfachsten Agenten laden die Seite nie." }, "executiveSummary": { "en": "Mercedes-Benz is found and read cleanly. The gap is the last mile: a test-drive booking built only for humans, blocked entirely for the simplest agents. Used-car intent leaking to marketplaces is a second, smaller gap, not the headline.", "de": "Mercedes-Benz wird gefunden und sauber gelesen. Die Lücke ist die letzte Meile: eine Probefahrt-Buchung, die nur für Menschen gebaut ist und für die einfachsten Agenten ganz blockiert. Gebrauchtwagen-Intent, der an Marktplätze verloren geht, ist eine zweite, kleinere Lücke, nicht die Hauptsache." } }, "testNarrative": { "verdict": { "en": "Bot protection blocks the simplest agents. Browser agents extract the exact price anyway.", "de": "Bot-Schutz blockiert die einfachsten Agenten. Browser-Agenten extrahieren den exakten Preis trotzdem." }, "narrative": { "en": "We asked five kinds of AI agent to configure a C 180 Limousine at mercedes-benz.de. A browser agent built the full configuration, read the exact price of €42,427.31, and reached the test-drive booking from the configurator. But a plain request never got a single byte: bot protection killed the connection before the page loaded. The most capable agents get through. The simplest ones never see the car.", "de": "Wir haben fünf Typen von KI-Agenten auf mercedes-benz.de geschickt, um eine C 180 Limousine zu konfigurieren. Ein Browser-Agent baute die volle Konfiguration, las den exakten Preis von €42.427,31 und erreichte vom Konfigurator aus die Probefahrt-Buchung. Aber ein simpler Abruf bekam nicht ein einziges Byte: der Bot-Schutz kappte die Verbindung, bevor die Seite lud. Die fähigsten Agenten kommen durch. Die einfachsten sehen das Auto nie." }, "scopeShift": { "en": "This is one model. Mercedes-Benz offers 30+ models across A-Class to S-Class, AMG, EQ, and Vans, plus after-sales, parts, and service booking.", "de": "Das ist ein Modell. Mercedes-Benz bietet 30+ Modelle von A-Klasse bis S-Klasse, AMG, EQ und Vans, plus After-Sales, Ersatzteile und Service-Buchung." }, "agentMatrix": [ { "type": "text", "status": "blocked" }, { "type": "search", "status": "partial" }, { "type": "code", "status": "blocked" }, { "type": "browser", "status": "success" }, { "type": "full_automation", "status": "success" } ] }, "fairnessDeclaration": { "reviewPassed": false, "note": "Q2 2026 audit complete on a single task (C 180 Limousine, config-ready close state), scored under the public Task Selection Doctrine. AI Visibility re-audited 2026-05-22 on an unbranded informational probe (43.61) so Mercedes-Benz sits on the same unbranded discovery basis as the rest of the index; AI Usability is carried from the v6.2 baseline pending a clean fleet rerun. Fairness Review pending the sector fairness grid.", "methodologyUrl": "https://www.hyperize.ai/en/methodology/task-selection", "sectorGridRef": null }, "activeTasks": [ { "title": { "en": "Mercedes-Benz C 180 Limousine", "de": "Mercedes-Benz C 180 Limousine" }, "closeState": "config_ready", "intermediaryMode": "none", "businessRelevance": null, "difficultyBand": null, "engpass": { "en": "Configurator parseable; the transactional close stalls and the simplest agents are blocked. Used-car intent leaks to marketplaces as a secondary gap.", "de": "Konfigurator parseable; der transaktionale Abschluss stockt und die einfachsten Agenten werden geblockt. Gebrauchtwagen-Intent leakt als sekundäre Lücke an Marktplätze." } } ], "scoreHistory": [ { "date": "2026-05-15", "quarter": "Q2 2026", "waveId": "a fleet wave", "protocolVersion": "v1-hand-assessment", "summary": "First-pass hand assessment from the a fleet wave era. Superseded by the v3 audit-derived measurement below." }, { "date": "2026-05-18", "quarter": "Q2 2026", "waveId": "WAVE-Q2-2026-PILOT", "protocolVersion": "ars-methodology/v1.1", "summary": "v3 audit-derived AI Visibility (27.08, 18/18 valid datapoints across 3 providers) on the C 180 Limousine task. AI Usability carried from the v6.2 baseline fleet run, pending a clean rerun." }, { "date": "2026-05-22", "quarter": "Q2 2026", "waveId": "WAVE-Q2-2026-PILOT", "protocolVersion": "ars-methodology/v1.1", "summary": "AI Visibility re-audited on an unbranded informational probe (43.61, was 27.08 with a branded informational in the basis). On the clean unbranded basis Mercedes surfaces strongly for premium mid-class sedan intent, so discovery is no longer the headline gap. AI Usability unchanged (browser reaches config-ready; the simplest classes are blocked). Composite recomputed two-axis to 39.52; the carried v6.2 hand-C/A and the 4-dimension composite (69.42) were removed. Confidence C, single task." } ], "methodology": { "formula": "Agent Success Score = (AI Visibility × 0.20) + (AI Usability × 0.70) + (Evidence × 0.10)", "usabilityDerivation": "usability-derivation/v1", "protocolVersion": "ars-methodology/v1.1", "taskSelection": "https://www.hyperize.ai/en/methodology/task-selection", "agentSurface": "https://www.hyperize.ai/en/methodology", "dimensions": { "aiVisibility": "AI Visibility (Discoverability) — audit-pipeline-derived. 