How the Research Desk tests seller tools

Tools are compared on common inputs and defined seller decisions. The protocol records account plan, collection window, task steps, outputs, exports, errors, and limitations before a verdict is written.

Vendor documentation is a source for what a product claims and how it describes a metric. It is not evidence that the claim holds in use.

The Divergence Report uses a fixed set of 50 Etsy-oriented keyword strings captured across eRank, EverBee, Alura, and Sale Samurai during a single 48-hour window on 2026-07-14/15.

Every displayed value is retained verbatim. Exact and K/M-rounded values enter spread calculations; reporting floors such as <20 and literal NA values remain visible but are excluded from ratio math. No tool is treated as a reference truth for Etsy’s private query logs.

Vendor pricing and feature pages are captured with the public URL, access date, currency, billing interval, and the precise claim used. A source details states what the vendor documents; it does not convert the claim into an independent finding.

If a public source cannot be accessed, the field stays unpublished. Search snippets, affiliate summaries, and third-party reviews do not fill the gap.

Criteria are defined before rating. Missing independent observations withhold the score rather than lowering it to an invented midpoint. When scoring begins, weighting, inputs, errors, and confidence will be published beside the result.

Commercial relationships, affiliate rates, and vendor access do not enter the rating model. Official-source review tiers therefore show structure and seller fit but no performance winner.

Product screenshots are published only when captured during an authorized test session or supplied through an approved reusable source. Each image must identify the product area, capture date, plan context, and any redaction.

The publication uses no product-interface screenshots because independent account sessions have not been completed.

Pricing, plan limits, interfaces, exports, and metric definitions are versioned in the data layer. Approved changes update affected pages and produce a dated changelog record with source and affected routes.

Until source details and independent observations are complete, affected product pages remain outside the sitemap through their page-level evidence gates.