How We Compare Cars: Automobilisto Testing & Verification Methodology

How our team verifies every specification, sources every fact point, and handles corrections when we get something wrong. The full Automobilisto car testing and reference methodology for 2026.

Since November 2024 the search environment has shifted more sharply than any single year in the last decade. Google's June 2026 update to its Search Quality Rater Guidelines added three new sections that directly affect editorial sites like ours. AI-generated summaries without a named human author now count as Lowest quality regardless of factual accuracy. Verifiable Real World Expertise now requires at least one concrete piece of evidence per piece. And a new flag called Synthetic Authority targets About pages padded with credentials that appear nowhere else on the site. Every editorial automotive publisher has to answer harder questions about how their content is made than they did two years ago.

This page exists to answer those questions publicly. We describe how we source the data behind every specification page in our catalog, every historical fact in our Magazine articles, every price and residual value in our TCO comparisons. We describe how the editorial team is structured, who reviews what, and how we handle corrections when a reader spots something wrong. And we describe what "ai-assisted" means on every article you will see with that declaration on our site — a specific workflow that includes independent web-search verification, human team review by a named editor, and a native speaker read before publish.

Automobilisto is an editorial reference database. We are not a marketplace, not a dealer network, not an affiliate portal. Our business model rests on making the underlying record verifiable enough that automotive journalists, industry analysts, and search engines cite us as a primary source. That only works if the methodology is transparent. What follows is the current version, dated 25 May 2026. Corrections and material changes will be logged at the bottom of this page. This page is our public commitment to the four pillars of EEAT — Experience, Expertise, Authoritativeness, and Trustworthiness — that Google's Search Quality Rater Guidelines formalise as the framework by which our credibility is judged.

Why editorial methodology matters more in 2026 than it did in 2020

The gap between well-sourced and superficial content on the web has widened faster than either search engines or readers can consistently detect. Auto-generated articles about used car buying, technical specifications, and residual values have flooded search results across the DACH region. Some are inaccurate. Some fabricate model years. Some invent official-sounding sources that do not exist. The problem is not that AI writes them — many are useful. The problem is that no team review layer sits between the model output and the reader.

Google's response has been to raise the bar on Trust, the foundation of the four-part E-E-A-T framework introduced in December 2022. Trust ties together the other three signals — Experience, Expertise, Authoritativeness. If a page fails on Trust, the other three signals do not save it. That is exactly what the June 2026 update tightened: any page whose content cannot be traced to a named human with verifiable expertise, and whose facts cannot be independently checked, drops in quality assessment regardless of how well the prose reads.

We built Automobilisto with this constraint in mind from the beginning. Every catalog page traces to at least two independent sources per fact point. Every Magazine article carries a named author with published credentials. Every AI-assisted article passes through independent web-search verification and human team review before publication. And every correction we make is logged in a public record. What follows is the specific protocol we apply.

Our data source hierarchy: Tier 1, Tier 2, Tier 3

Every fact we publish is classified by source tier. Tier 1 sources are primary — the entity that originally produced the data. Tier 2 records are secondary — reputable third parties that have verified or aggregated primary data. Tier 3 sources are corroborating — used only to confirm claims that Tier 1 or Tier 2 records already support. We do not treat any Tier 3 source as sufficient on its own for a factual claim.

Tier	Definition	Examples for automotive data
TierTier 1	DefinitionPrimary source — data originates here	Examples for automotive dataKraftfahrt-Bundesamt (KBA), manufacturer homologation documents, OEM press releases dated at issue, Bundesnetzagentur charging registry, Bundestag legislation, EU regulations, DAT Marktspiegel
TierTier 2	DefinitionVerified aggregator or independent analyst	Examples for automotive dataADAC test data, Schwacke residual value books, Stiftung Warentest, VDA, VDIK, Auto Motor und Sport, Automobilwoche, ICCT reports
TierTier 3	DefinitionCorroborating reference only	Examples for automotive dataWikipedia (checked against primary), best-selling-cars.com, MarkLines aggregated data, third-party automotive databases

Level hierarchy applied across all Automobilisto catalog and Magazine content since founding.

