Methodology · Doctrine

Task Selection.

A fair score starts with task selection, not measurement. The Hyperize DAX 40 Index v3 refuses to score a brand from one lucky or unlucky path. Tasks must be frozen, public, and defensible before any wave runs. Direct Surface Score and Third-Party Interception are reported separately so structural intermediary capture is never silently folded into surface failure. This page sets out the doctrine, the five failure modes it prevents, and how it applies to Allianz, Mercedes-Benz, DHL, and Bayer.

A score from one task is not a score for a brand. Tasks must be frozen and public before the wave is run.

Why this exists

One task is not the brand.

Most AI-visibility scoring runs one query, finds a gap, and publishes the gap as the brand's score. That gap is sometimes real and sometimes a single-path artefact. Once it is published, the brand has to argue against an unfair number — or live with one.

The Hyperize Agent Surface methodology separates this question into two: which tasks are fair to test, and how the brand performs on those tasks. Task selection comes first. Measurement is downstream.

We call the prototypical mistake the Hero-task Fallacy — using one heroic-or-broken task to stand in for a brand. The doctrine below exists to make this fallacy impossible.

The doctrine

Six rules. Public. Fixed before measurement.

Every DAX 40 brand under v3 is selected against the same six. No exceptions. No retroactive tuning.

A task must represent a real customer need, not a synthetic trap.

A task must test a surface the brand actually owns, not a channel it can never own directly.

A task set must contain multiple tasks, never a single path.

Tasks within a sector must have comparable difficulty across brands.

Structural intermediary capture must be reported as Third-Party Interception, a separate axis from Direct Surface Score — never silently folded into surface failure.

The task list must be frozen and publicly published before the wave is run. Tasks cannot be retroactively adjusted to favourable outcomes.

What the doctrine prevents

Five failure modes.

Each rule above answers to one or more of these. Source for the failure-mode set: the v3 sector task grid template [S3].

Failure mode

What goes wrong

Prevention

FM1

Needle-in-a-haystack bias

Picking the exact path where a brand fails and treating it as the whole brand.

Use 3–5 tasks across distinct task classes; never score a brand from a single path.

FM2

Cross-brand unfairness

Comparing brands on tasks of different difficulty or different task classes.

Use one frozen sector grid for all brands in the sector.

FM3

Market-structure distortion

Scoring structural intermediary capture as if it were purely a brand-surface failure.

Declare intermediary_mode per task; render Direct Surface and Third-Party Interception separately.

FM4

Task drift over time

Changing tasks quarter-to-quarter without versioning, breaking trend comparability.

Freeze task schemas per quarter and append new versions rather than editing silently.

FM5

Defensibility overfit

Choosing only easily measurable tasks instead of commercially relevant ones.

Require both business relevance and measurement tractability before inclusion.

The load-bearing decision

Direct Surface and Third-Party Interception are scored separately.

Axis 1

Direct Surface Score

Measures what the brand actually owns: structured product data, accessible documentation, retrievable content, completable journey paths. This is the surface the brand can build, fix, and verify.

Axis 2

Third-Party Interception

Classifies intermediary capture per task as none, avoidable, or structural. Structural capture is reported, never silently folded into surface failure. Avoidable capture is a surface opportunity.

A brand can have a strong Direct Surface and weak Channel Position — or the reverse. Collapsing both into one score is exactly the FM3 distortion the doctrine prevents [S3].

Applied

Four brands. Four channel structures. One doctrine.

The doctrine holds across structurally different brands. Allianz, Mercedes-Benz, and DHL are the v3 pilot trio. Bayer is the edge case the doctrine was stress-tested against [S1] [S2].

Insurance · Financial Services

Allianz

Channel position · avoidable

Allianz could own most of the direct-quote path. Verivox and Check24 capture demand, but the capture is avoidable — not structural.

Fair to test

Get a household-contents insurance quote; compare two policy variants; find evidence for a coverage claim; reach the next-step contact handoff.

Not fair

Scoring the broker comparison page as if Allianz had to win it. That would smuggle the aggregator capture into Direct Surface failure (FM3).

Controls

FM1 · FM3

Automotive · Premium

Mercedes-Benz

Channel position · structural+avoidable

Mercedes owns the configurator end-to-end. Final purchase is structurally dealer-mediated in the DACH automotive market. Two task classes are scored separately.

Fair to test

Configure a specific model; find financing scenarios; locate documentation for an existing vehicle; identify the right dealer for a test drive.

