How this index is built

Methodology

PURRINDEX is a specification index, not a review site. We do not buy, test, or photograph litter boxes. We read what manufacturers publish, record where each value came from, and say so when they publish nothing.

What PURRINDEX does not do

  • We do not test products. No lab, no long-term trials, no scoring of performance.
  • We do not rate safety. We list the mechanisms a manufacturer documents. We never apply comparative safety adjectives to a product, and we never turn a count of documented mechanisms into a score.
  • We do not infer. If a manual does not mention tofu litter, the cell says the manufacturer does not state it — not "probably fine".
  • We do not show live retail prices or retailer ratings. Where a price appears it is the manufacturer's own published figure with a capture date; otherwise you get a band.
  • We do not reproduce manufacturer photography. Every drawing on this site is generated by us from published dimensions.

Source hierarchy

When two sources disagree, the higher-ranked one controls.

ClassMeaning Weight
M1Manufacturer manualEngineering/instruction document. Outranks marketing pages.
M2Manufacturer product pageBrand's own published specification.
M3Manufacturer support/FAQBrand's own support documentation.
R1Retailer-declared attributeDeclared by the seller in a retail listing, not by the manufacturer.
SSite probe / brand-site checkObserved state of a brand's own web presence.

Evidence states

StateWhat it means
ManualThe value appears in the manufacturer's own manual or instruction document. This outranks everything else.
ManufacturerThe value appears on the manufacturer's own product or support page.
Retailer-declaredThe value was declared by a seller in a retail listing, not by the manufacturer. Treated as weaker evidence and labelled as such.
ConflictingTwo sources disagree. Both values are shown, a ruling is recorded, and the higher-ranked source controls. Where the gap exceeds the manufacturer's own stated tolerance, nothing is published.
Not statedNo manufacturer document states this value. Not an omission on our part.

Recorded conflicts

11 conflicts between sources were found and ruled on. Two were ruled unpublishable: where the disagreement exceeded the manufacturer's own stated tolerance, PURRINDEX publishes neither value.

ProductField RulingControlling source
Whisker Litter-Robot 5 / 5 Proexterior heightPublish 29.50 in; show bothManual (M1)
Whisker Litter-Robot 5 / 5 Prointerior headroomPublish the manual's range; show the page figure as the manufacturer's single-point marketing valueManual (M1)
Whisker Litter-Robot EVOproduct weightPublish 21.02 lbsManual (M1)
PETKIT Purobot Max 3exterior dimensionsDO NOT PUBLISH as a comparison value; show both with the conflict statedNeither — unresolved
PetSafe ScoopFree SmartSpinlitter compatibilityPublish the manual's exclusion; treat the page's blanket claim as marketingManual (M1)
PetSafe ScoopFree OpenSkylitter compatibilityPublish the manual's exclusionsManual (M1)
Whisker Litter-Robot 5 vs 5 Procrystal litter compatibilityResolved by the manual: crystal beads are acceptable if small enough to pass the screen, with a sensor-accuracy warning for translucent or white litterManual (M1)
HomerunPet CS106noise levelDO NOT PUBLISHNone — value discarded
CATLINK Scooper Open-Xexterior dimensionsDO NOT PUBLISH as a comparison valueNeither — unresolved
PETKIT Purobot Max 3noise levelDO NOT PUBLISHNone — value discarded
OEM family Cumrige / Fumoi / Paruuntysmaximum cat weightDO NOT PUBLISH a family-level figure; publish per listing with the conflict statedNeither — unresolved

Coverage, stated honestly

Across 44 products and 30 tracked fields, 34% of cells are documented from some source and 25% from the manufacturer's own document. The documented tier reaches 46%; the budget tier reaches 13%, with 0.2% primary-source coverage.

That budget number is not a research failure. It is the most important finding in this index: for the products that account for roughly half the units sold, there is essentially no manufacturer documentation to read.

Editorial independence

Product fields are partitioned in code. 49 editorial fields may inform ranking, comparison and recommendation. 13 commercial fields — commission rate, cookie duration, merchant, network, payout estimate, EPC, merchant priority, deep link, tracking id, programme AOV, reversal rate and creative licence — may not. The two sets are asserted disjoint when the module loads.

Three checks run on every build:

  • Static. The ranking module's syntax tree is scanned for any reference to a commercial field name.
  • Runtime. Ranking runs against a read-only projection that raises an exception on a commercial read rather than quietly returning nothing.
  • Permutation. Commission, cookie, merchant and payout values are shuffled across all products 200 times; the ranking must come out byte-identical every time. That is 1,600 rankings per build.

The audit was verified against a deliberately injected leak, which all three checks caught. The test can fail, so its passing means something. A build that fails it does not ship.

Price is deliberately an editorial field: what a product costs is a property a buyer ranks on. What we might earn from that price is not.

Affiliate disclosure

PURRINDEX intends to earn commissions from qualifying purchases made through retailer links. No such link is live in this build. Merchant destinations are resolved only after a product has been selected editorially, and the commission attached to a destination is invisible to the code that did the selecting.

We have joined no manufacturer affiliate programme. Programme terms recorded in this index were read from each merchant's own published page and are stated as of 2026-09-01. Acceptance is assumed nowhere, and no revenue projection depends on it.

Known limits

  • Specifications were captured on 2026-09-01. Manufacturers change pages without notice.
  • Two manufacturers' manuals (CATLINK, Neakasa's own-site PDF) use embedded font encodings that could not be machine-read. They are archived and await human transcription.
  • Entry-height and cat-fit comparisons are deliberately not offered: coverage is too thin to compare fairly.
  • Ownership-cost modelling is not offered: consumable prices and replacement intervals have not been collected.
  • This index covers the United States market.