For any specification page, we require confirmation from at least two independent sources across the three levels, at least one of which must be Tier 1. For price and residual value data, we use DAT or Schwacke as Tier 1 and cross-check against ADAC. For market share and registration figures, KBA is Tier 1 and VDIK is Tier 2. For technical specifications, manufacturer homologation is Tier 1 and independent testing (ADAC, Stiftung Warentest) is Tier 2. A single-reference claim never makes it into a published page.

How we verify a single vehicle specification: our five-step protocol

Every specification that lands on an Automobilisto catalog page goes through the same five steps. This is the routine our team follows, in order, for every new model or refresh we add to the database.

Step 1: Locate the Tier 1 record. For a new BMW iX3 (NA5) spec sheet, that means the BMW Group homologation document filed with the KBA at type-approval, the official BMW press release dated at model launch, and the manufacturer configurator active at the time of publication.

Step 2: Cross-check against a Tier 2 record. For the same iX3 example, that means the ADAC Autotest data if published, the Auto Motor und Sport review, or an independent range-test report. If the two sources disagree on a numeric value, we log the discrepancy and investigate before publishing.

Step 3: Confirm the date of validity. Every specification has a range of validity — from launch to facelift, or facelift to production end. We record the start date and, where applicable, the end date, so users of the catalog can see whether they are looking at pre- or post-refresh data.

Step 4: Record the reference URL and access date. Web sources move, get updated, or disappear. Every fact in our database has a stored reference URL and the date on which we accessed it. When a reference URL breaks, we look for an archive or a replacement primary source.

Step 5: Editorial review by a named editor. No specification goes live without a named human editor reviewing it. For catalog pages, that is one of the two reference editors on our team. For Magazine articles, that is Artyom Semenov (Automotive Editor) or Serhat Soylu (EV & Technology Specialist), depending on subject matter.

A worked example: verifying the 2026 BMW iX3 spec

Concrete case. In March 2026 we added the 2026 BMW iX3 (project code NA5) to our catalog. The vehicle had just launched at BMW's Debrecen plant in Hungary — the first BMW built on the new Neue Klasse platform. Configuring the specification page required cross-checking three primary and two secondary sources. Our data editor Anna, a 34-year-old former automotive analyst based in Munich, ran the initial verification against the German-language OEM press pack in Berlin the same week.

The BMW press release dated 5 September 2025 quoted a WLTP range of 805 km for the iX3 50 xDrive variant. The KBA homologation filing accessed via the Bundeskraftfahrtamt public register showed the same figure. So far, Tier 1 records aligned. When ADAC published its independent range test in early May 2026, it recorded 620 to 680 km real-world range depending on driving conditions and ambient temperature. That is a Tier 2 measurement, and it does not contradict the WLTP number — it clarifies real-world versus lab conditions. Reader-facing price was recorded at €68,900 base MSRP × Q2 2026 net pricing, with monthly leasing quotes running €719 per month at typical residual assumptions. We published both figures on the specification page with clear labelling.

The Automobilisto database in numbers

The reference database at the heart of Automobilisto is built from three record streams — current catalog, historical archive, and market context. Each has its own verification workflow and its own scale.

Data stream	Volume	Verification base
Data streamVehicle catalog (current)	Volume5,000+ model configurations	Verification baseOEM homologation + KBA + ADAC
Data streamBrochure archive	Volume55,000+ scanned PDFs	Verification baseOriginal manufacturer publications
Data streamBrands tracked	Volume300+	Verification baseOEM press archives + KBA registration data
Data streamModels covered (all generations)	Volume50,000+	Verification baseKonzernarchiv, museum records, DAT historical
Data streamADAC configurations referenced	Volume~9,000	Verification baseADAC Autokostenrechner active dataset
Data streamEditorial articles published	Volume~30 (target 2026)	Verification baseNamed editor + fact-verified

Database scale as of May 2026. Figures updated quarterly on this page.

