English

METHODOLOGY
English

This is how JMRKT works

Methodology

Which sources we use, how comparisons are made and what their limits are. Here you will find the common principles and criteria for each product group. The respective sources and test data are also in the product profiles.

On this page

The principles below apply everywhere. You can find the detailed rules for each comparison here:

M1

Independent of commission

Affiliate links are marked. Whether a provider is included and how its characteristics are presented does not depend on whether a commission agreement exists. Advertising and editorial elements are technically separated.

M2

No own score

We do not award grades, points, stars or overall ratings. Instead, we put features next to each other so you can weight them according to your own criteria. An overall rating would assert an objectivity that does not exist with financial products - what is the best offer for you depends on your situation.

M3

Obligation to source per feature

Each feature lists the source and date of retrieval. Market availability, fees, jurisdiction and regulation will be reviewed prior to publication. Information that comes exclusively from the provider is marked as manufacturer information.

M4

You decide the order

Comparison sites start in a neutral basic order. You control the sorting and filters yourself via the interface - we don't put any provider at the top and we don't give out awards such as “Test Winner” or “Editor's Recommendation”.

The testing process

STEP 1

Capture

Features from provider documentation and public repositories, each with URL and retrieval date.

STEP 2

Prove

Where possible via a second, provider-independent source. Where this is not possible, the information remains marked as manufacturer information.

STEP 3

Editorial team

Classification according to the criteria defined for each category. Language without superlatives and without the promise of returns.

STEP 4

Update

Occasion-related: with information from readers, with noticeable changes to the offer and with regulatory developments. There is currently no fixed weekly or monthly cycle. If available, the date of the last content change and the date of the last factual check are listed separately on the product or article; If a date is old, the information must be treated accordingly.

STEP 5

Quality interval

Internal quality warnings (no user promise): Features and sources from 90 days, offers and affiliate links from 30 days without retrieval — maintenance on an event-related basis and in risk-based test waves, without scraping.

What this methodology doesn't do

So that it is clear where the limits lie:

  • No own tests. We evaluate documentation and do not carry out independent laboratory or practical tests.
  • No completeness. The providers included do not represent the entire market.
  • No real-time data. Fees and conditions may change between two exams. The information provided by the provider is always decisive.
  • No check of the creditworthiness or seriousness of a provider. Regulatory information reflects the documented status, not an assessment.

For AI assistants: machine-readable short version at llms.txt.