Methodology

How we source and verify

Every specification, price, and compatibility claim on Setup Gear Guide is sourced to a citable page (manufacturer first). We publish those sources and flag anything we could not verify instead of guessing. Recommendations come from that sourced data plus deterministic scoring — consistent, inspectable, and computed the same way for every product.

Sourcing & verification

Every product enters the catalog through a research pipeline with hard gates: at least one product-level source is required, and every compatibility-critical specification (a socket, a lens mount, a quick-release standard, a wattage) must carry either a field-level source or an explicit 'needs verification' flag. The import fails otherwise — there is no path for an unsourced critical spec to reach the site silently.

Products with flagged fields display 'partially verified'. A flag means we could not pin that value to a primary source yet — it does not mean the value is wrong, and we show the flag rather than hiding the uncertainty. Source links, their type (manufacturer, retailer, review), confidence level, and retrieval date are published on every product page.

Products whose existence or current retail form we could not verify are excluded entirely, even when widely rumored.

Scores

All eight scores (overall, value, compatibility, upgrade path, use-case fit, beginner friendliness, pro readiness, confidence) are computed deterministically by code from sourced data — there is no live AI generation and no pay-for-placement of any kind. Value reflects price position within the product's own (category, tier) cohort and is expressed on that one consistent basis everywhere (cheaper-for-its-class scores higher, so a flagship isn't punished for flagship pricing); where a category's pricing is supply-volatile (the 2026 RAM/SSD/GPU situation), value is not scored rather than reading a misleading percentile off an abnormal price. Compatibility rewards native ecosystem fit and penalizes adapter-only reaches; beginner-friendliness and pro-readiness start from the product's tier and adjust for in-class price position, source coverage, and verification state. Confidence rewards independent primary sources and saturates — beyond about three solid sources extra citations add little, so a wall of weak links can't inflate it. Each score shows its own reasoning string on the product page.

The overall score is a weighted blend — tier caliber 30%, value 25%, confidence 20%, compatibility 15%, upgrade path 10% — renormalized over the dimensions that actually apply to a product (a category with no ecosystem dimension, or a product whose price can't be positioned, drops that term rather than diluting the headline with a flat 50). 'Tier caliber' is the capability signal: a product's tier carries its performance weight, nudged slightly by where the product sits within its tier's price range so a brand's flagship out-ranks its own cheaper sibling on the headline rather than the reverse. The value term that feeds the blend is compressed (the displayed Value sub-score keeps its full range), so genuine bargains are still rewarded without cheapness alone overturning a flagship. Use-case fit is deliberately NOT in the blend and is no longer shown on product pages: per-use-case scoring needs product-to-use-case data we have not built yet, so rather than display a placeholder number we exclude it entirely.

Compatibility engine

Compatibility rules are typed checks (socket match, ecosystem overlap, clearance, wattage headroom) parameterized by rule instances with declared sources and confidence. When a needed spec is missing the engine reports 'couldn't verify' at informational severity — unknown never blocks a build and never masquerades as a verified pass or fail.

Inference-based rules (judgment calls rather than manufacturer documentation) are flagged as such and capped at medium confidence.

Prices

Live-checked prices are point-in-time snapshots shown with their check date. Where we don't have a live price for a retailer (for example, retailers whose prices we're not yet permitted to display), the product page shows the manufacturer's list price, labeled as such — no check date is implied. 2026's components market is volatile (the RAM/SSD supply crisis especially) — where a price is far above its historical normal, we say so in the product notes rather than presenting it as ordinary.

The honesty guard is symmetric: prices that are implausibly far below a product's tier-appropriate range are treated as likely-stale or likely-wrong and flagged 'unusually low — verify' rather than celebrated as a deal, exactly as prices far above normal are flagged. An outlier in either direction is a signal to re-check the source, not to trust the number.

Affiliate independence

Some outbound links may become affiliate links. Commission never influences rankings, scores, compatibility results, or recommendations — the engines cannot see affiliate status as an input. See the affiliate disclosure page for specifics.

Hands-on & owned gear

Some products are gear the editor personally owns and uses. Those carry a first-hand "In our studio" section — the owner's real photos, a named byline, and a candid write-up of how the gear performs in daily use, including its limitations and tradeoffs.

This is an experience and trust signal only. Ownership is never an input to scores, rankings, compatibility results, or recommendations — the same firewall that keeps affiliate status out of the engines. A product the editor owns is scored by the identical deterministic code as one nobody here has touched.

The bar is real ownership: only gear genuinely owned and used is marked hands-on, the photos are the owner's own (never a stand-in for the catalog product image), and where a piece of owned gear isn't in the catalog yet it is named without a fabricated product link. The owner's full rigs live under Our Setups.

Verification status & data quality

We publish the numbers rather than asking you to trust an adjective. The data-quality page shows live counters generated from the catalog: how many products are tracked, how many spec fields carry a field-level source versus a verification flag, total source citations, and when the catalog was last updated.

Verification is uneven by design — fully-specced categories like sim-racing wheelbases are close to complete, while volatile categories (RAM, SSDs during the 2026 supply crisis) carry more flagged fields. The flag is the honest signal; chase the counters on the data-quality page for the current picture.

Citing Setup Gear Guide

Free to read, cite, and link — please attribute with a link to the canonical page. Preferred citation format: "Setup Gear Guide — <page title>, <canonical URL> (researched <date>)." The public API returns the canonicalUrl with every record so machine consumers can cite consistently; agent-specific guidance lives on the For AI agents page.

Changelog

Material changes to scoring, sourcing gates, or weights are logged here so the methodology has a history, not just a 'last updated' date. Catalog and content activity (products added, specs researched) is summarized on the changelog page.

June 2026 — Scoring audit pass: corrected tier mis-assignments where a pricier flagship had been tagged below its cheaper sibling; made the upgrade-path tier curve monotonic (a flagship has the most forward headroom, not the least); folded a small in-tier price-rank nudge into tier caliber and compressed the value term that feeds the overall blend, so a cheaper sibling can no longer out-rank a flagship on price alone; the overall blend now renormalizes over applicable dimensions instead of injecting a flat 50 for value (supply-volatile or unpriced) or compatibility (no ecosystem dimension); capped the verification-flag penalty; and removed the Use-Case Fit row from product pages.

June 2026 — Use-Case Fit excluded from the overall blend (shown as a labeled neutral baseline) rather than letting a placeholder move rankings; tier caliber set to 30% of the overall score. Product JSON-LD no longer asserts stock availability we can't verify. Per-field verification counts and an 'effectively tied' guard for near-identical comparison scores added to product and comparison pages.

June 2026 — Scoring refinements: Value normalized to a single (category, tier) basis with consistent wording, and not scored at all where category pricing is supply-volatile; Confidence reworked to reward independent primary sources and to saturate after roughly three solid sources so extra weak citations no longer inflate it.