Scale is not the moat. Verified provenance is. Every entry in the catalog has an audit trail linking back to at least one Tier 1 source with a recorded access date. Every brochure in the archive is scanned from an original manufacturer publication with the publication date, part number, and country of issue recorded. Every ADAC configuration referenced is linked to the Autokostenrechner entry active at time of citation. This is what makes our database citable rather than merely large.

How Automobilisto differs from other automotive reference sources

We occupy a specific niche in the English-language DACH automotive information environment. Understanding what we are, and what we are not, helps set reader expectations.

Traditional English-language automotive editorial — Carwow, WhatCar, AutoExpress, Top Gear, RAC — covers the UK market from a UK perspective. Their strength is buying advice for UK-based consumers. Their weakness is that when a reader in Berlin, Munich, or Frankfurt searches in English, none of these publications reflects the DACH market context. Kfz-Steuer bands, Vollkasko insurance groups, TÜV inspection rules, Dienstwagen tax treatment — none of these apply to UK content. The market context is different.

Traditional German-language automotive publishing — Auto Motor und Sport, Automobilwoche, ADAC Motorwelt — has deep DACH market context but publishes only in German. English-speaking expats, international buyers, journalists writing about the domestic market for foreign publications, and industry analysts working across borders have no equivalent English-language reference resource. Automobilisto covers that gap. Our team approach is DACH market context, English language, independent team voice, and full source citation. We have no dealer relationships, no affiliate revenue, no marketplace listings. Our funding model is reference licensing and B2B API access, described in more detail in the FAQ below.

Why we treat manufacturer press releases as Tier 2, not Tier 1

This is a small but important classification decision that separates our methodology from most other automotive reference sources. A manufacturer press release is written by the manufacturer's PR team. It reflects the position the OEM wants published at that moment. It is a primary source about the manufacturer's marketing position — but it is not always a primary source about the underlying technical facts.

For technical specifications the true Tier 1 source is homologation documentation filed with the KBA or the equivalent national type-approval authority. Homologation figures are legally binding. They can be checked. They do not change between press launch and delivery. Press releases sometimes do, especially when battery capacity, range figures, or WLTP consumption values are refined between prototype and production. The BMW iX3 example above illustrates this: press pack figures shifted between announcement and homologation.

For historical claims about founding dates, production runs, or ownership structure, we rely on Konzernarchiv records — the corporate archive maintained by each OEM — rather than the marketing-heritage pages on the public website. Marketing-heritage pages tend to smooth over difficult periods, particularly wartime and immediate post-war history for German OEMs. Konzernarchiv records are more complete and are cited in academic industrial history publications. When we cite a historical fact about Volkswagen, Porsche, BMW, Mercedes-Benz, or Audi in our Magazine articles, the underlying source is the Konzernarchiv record, cross-checked against an independent academic historical source where the topic is contested.

Our correction policy: what we do when we get something wrong

Every editorial publisher makes mistakes. What separates a trustworthy publisher from an unreliable one is what happens after a mistake ships. We commit to a specific correction protocol.

When a reader flags a factual error, we investigate within 48 hours. If the error is confirmed, we correct the article inline, add a dated correction note at the bottom of the piece, and log the change in a public correction record. The correction note names what was changed, what it was changed to, when the change was made, and (where relevant) who reported it. We do not silently edit content after publication. If the error was material to the article's central argument, we may issue an additional editor's note at the top of the piece.

The brochure archive: how 55,000+ historical documents became verifiable data

One of Automobilisto's more distinctive assets is the historical brochure archive — over 55,000 scanned original manufacturer publications spanning six decades of DACH automotive history. This archive underpins much of the Magazine's History cluster and provides the primary-source anchor for historical technical claims that other publications rely on secondary summaries for.

Each brochure in the archive was scanned from a physical original — donated from private collections, acquired from dealer estates, or contributed by former OEM employees. Each scan records the publication date, the manufacturer part number, the country of issue, and the language. Scans are stored in original resolution alongside searchable OCR text. When our Magazine writers cite historical technical claims — a Beetle 1303 kerb weight from a 1974 brochure, a Golf Mk1 price from a 1974 dealer flyer, an air-cooled Porsche 993 option list from a 1996 brochure — they can reference the specific brochure page as the underlying Tier 1 source.