Not fair

Demanding direct online checkout for a new car. That ignores the structural dealer-mandate in DACH automotive distribution (FM3).

Controls

FM2 · FM3

Logistics · Postal Services

DHL

Channel position · none

DHL is the channel. There is no intermediary between the customer and the shipping action. Direct Surface carries the whole score.

Fair to test

Calculate a parcel price; track an existing shipment; schedule a pickup; find the right customs documentation for an EU-to-non-EU shipment.

Not fair

Comparing DHL's price calculator to an aggregator that doesn't exist for the use case. That would invent a channel layer that isn't there (FM2).

Controls

FM5

Pharma · Consumer Health

Bayer

Channel position · structural

OTC commerce is pharmacy-mediated by law and habit. Bayer is the hardest stress test for the doctrine — multi-brand portfolio, structural intermediary capture, uneven content-to-commerce bridges.

Fair to test

Aspirin family symptom path; Bepanthen wound-care path; Canesten antifungal path; one clinical-documentation retrieval task. Five tasks, three product families, one trust-info anchor.

Not fair

Single-SKU measurement (FM1); direct Bayer checkout as a success criterion for OTC (FM3); generic 'best headache medicine' without a Bayer-surface constraint (FM2 + FM3).

Controls

FM1 · FM2 · FM3

Governance

The Frozen Wave Rule.

Once a wave is published, its task set is frozen. Re-tests append to history; they never overwrite prior measurements. Each published value traces to a derived record, which traces to raw evidence [S3].

Task changes between quarters are versioned — pharma_healthcare/v1 becomes pharma_healthcare/v2 with the change documented, not swapped silently. Trend comparability is preserved by reading within a version, not across silent edits (FM4 controlled).

Brands that disagree with their score have a public challenge intake; brand records carry a history.challenges block. Overrides are public-history-visible. The score is contestable; the doctrine is not.

On scope

The doctrine is public. The mechanics are not.

The five failure modes, six rules, and Direct Surface / Third-Party Interception split are published. The specific task YAMLs, the prompt phrasings, the agent fleet configuration, the scoring formulas, and the run protocols are proprietary. The category belongs in the open. The operating system stays inside the engagement.

Sources

Evidence and provenance.

internal

Life Sciences Wave Scorecard (Bayer)

Hyperize Internal — Fleet Operations · April 2026

Hyperize fleet · Life Sciences wave (internal)

Supports: Bayer architecture-gap finding, Aspirin Forte vs. 500mg contrast, FM3 pharmacy-mediation observation.

internal

Life Sciences Wave v6 output (Bayer entry)

Hyperize Internal — Fleet Operations · April 2026

Hyperize fleet · Life Sciences wave v6 (internal)

Supports: Bayer task framing, phase scores, failure mode, surface type. Legacy note: Fleet D values precede the v3 audit-derived Discoverability dimension.

internal

DAX 40 Sector Task Grid Template v1

Hyperize Internal — Methodology · May 2026

Clients/DAX40-GEO-Index/v3/sector-task-grid-template-v1.yaml

Supports: Five failure modes, fairness criteria, task-class logic, frozen-wave + append-only-history governance hooks.

internal

Pre-v3 Discovery and Journey scoring

DAX 40 Index v1 — Provisional Hand-Assessment · March 2026

Clients/DAX40-GEO-Index/dax40-geo-index.json

Supports: Allianz channel-context observation (Verivox/Check24 aggregator competition); Mercedes-Benz configurator strength + dealer-dependent completion; DHL end-to-end digital journey signal. Provisional, superseded by v3 audit-derived measurement.

Internal sources reside in the Hyperize project repository (Giorgio-Hyperize org). Public derivative artifacts are linked from the brand reference pages once the v3 audit pipeline is operational.

Last updated 2026-05-23

Next review 2026-08-15

Confidence A

Evidence tier Proprietary

Where this connects.

Upstream

Agent Surface

The methodology this doctrine serves. Why brands have surfaces, what makes them strong, what Hyperize builds.

Read the methodology

Downstream

Agent Success Score

The metric the fair tasks feed into. Two axes, AI Visibility and AI Usability, plus Third-Party Interception, scored per brand on the frozen task set.

Coming next · Anchor page

Applied

DAX 40 Index

The public dataset where the doctrine is applied. Allianz, Mercedes-Benz, and DHL anchored the v3 pilot wave; the index grows by wave.

Phase 2 · Index hub