Sources - RoboVac News

Sources

Source	Language	Score label	Scale max	Scale note
Vacuum Wars	en	Vacuum Wars Stars	5	Read from the 'Vacuum Wars Stars X.XX' line under the product name in every review (e.g. 'Vacuum Wars Stars 0.89'), also shown as a 0-5 star glyph ('★ 0.89') in the same spot.
Just A Dad Approved	en	3-Month Real World Test Score	10	Read from the scorecard headed '🏆 3-Month Real World Test Score X.X /10', whose 'Score Summary X.X' total is the overall figure; the five category bars above it (Hassle Free, Mop Performance, etc.) are sub-scores, not used here. Explicit /10 scale stated in the heading.
Basic Tutorials	en	Basic Tutorials Score	100	Read from the Conclusion card's large number (e.g. '91') shown beside a 5-star graphic filled to that percent. Not stated in the visible text, but the archived page's own CSS makes the scale explicit: '.bt-review-score-number::after { content: "/100"; }'. The 'Detailed Rating' category bars (Suction Performance, Navigation, ...) above it are sub-scores, not used here.
SmarthomeAssistent	de	Gesamt Performance (6 − grade)	5	Read from the 'Gesamt Performance X,X · <word>' line in each review's results card (e.g. 'Gesamt Performance 2,9 · Befriedigend'). The word bands (Sehr Gut/Gut/Befriedigend/Ausreichend/Mangelhaft/Ungenügend) confirm this is the standard German school-grade scale, 1 (best) to 6 (worst) -- LOWER is better, opposite every other source here. The page's own schema.org Review JSON-LD confirms the scale by computing ratingValue = 6 - grade with bestRating 5 (grade 2.9 -> ratingValue 3.1). We store that same (6 - grade) transform as 'raw' on a 0-5 higher-is-better scale so normalising to /10 is comparable with every other source; the literal printed grade is quoted in each review's score.evidence. 'Note Saugen'/'Note Wischen'/etc. are per-category sub-grades, not used here. The 3 preview_only pages are pre-release news with no grade.

How we aggregate

Every review we track reports a score somewhere — stars, a percentage, a grade out of 5, or a German school grade (where lower is better). We convert each one to a common /10 scale ("normalized"), preserving the publisher's own raw score and scale alongside it so you can check our math. A model's House Recommendation score is the mean of every review's normalized /10 score, one vote per source. Reviews with no numeric score (e.g. preview coverage) are shown but excluded from the average. This is separate from Owner Feedback, which will reflect real reader/owner reports once there are enough of them (5+) to be meaningful — not our own aggregate.