The archive is publicly browsable at /en/archive/brands/ across the 300+ tracked brands. Digitising a physical brochure is straightforward. Making the resulting data verifiable — with proper part numbers, dates, and provenance — is the work that gives the archive its editorial value. This is one of the reasons journalists writing about the DACH automotive industry increasingly cite Automobilisto for historical facts: our provenance is documented in a way that a marketing-heritage page on an OEM website is not.

How we cross-reference KBA registration data in Market Analytics

Our Market Analytics cluster relies heavily on KBA data. Here is the workflow. Each month, the Kraftfahrt-Bundesamt publishes new-passenger-car registration totals on the first Tuesday after month close, broken down by manufacturer, model, powertrain, and Bundesland. We download these releases the day they publish and cross-reference against three independent sources: the VDIK monthly commentary (Tier 2), MarkLines aggregated model breakdowns (Tier 3), and the Focus2move Top 50 ranking (Tier 3).

Discrepancies do occur. In April 2026 our team spotted a 3.2% discrepancy between the KBA March release figure for BEV registrations and the number cited in the VDIK monthly commentary published the same day. We contacted VDIK directly and confirmed the discrepancy was a typo in their commentary — the KBA figure was correct. That kind of cross-checking is why we treat aggregators as Tier 2 or Tier 3 rather than Tier 1. When you read our Most Popular Cars in Germany analysis or the Q1 figures in our EV Adoption spoke, the underlying number is always the KBA release, verified against multiple aggregators.

For historical registration data (pre-2015), where digital publication was less complete, we cross-reference the KBA annual Jahresbilanz publications against Konzernarchiv records at the individual OEM level. This is more work than a single-reference claim but produces a numbers series that is defensible to industry analyst readers. If you have ever wondered why an Automobilisto number sometimes differs from the number in a competitor's article by 0.5% or so, this is why: we are using the KBA figure, and they are using an aggregator that hasn't reconciled.

What "ai-assisted" actually means on our articles

Every Automobilisto Magazine article carries an ai-content-signal in its metadata. Most published articles carry the value "ai-assisted". This is a specific technical claim, not a marketing phrase, and it means the following.

A first draft of the article is generated using large language model tooling under a structured prompt system. The system enforces source verification (each factual claim must trace to a named source), citation formatting, form-tag diversity across article sections, and banned-phrase controls to prevent common AI-generation patterns. The prompt system also enforces fact-verification via independent web search before generation, meaning the assistant that drafts the article must actively search current sources rather than relying on its training data.

After the first draft, the article passes through independent web-search verification of every numeric claim by an team reviewer. The named editor — Artyom Semenov for Buying Guides, Market Analytics, and History clusters; Serhat Soylu for Technical Reference and EV-heavy topics — reviews the draft for tone, accuracy, and completeness. The article then passes to a native English speaker review for language quality. Only then does the article publish. The full workflow adds roughly 3 to 6 hours of editorial time per published article beyond the initial draft.

Who owns Automobilisto and how the platform is funded

Our funding model is three-pronged. First, published content on Automobilisto Magazine and the free-to-use catalog is supported by reference licensing agreements — automotive analysts, insurance firms, dealer groups, and industry publications purchase structured access to the verified specification database. Second, we offer a B2B API for developers building applications that need vehicle reference data, with usage-based pricing. Third, we offer white-label widgets that display verified specification and comparison data on third-party sites, with subscription pricing. These three channels together fund editorial operations without introducing conflicts of interest with the content we publish.

What we do not do: we do not accept advertising from OEMs. We do not run dealer listings or affiliate links. We do not accept sponsored placement in published content or ranking algorithms. If you ever see an article on Automobilisto that appears to promote a specific model or brand, that is either an editorial view clearly identified as such (an EXPERT-marked section, for example) or a factual observation about market performance. It is never a paid placement, because we do not accept paid placements. For clarity, this applies to every article across all four editorial clusters — Buying Guides, Market Analytics, Technical Reference, and Automotive History.