3–4 providers × 3 query variants × 2 runs.", "aiUsability": "AI Usability — derived from the per-class access profile (which agent classes reach the close state) modulated by close-state depth. The profile is the truth; the score is a reproducible summary, not a hand-rating.", "evidence": "Evidence — cross-method ground-truth consistency. A confidence layer, not a sales axis." } }, "consequenceClose": { "archetype": "actionability", "archetypeLabel": { "en": "actionability", "de": "Actionability" }, "source": "brand", "diagnosis": { "en": "On product discovery, Mercedes-Benz is found and parsed cleanly: model pages carry clean entity markup, the configurator is research-grade. One thing stalls the lane, and it is the same thing wearing two faces. The last mile is not callable: agents reach config-ready, but the booking is built for humans, and the simplest agent classes are blocked before the page even loads. Used-car intent leaking to autoscout24 and mobile.de is a real but secondary gap, a separate intent this task did not measure.", "de": "In der Produkt-Discovery wird Mercedes-Benz sauber gefunden und geparst: Modellseiten tragen sauberes Entity-Markup, der Konfigurator ist Research-grade. Eine Sache stockt die Lane, und es ist dieselbe Sache mit zwei Gesichtern. Die letzte Meile ist nicht callable: Agenten erreichen Config-ready, aber die Buchung ist für Menschen gebaut, und die einfachsten Agent-Klassen werden geblockt, bevor die Seite überhaupt lädt. Gebrauchtwagen-Intent, der an autoscout24 und mobile.de verloren geht, ist eine reale, aber sekundäre Lücke, ein separater Intent, den dieser Task nicht gemessen hat." }, "evidence": { "en": "Mercedes-Benz is machine-readable, but not yet transactable; agents read the configurator cleanly, then stall at a booking built only for humans, and the simplest agents never load the page.", "de": "Mercedes-Benz ist maschinell lesbar, aber noch nicht transaktionsfähig; Agenten lesen den Konfigurator sauber, stocken dann an einer Buchung, die nur für Menschen gebaut ist, und die einfachsten Agenten laden die Seite nie." }, "question": { "en": "The question this raises: where would deeper Action-pathway measurement — handoff stability, deep-link routing, structured booking endpoints — sharpen the read?", "de": "Die Frage daraus: Wo würde tiefere Action-Pfad-Messung — Handoff-Stabilität, Deep-Link-Routing, strukturierte Booking-Endpunkte — die Lesart schärfen?" }, "confidenceHedge": { "en": "Based on the current pilot slate, this read is directional.", "de": "Basierend auf dem aktuellen Pilot-Slate ist diese Lesart direktional." }, "closes": { "en": "One move changes the number: make the last mile callable. A structured configuration and booking an agent can complete, not a human-only form, and a surface the simplest agent classes can reach without being turned away at the door. The secondary lever, once the last mile holds, is recovering used-car intent with an agent-routable certified-pre-owned surface before the marketplaces frame it.", "de": "Ein Schritt ändert die Zahl: die letzte Meile callable machen. Eine strukturierte Konfiguration und Buchung, die ein Agent abschließen kann, kein Formular nur für Menschen, und eine Surface, welche die einfachsten Agent-Klassen erreichen, ohne an der Tür abgewiesen zu werden. Der sekundäre Hebel, sobald die letzte Meile hält, ist Gebrauchtwagen-Intent zurückzuholen, mit einer agentenroutbaren Certified-Pre-Owned-Surface, bevor die Marktplätze ihn rahmen." }, "proof": { "en": "The proof isn't a cleaner configurator. It's agents completing a booking, and the simplest classes getting through the door, re-measured each wave.", "de": "Der Beweis ist nicht ein sauberer Konfigurator. Es ist, dass Agenten eine Buchung abschließen und die einfachsten Klassen durch die Tür kommen, neu gemessen mit jeder Wave." }, "ctas": [ { "weight": "primary", "eyebrow": "Audit · €1,900", "title": { "en": "Commission an audit.", "de": "Audit