Quantdiv was built to compare the best value between competing products, category by category. The idea is simple: with every purchase, you want to maximize what you get per dollar spent.
Not necessarily the cheapest — cheap often costs more later. Not automatically the most expensive — a premium price tag isn't the same as premium value. What most people actually want is the product that delivers the most quality for each dollar — and to see how that value stacks up against every competitor on the shelf.
One "editor's pick" buries what you care about under a single opinion. Instead, we assembled the raw specs from manufacturer sheets and standards bodies, published the exact scoring formula, and put the weightings in your hands. Drag a slider, and the ranking recomputes live around what you care about.
No black box. No paid placements changing the math. The score runs on published specs, not payouts — entirely unbiased Quantdiv.
Every spec is checked against manufacturer sheets and standards bodies (AHAM, DOE, ENERGY STAR, NSF), and we keep adding categories as we go. Want one covered next, or spotted a number that looks off? Tell us on the feedback page — your input directly shapes what we build and re-check.
Every score is quality normalized to price — the most you get per dollar, ranked against the field.
Slide what matters to you. The ranking is yours, not our opinion — the tool recomputes it live.
The formula is published on every page. Affiliate links never change the math.
No black box, no editor's pick. Every ranking on this site is produced by the same published formula, running on specs verified against manufacturer sheets and standards bodies. Here's exactly how it works — including where judgment enters and where it doesn't.
Value Score = ( Σ weightᵢ × normalize(specᵢ) ) ÷ (price ÷ category-average price)
Each spec is scored relative to the other products in the category — the weakest model on that spec sits at the bottom, the strongest at the top, everyone else in between, so a score means "how this model compares to its competitors," not an absolute grade. (This also lets us fairly combine specs on totally different scales — a rating in the hundreds, a cost in dollars, a yes/no feature.) We multiply each normalized spec by its weight, sum them, then divide by price-relative-to-average — which turns raw quality into value. A great-but-overpriced unit ranks below a good one at the average price. That's the whole point: value, not price, and not raw quality either.
This is the part most comparison sites hide. The weights you see on load aren't our guess about what's "best." They're computed from the data itself, in three steps:
For every spec we compute its coefficient of variation (standard deviation ÷ mean) across all products in the category. This is scale-independent, so a spec measured in the hundreds (CADR) is comparable to a true/false spec (HEPA). The logic: a spec where every product scores about the same can't tell winners from losers, so it earns little weight; a spec that varies widely is a real differentiator, so it earns more.
Each spec's weight = its share of the total variation across all specs, scaled to sum to 100. Pure information content — no thumb on the scale.
Pure variance has a known blind spot: it can crush a spec that's genuinely important but happens to be similar across products (e.g. True HEPA — arguably a must-have, but low variance because most units have it). So we set a floor: no spec is weighted below 30% of its equal share (equal share = 100 ÷ number of specs). Anything below the floor is lifted to it; the rest are rescaled to fill the remainder. This floor is the only judgment input in the default weights, and we state it openly. The k = 0.30 ratio is fixed and identical across every category.
Twelve units, five specs. Here's what the formula actually decided, and why:
| Spec | Variation (CoV) | Default weight |
|---|---|---|
| Room coverage | 0.66 (widest — 220 to 1,560 sq ft) | 41 |
| True HEPA | 0.30 | 18 |
| Smoke CADR | 0.24 | 15 |
| Filter cost / yr | 0.25 | 15 |
| Quiet (low dB) | 0.18 (narrowest) | 11 |
Room coverage carries the most weight because it genuinely separates these units — not because we decided it should. Notably, this overruled our own instinct: we'd have hand-weighted Smoke CADR highest, but the data showed these units cluster fairly close on CADR, so it earns moderate weight instead. That's the method working as intended — the math corrects the narrative.
Tapping a use-case chip re-weights the sliders for a specific scenario. Unlike the defaults, these do carry domain judgment — “for pets” emphasizes running cost because pet households run purifiers around the clock, which is a fact about pets, not something derivable from the spec spread alone. We're upgrading these to carry explicit, sourced rationale over time. Either way, every preset is just a starting point: drag any slider and the ranking becomes entirely yours.
Reproducible: the weighting engine is a single documented script; the same method runs on every category. Spot something off? Tell us here.