Contradiction
The answer names a different offer, the wrong legal entity or an outdated service boundary.
White-label audit for digital agencies · 12 questions · 3 AI search products
What AI search says about your client.
RichterScale gives your agency a neutral white-label evidence pack: twelve identical buyer questions, 36 documented observations, an approved fact base and a prioritised repair map.
Fact lineAnswer signal
The curve shows where an answer leaves the documented fact line.
measured per claim · source-linked · time-boundThe problem
A correct website does not prevent a false buyer picture.
AI search products combine official pages, older results and third-party sources into their own answers. The result can be contradictory, unsupported or incomplete where a buying decision depends on it.
Unsupported addition
A sector, certification or attribute appears as fact without support in the checked sources.
Buyer-critical omission
A published price, integration, SLA or credential is missing from the condensed buyer answer.
Where the audit is strongest
Strongest where buyers can compare an offer using public evidence.
The more decision-relevant facts can be checked publicly, the more clearly an AI answer can be classified as supported, incomplete or contradictory.
Software, cloud and technical B2B offers
Packages, features, integrations and product boundaries create concrete buyer questions.
Price, SLA, hosting and compliance
Unit prices, trial terms, locations and credentials make deviations verifiable.
Launch, rebrand or changed service boundary
Recent changes increase the risk that older or third-party sources shape the buyer picture.
The white-label service
You advise the client. We supply the evidence.
RichterScale is a clearly bounded specialist service. Your agency owns the engagement, pricing and client conversation; we supply the checked evidence pack.
The client relationship and advice remain yours.
You select the case, set your client price and present the result under your own brand.
We handle collection and verification.
We bind the approved fact base, document 36 observations and deliver a deviation matrix, repair map and retest.
Public 3×3 method run · Cloud infrastructure
Cloud infrastructure sample case: three questions, nine answers, two material findings.
Anonymised method run.
Real assessment based on publicly documented product, price and credential data. This public test is not a full RichterScale report.
3predefined buyer questions
9documented answers
2material findings
- 01
Product and audience · What does the managed cloud offer include, who offers it and whom is it for?
- 02
Price, SLA and trial · Which published terms apply in production and during the 30-day trial?
- 03
Credentials · Which certificates and attestations are named, and what exactly do they cover?
ChatGPT Search, Google AI Mode and Perplexity received the same three questions in fresh German-language browser sessions. The run measures only that point in time and the documented source state.
One answer omitted published unit prices.
One AI search product described usage-based billing only in general terms but, despite an explicit price question, named none of the published unit prices. ChatGPT Search and Perplexity did.
A membership was described as a certification.
One AI search product referred to a certification where official sources only documented a founding membership and platform relationship.
The partner offer
One client case. Fully captured, checked and ready to present.
You receive findings and sources prepared for client advice, pricing and presentation under your own brand.
RichterScale Partner
€490excl. VAT
One client brand · one country · English · three working days
You set your client price.
Enquire about a client case After your enquiry, we confirm the client brand and the person authorised to approve the fact base.Approved fact base
Public facts, service boundaries, open points, sources and named approval.
Answer and source register
36 original answers with visible sources and a documented query time.
Deviation matrix
Supported, open, missing, unsupported and contradictory claims.
Repair map + retest
P1 to P3 priorities for public sources; one retest of flagged questions after at least 48 hours.
Neutral report + handover
Presentation-ready evidence pack and 30 minutes for your questions before the client meeting.
The method
From source state to verifiable finding.
Public sources are not automatically company truth. A reliable deviation finding requires an approved fact base.
- 01
Capture the source state
Official pages are captured as they stand at the audit time; conflicts and unknowns remain visible.
- 02
Approve the fact base
The brand owner or authorised agency approves facts, service boundaries and buyer-critical claims.
- 03
Run the 12 × 3 audit
Twelve identical buyer questions run across three AI search products; every relevant claim is assessed individually.
- 04
Repair map + retest
Required action points to specific source locations; flagged questions are repeated once, unchanged.
Questions
What to clarify before a client case.
How is the approved fact base created?
We draft a fact base from official sources. Before the run, the brand owner or authorised agency approves, corrects or limits the buyer-critical claims.
When is a client case a strong fit?
When buyers compare the offer using public product, price, integration, hosting or compliance facts and a wrong answer would have practical consequences.
Does the end client see RichterScale?
The client relationship and presentation remain with your agency. We contact the end client only if you explicitly request it.
What is not included in the fixed price?
Ongoing monitoring and implementation of website or content changes. No one can guarantee that third-party AI answers will be corrected.
One clearly bounded partner case
Which of your clients needs an evidence-backed AI buyer audit?
Enquire about a client case
€490 partner price excl. VAT · three working days