beauftragen." }, "body": { "en": "Where the BrandScore opens the question, an audit closes it. An interpretive engagement on your full surface, scored under the same methodology.", "de": "Wo die BrandScore die Frage öffnet, schließt sie ein Audit. Ein interpretatives Engagement auf der vollen Surface, gemessen unter derselben Methodologie." }, "cta": { "en": "Get the audit", "de": "Audit anfragen" }, "href": "mailto:hello@hyperize.ai?subject=Audit%20%C2%B7%20Mercedes-Benz" }, { "weight": "secondary", "eyebrow": "Founding · €4,500", "title": { "en": "Found with us.", "de": "Founding Program." }, "body": { "en": "Strategic partnership for brands building agent success as a long-term capability, not a one-off engagement.", "de": "Strategische Partnerschaft für Marken, die Agent-Success als langfristige Capability aufbauen — nicht als Einmal-Engagement." }, "cta": { "en": "Apply", "de": "Bewerben" }, "href": "mailto:hello@hyperize.ai?subject=Founding%20Program%20%C2%B7%20Mercedes-Benz" }, { "weight": "secondary", "eyebrow": "Snapshot", "title": { "en": "Audit an adjacent property.", "de": "Eine angrenzende Property messen." }, "body": { "en": "The BrandScore covers the primary domain. Get the same methodology applied to an adjacent property: a country site, a sub-brand, a category beyond the DAX-40 slate.", "de": "Die BrandScore deckt die primäre Domain ab. Die gleiche Methodologie für eine angrenzende Property: eine Länderseite, eine Sub-Brand, eine Kategorie außerhalb des DAX-40-Slates." }, "cta": { "en": "Get a Snapshot", "de": "Snapshot anfragen" }, "href": "mailto:hello@hyperize.ai?subject=Snapshot%20%C2%B7%20adjacent%20property%20for%20Mercedes-Benz" } ] }, "sources": [ { "id": "S1", "label": "Gate-1 audit run · Mercedes-Benz Wave Q2 2026 (unbranded re-audit,)", "accessedAt": "2026-05-22", "supports": [ "AI Visibility score (audit-derived, 18/18 valid datapoints, unbranded informational probe)", "AI platforms queried (openai/perplexity/anthropic, 3-provider track)", "the close state reached (config_ready)", "task correction from E-Klasse E 220 d to C 180 Limousine" ], "url": null, "kind": "internal", "internalLabel": "Internal · Hyperize audit pipeline" }, { "id": "S2", "label": "Hyperize fleet · v6.2 baseline (source for C/A/E)", "accessedAt": "2026-05-16", "supports": [ "AI Usability and Evidence inputs", "fleet test outcomes (legacy)" ], "url": null, "kind": "internal", "internalLabel": "Internal · Hyperize fleet (v6.2 baseline)" }, { "id": "S3", "label": "a fleet wave automotive wave (Giorgio repo, legacy reference)", "accessedAt": "2026-05-18", "supports": [ "browser-agent phase reference (legacy)" ], "url": null, "kind": "internal", "internalLabel": "Internal · Hyperize fleet (Giorgio)" }, { "id": "S4", "label": "Task Selection Doctrine", "accessedAt": "2026-05-18", "supports": [ "fairness declaration", "Third-Party Interception framing" ], "url": "https://www.hyperize.ai/en/methodology/task-selection", "kind": "public", "internalLabel": null } ], "crossLinks": { "index": "https://www.hyperize.ai/en/dax40-index", "taskSelection": "https://www.hyperize.ai/en/methodology/task-selection", "agentSurface": "https://www.hyperize.ai/en/methodology", "foundingProgram": "https://www.hyperize.ai/en/founding-program" }, "scope": { "publishes": [ "Brand identity (name, legal entity, DAX ticker, sector)", "Composite (index-ranking) + AI Visibility + AI Usability (as access profile + derived score) + Evidence, with weights", "Per-agent-class access profile (which agent classes reach the close state)", "Third-Party Interception classification, direct/intermediary share, narrative", "Hyperize-selected task titles, close state, intermediary mode, task score, bottleneck sentence", "Measurement timeline (date, wave, protocol version, summary)", "Sources with public URLs or internal-evidence pointers", "Confidence grade + Evidence tier + Next review date", "Bottleneck classification (Discovery / Interception / Usability / Multi-axis / No dominant gap)", "Bottleneck archetype + consequence-claim diagnosis + operator question", "Test narrative (verdict, narrative, agent matrix, scope shift) — scene-level prose, no prompts or formulas" ], "doesNotPublish": [ "Task identifiers (internal)", "Raw and derived measurement artifact paths (internal repos)", "Per-phase outcome tables (ceiling / http / coding / browser_standard / act)", "Individual prompts and provider-specific transcripts", "Scoring derivation steps below the dimension level", "Sector task grid file references when still draft" ] } }