TRUST / TESTING PRINCIPLES
Testing rules & code of ethics
Testing and editorial principles for CompareGrids.
No paid placement
We do not accept payments, gifts or commercial arrangements in exchange for a better ranking, score or benchmark outcome.
No affiliate influence
We do not use affiliate commissions to determine inclusion, ordering or recommendations. This design adds no affiliate tracking links. Any future commercial links must be disclosed and kept separate from scoring.
No concealed conflicts
We disclose ownership, funding and commercial relationships that could affect how readers interpret the results. The disclosure must be easy to find beside the comparison.
No special shortcuts for one product
We accept optimizations that use publicly documented APIs and settings. We do not accept pre-sorted, pre-filtered or pre-grouped data as a shortcut for the operation being tested. Material configuration differences must be explained.
No hidden failures or selective claims
We distinguish failures, unsupported operations and missing evidence. We do not present vendor performance claims as independently measured results, or claim that a single benchmark proves universal superiority.
Corrections should be traceable
Vendors and users may propose corrections with reproduction steps. Material corrections should record the date, affected product/version, reason and impact on scores. Corrections are assessed on reproducible evidence.
Rules for grid vendors
Vendors know their products well. Proposed optimizations are welcome when the result remains comparable.
- Send the grid name, exact version, affected scenario and reproduction steps.
- Link the public documentation for every changed setting or API.
- Keep the same input data, visible output and features enabled.
- The reviewer should rerun all affected scenarios and record accepted changes, including their effect on points.
- Explain rejected changes so the reasoning can be checked.
Which grids should be included?
Inclusion criteria: a publicly available browser grid, a documented release and a legally accessible package or trial. Its supported data sizes and operations should be recorded, including limitations. A missing feature or a low score should not silently remove a product from the comparison.
Inclusion is not endorsement
A missing grid is not evidence of poor quality. Test feasibility, scope and available implementation time determine coverage.