Methodology
How Best Robot Match scores every robot on the site — and what those scores do and do not represent.
What the scores are
Every score is an editorial score built from public, verifiable data: manufacturer specification sheets, warranty and support documentation, published parts pricing, SDK and API documentation, app-store listings, and public reliability reporting. We do not claim first-person lab testing. Where a claim can't be verified from public data, it doesn't move a score.
The eight factors (weights locked at v1.0)
| Factor | Weight | What it measures |
|---|---|---|
| Capability | 20% | What the robot actually does versus what the marketing implies |
| Reliability & Longevity | 20% | Build-quality reporting, manufacturer stability, server-dependency risk |
| Value | 15% | Price against capability, including subscriptions and true cost of ownership |
| Support & Warranty | 15% | Warranty terms, support responsiveness, delivery and returns reality |
| Repairability & Parts | 10% | Parts availability, part costs, ease of fix, documentation |
| Software & Updates | 10% | App quality, update cadence, SDK/API openness |
| Ecosystem | 5% | Connectivity, smart-home integration, community strength |
| Privacy & Safety | 5% | Camera/microphone data policy, certifications, data minimisation |
The overall score is the weighted sum, rounded. Weights are versioned — any change is logged on the ethics page.
The server-shutdown risk rating
Our signature rating. Consumer robots have a unique failure mode: when the manufacturer shuts its servers, cloud-dependent robots can stop working entirely. This has already happened — Anki's Vector, Jibo, and Embodied's Moxie. Every robot is rated on two axes:
- Server dependency — works offline (no cloud account needed for core functions), degrades (works, but loses features), or bricks (non-functional without the maker's servers).
- Maker stability — an editorial judgement of the manufacturer's longevity based on funding, track record, and install base.
The BestMatch Longevity Grade (A–F)
The server-shutdown axes above feed a single letter grade, so it can be cited at a glance. Formula v1.0 combines four evidenced components, each scored 0–20 (maximum 80), then normalised to a percentage:
| Component | How it is scored |
|---|---|
| Server dependence | works offline = 20 · degrades = 12 · bricks = 4 |
| Subscription dependence | no required subscription = 20 · has one = 8 |
| Company stability | low risk = 20 · medium = 12 · high = 4 |
| Repairability | repairability score ÷ 100 × 20 |
Grade bands: A ≥85% · B 70–84% · C 55–69% · D 40–54% · F below 40%. The exact per-robot component values are shown on each robot's profile page, and the calculation lives in data/scoring.js (function longevityGrade).
True 3-Year Cost
The sticker price hides the real cost of robots that charge a subscription. We publish a True 3-Year Cost on every profile:
This exposes cases where the subscription dwarfs the hardware — ElliQ 3, at $249 up front, reaches roughly $1,690 over three years once its $39.99/month plan is counted.
Evidence confidence
Factual figures — price, subscription, availability, server dependency, company status — are individually classified by how well we can source them: Verified (official source), Claim (manufacturer's own figure), Reporting (credible third party), Inference (editorial estimate) or Unknown (shown as such, never guessed). Every figure and its source is recorded in the public evidence ledger, which is updated in the same change as any data edit.
Categories
The main comparison covers companion, quadruped and humanoid robots. Robot vacuums and robot mowers are scored with the same eight factors but ranked in their own tables — they're mature appliance categories with different buyer expectations.
Discontinued products
Discontinued robots (like Moxie) stay listed, greyed out, with no purchase link. They exist as buyer warnings and evidence for the shutdown-risk rating — removing them would erase the category's most important lesson.
Commercial independence
No affiliate programs are currently active. Every outbound link goes to the official manufacturer site. If affiliate partnerships are added later they will be disclosed on every page that uses them, and they will never influence scores or rankings. Full policy on the ethics page.