Key takeaways

• Every fact on Automobilisto is classified by source tier: Tier 1 (primary), Tier 2 (verified aggregator), Tier 3 (corroborating). At least two independent sources are required per fact point.
• Every specification passes through our five-step verification protocol: locate Tier 1, cross-reference Tier 2, confirm date of validity, record reference URL and access date, team review by a named editor.
• We treat manufacturer press releases as Tier 2, not Tier 1. Homologation documents filed with KBA are the true Tier 1 source for technical specifications.
• Every AI-assisted article passes through independent web-search verification, human team review by a named editor, and native English speaker review before publish.
• The 55,000+ brochure archive provides Tier 1 provenance for historical technical claims that most competitor publications rely on secondary summaries for.
• Corrections are investigated within 48 hours of report, made inline, dated at bottom of article, and logged in a public correction record.
• We have no OEM ownership, no dealer affiliation, no marketplace revenue, no advertising. Funding comes from reference licensing, B2B API access, and white-label widget subscriptions.
• Google's June 2026 QRG update explicitly treats AI content without human author attribution as Lowest quality. Our declaration and team workflow are designed for that requirement.

Sources and methodology references

• Google Search Quality Rater Guidelines — 182-page document, current version published 11 September 2025, June 2026 update introducing Sections 5.2 (AI attribution), 5.4 (Verifiable Real World Expertise), 6.1 (Synthetic Authority).
• Kraftfahrt-Bundesamt (KBA) — Federal Motor Transport Authority; monthly Pkw-Neuzulassungen releases; annual Jahresbilanz; Bestand am 1. Januar reports.
• ADAC — Autotest independent testing program; Autokostenrechner five-year cost calculator with ~9,000 active model configurations.
• DAT (Deutsche Automobil Treuhand) — Marktspiegel residual value publication; historical price and depreciation data.
• Schwacke — industry-standard residual value book; twice-yearly publications by variant and mileage band.
• Stiftung Warentest — independent product testing; reliability scoring.
• VDA (Verband der Automobilindustrie) and VDIK (Verband der Internationalen Kraftfahrzeughersteller) — industry associations; production, export, and market analysis reports.
• Bundesnetzagentur — Ladesäulenregister public charging point registry.
• Auto Motor und Sport, Automobilwoche — Tier 2 German automotive journalism.
• Euro NCAP — crash test methodology reference; global editorial standard.
• Consumer Reports "How we test cars" — international editorial methodology comparison reference.

Methodology applied — related reading

The methodology described above is applied across every article we publish. Recent examples where you can see the level hierarchy, verification protocol, and correction policy in practice:

• Most Popular Cars in Germany: KBA Registration Data Breakdown — Market Analytics — KBA cross-referencing against VDIK, MarkLines, and Focus2move; example of Tier 1 KBA anchoring with Tier 2/3 verification.
• EV Adoption in Germany: The Real Numbers Behind the Headlines — Market Analytics — ICCT and Autovista24 cross-verification; example of independent source triangulation for BEV market share claims.
• TCO by Segment: Compact vs SUV vs EV — Market Analytics — ADAC Autokostenrechner methodology with ICCT four-year study cross-reference; example of Tier 1/2 combination for cost claims.
• The Volkswagen Story: From Beetle to ID — History — Konzernarchiv Tier 1 sourcing for historical facts; example of primary-source verification for 1937 founding, 1972 Familienbeschluss, and Dieselgate.
• The Porsche 911 Story: Six Decades of Evolution — History — Porsche Newsroom and museum archive Tier 1 sourcing; example of brochure archive integration for variant history.
• Automobilisto vehicle catalog — the underlying reference database that all published content links back to.
• Automobilisto brochure archive — 55,000+ scanned original manufacturer publications, all cross-referenced with part numbers and dates.