How we verify data
Every specification in the catalog carries per-field provenance: we know where each number came from, and we show you. Nothing is scraped and forgotten — specs are extracted from manufacturer documentation, normalized into one schema, and cross-checked against independent lab measurements where they exist.
The compatibility engine runs the same checks a good dealer would — speaker load against amplifier rating, amplifier power against speaker sensitivity, projector throw against screen material — and explains every flag in plain English. Departments only open once their data meets this bar — the standard all nine now clear. Seating was the most recent to open: it lists the models with published fit-and-safety specs, shows an honest “unstated” where a maker goes quiet, and holds back models whose published specs fall too far short until the numbers exist.
No sponsored placements, ever. Retail links exist so you can buy the thing — some may be affiliate links where an affiliate relationship exists. We may earn a commission at no cost to you; it never affects rankings or compatibility verdicts — those come from the data.
Every value carries its context
Two sources can publish the same field under completely different conditions — amplifier power at different distortion levels or with a different number of channels driven, frequency response at different tolerances, speaker sensitivity measured in-room versus anechoic. We treat value + context as an atomic pair: when a value arrives without its context, we store the context as “unstated” rather than guessing or silently dropping it, and a validation stage enforces those companion fields instead of leaving them to chance. A side effect of storing the silence is that we can count it — the Spec-Sheet Honesty Report tallies, brand by brand, what manufacturers don’t say.
Three-state booleans: yes, no, and unknown
A feature is either present, absent, or simply not stated — three states, not two. A retailer that never mentions a phono input hasn’t told you the unit lacks one, so an unstated feature is recorded as unknown, never quietly defaulted to “no.” That is why you’ll sometimes see “unknown” where other sites confidently show an X. It is the honest answer, and it keeps the data from inventing facts.
Catching conflicts and variants
When two sources disagree about the same product beyond a set tolerance, a cross-source conflict detector routes the value to human review rather than picking a winner automatically. Some disagreements aren’t errors at all: when the same model number genuinely ships with different internals by production lot — the panel-lottery TVs are the classic case — we flag it as a variant, blank the conflicting field, and note the variance instead of thrashing between values. The goal throughout is to fail loudly and visibly, never to smooth over a discrepancy.
Where the data comes from
Specifications are drawn from manufacturer spec-sheet PDFs (preferred, because their footnotes carry the measurement conditions marketing pages omit), manufacturer product pages, and retailer and affiliate feeds. Prices and availability are gathered from public retailer and shopping listings — Google Shopping results and retailer product pages — refreshed on a schedule and tied to each product’s most recent fetch. Prices are indicative rather than a live checkout quote and can lag the retailer.
Honest limitations
We don’t claim to be perfect. Coverage grows category by category and not every product carries every field yet. Prices lag the retailer by however long since the last fetch, so always confirm the final price and availability on the retailer’s own page. A clean compatibility result rules out the common mechanical mistakes, but it can’t judge how a system will sound or look in your room — treat it as confidence to dig deeper, not the last word. And where we genuinely don’t know something, we’ll tell you we don’t know rather than fill the gap with a guess.