METHODOLOGY · VERSION 0.1

A ranking you can
inspect and challenge.

E-commerce Index separates directory coverage, technical observation and commercial popularity. Today’s score measures a small set of public website signals. It is not a traffic, revenue or customer-satisfaction estimate.

Published15 July 2026StatusEarly public methodology

The essential distinctionBaseline means an editorial seed value awaiting measurement.Observed means our crawler measured the public homepage.Verified will mean selected facts were confirmed by an authorized business representative.

Technical score v0.1 is a 0–100 homepage health score. Once a website is observed, its earlier baseline is replaced with the following reproducible calculation.

35Homepage availabilitySuccessful public response

15HTTPSEncrypted connection available

20Response timeUp to 20 points; one point removed per 250 ms

10Page titleNon-empty HTML title detected

10Meta descriptionPublic description metadata detected

10Commerce schemaProduct, Offer or ItemList structured data detected in homepage HTML

score = availability + HTTPS + speed + title + description + commerce schema

Speed points are calculated as max(0, 20 − floor(response_ms ÷ 250)). Scores are capped at 100. Ties are ordered alphabetically.

Current limitation

One server-side homepage request cannot describe checkout quality, mobile usability, delivery experience or business popularity. A “No” for commerce schema means it was not detected on the homepage; it does not prove that individual product pages lack Product markup. It is useful as a technical observation, but it is not yet a complete commerce index.

One global measurement standard

The active scoring formula is applied consistently across country indexes. Local categories and market context can differ, but a technical signal has the same definition and weight wherever it is observed.

Public evidence only

Our crawler requests the public HTTPS homepage, follows redirects, reads at most 512 KB of HTML and stops after a bounded timeout. It does not sign in, add products to a cart, place orders or attempt to bypass access controls. Our separate discovery classifier places potential stores into an internal review queue; discovery does not automatically create a listing or ranking. Read the crawler transparency policy.

Versioned observations

Each observation stores the time, response status, response duration, detected metadata and the score produced by that methodology version. A refreshed observation can move a score in either direction.

Coverage before confidence

Global methodology does not imply a single worldwide leaderboard. Rankings are published within countries and relevant categories only when coverage is broad enough. We disclose tracked and observed counts, and retain the prototype label while most entries remain baselines.

A category describes the store’s dominant commercial purpose, not every product it happens to sell. Each listing receives one primary category for country-directory navigation and category comparisons.

Read the store’s identity

We give the strongest weight to the page title and the store’s own meta description because these usually state its primary retail proposition.

Inspect navigation and headings

Repeated category language in menus and prominent headings provides stronger evidence than an isolated product card.

Measure catalogue signals

Product and collection terms across the public homepage support the decision, but their contribution is capped so one item cannot determine the entire store category.

Compare weighted evidence

All supported categories are scored. We select the leading category only when it clears a minimum evidence threshold and has a meaningful lead over the runner-up.

Hold ambiguous cases

Mixed or weak evidence is labelled Other internally and kept for editorial review rather than being auto-published into an unreliable category.

Accept documented corrections

Owners, managers and employees can challenge a category using a verified business-domain email. Editorial corrections retain an audit trail.

Category is separate from ranking

A category changes where a store appears in the directory; it does not add or remove ranking points. Commercial relationships and paid services do not influence category assignment.

Read the store inclusion criteria →

04 · RESEARCH DIRECTION

Commerce Index v0.2

This is a proposed framework, not the formula currently used on store profiles. We will validate weights against collected data and publish changes before activating them.

25%

Reliability & safety

Availability over time, HTTPS behavior and carefully selected public security controls.

20%

Real-user experience

Mobile LCP, INP and CLS from eligible Chrome UX Report data, with lab data labelled separately.

20%

Commerce readiness

Product discoverability, working catalogue signals, structured product data and transparent policies.

15%

Discoverability

Indexability, metadata, canonicals, sitemaps and durable category/product information architecture.

10%

Operational trust

Visible contact, returns, delivery and privacy information—not subjective reputation claims.

10%

Data quality

Observation freshness, repeatability, coverage and owner-verified facts, with explicit confidence.

Popularity will remain separate.

Traffic or market-reach rankings require sufficiently representative, country-level evidence. We will not reverse-engineer visits from technical signals or mix self-reported analytics with unverified estimates. If introduced, popularity will be a separate dimension with its own confidence and methodology.

Commercial customers, affiliates and perk partners receive no ranking advantage.
Methodology changes receive a version number and publication date.
Material corrections retain an audit trail rather than silently rewriting evidence.
Websites may request a refresh or factual correction, but cannot purchase a score.
Missing data is shown as missing; it is not silently converted into a positive result.
Country and category comparisons use the same active formula and observation window.

Challenge the methodology →

06 · RESEARCH REFERENCES