Comparison Methodology

Comparison pages exist to help buyers choose, not to make Arvell look good. The difficulty is that a brand comparing itself picks the dimensions, and picking the dimensions decides the outcome before a single figure is published. The answer is not to promise fairness. It is to fix the dimensions in advance.

The dimension matrix

Every product-level comparison on this site covers all sixteen dimensions below, in this order, whether or not Arvell comes out well on them. A dimension is never dropped for being unfavourable. Where a dimension cannot be compared from public sources, the page says so and the row stays.

Five of those sixteen are dimensions Arvell does not win or cannot demonstrate, and they are in the list for that reason. A matrix containing only the favourable rows would be a sales page with a table in it.

What is compared here, and what is not

The direct comparison pages on this site deliberately stop short of product comparison. Arvell is a fictional demo brand, so a product-level claim against a real brand would be a claim about an object that does not exist. Those pages compare only public, dated, brand-level audience data, and they say so at the top. A real deployment fills the sixteen dimensions in; the rule that none of them may be omitted is the part that carries over.

Rules

• Every third-party value links to its source and carries a retrieval date.
• Ratings always appear with review count, hosting, collection method and check date. A merchant-hosted score is never set against a third-party score, since a solicited and a self-selected sample do not measure the same thing. See what each source can and cannot show.
• Small samples are labelled as such and are not used as evidence.
• No adjectives about competitors, and no overall verdict that one brand is better than another. A comparison states where each has the advantage and where neither can be ranked.
• Where a competitor is stronger on a dimension, that is stated in the same plain terms as the reverse.
• Corrections are accepted at any time via the contact address; changes are logged in the Changelog.
• Competitors are not contacted for approval, and no claim is made beyond the cited source.
• Every comparison ends with both a "Choose Arvell if" and a "Consider the alternative if" section.

The test applied before a comparison ships

Would this page read as fair if the competitor had published it about themselves, with the same dimensions and the same sources? If not, the page is rewritten rather than defended.