A sentence about a named company is a claim about that company. The corpus behind this study was read weeks before publication, and a certifier's directory moves: a model gets added, a certification completes, a search that never executed comes back looking exactly like a search that found nothing. So every brand named below had its listing re-read on 2026-09-20, and the read is published with it — the URL, the status, the bytes on the wire and what the page said. A brand that could not be re-read is not named.
One of the five entries below corrects this study rather than a manufacturer — Waterdrop. That is not balance for its own sake; it is what the re-reads found.
iSpring — RCC7AK
Evidence grade: listing-read. Names queried: iSpring Water Systems LLC (company id C0536866); iSpring.
NSF's listing for iSpring Water Systems LLC carries the RCC7AK under NSF/ANSI 58 for TDS Reduction and nothing else, and a manufacturer-name search across every standard returns the same single standard — iSpring holds no NSF/ANSI 42, 53 or 401 listing. Mercury and microplastics, both marketed for this model, appear on no iSpring row at any model. Fluoride does appear on iSpring rows, but not on this one: it is carried by the RCC7AK-BN, RCC7AK-BLK and the two UV-faucet siblings, which the listing certifies out of a different manufacturing facility from the plain RCC7AK. Those are separate certifications of separate products, not finishes of this one, so the fluoride the RCC7AK's own product page markets is not backed for the RCC7AK.
What this does not establish: This is a read of NSF only. IAPMO and WQA returned no listing for the RCC7AK when the corpus was built on 2026-08-30 and were not re-executed today, so no absence is asserted against them here, and none of these reads can speak for the accredited bodies this study does not index. One complication is recorded rather than resolved: the record notes that the Amazon destination it opened on 2026-08-30 resolved to the separately named RCC7AK-BN, which is why it assigns this product no ASIN at all. That observation is more than a month old, it was not re-executed today, and the claims scored here were read from iSpring's own RCC7AK page — so it cannot move a verdict about the RCC7AK, and it is published here as the open question it is.
Rows read today, in the certifier's own wording
- RCC7AK — TDS Reduction
- RCC7AK-UV — TDS Reduction
- RCC7AK-BN — Asbestos Reduction, Barium Reduction, Cadmium Reduction, Chromium (Trivalent) Reduction, Copper Reduction, Fluoride Reduction, Lead Reduction, Selenium Reduction, TDS Reduction
- RCC7AK-BLK — Asbestos Reduction, Barium Reduction, Cadmium Reduction, Chromium (Trivalent) Reduction, Copper Reduction, Fluoride Reduction, Lead Reduction, Selenium Reduction, TDS Reduction
Directory reads behind the iSpring finding| Directory | Query | Read on | HTTP | Bytes | What it said |
|---|
| NSF | company id C0536866, NSF/ANSI 58 | 2026-10-02 | 200 | 20,586 | model rows returned: 14; rows under the facility that certifies the RCC7AK: 3; rows reading "Fluoride Reduction", all under a different facility: 7; rows reading "Mercury Reduction": 0; rows naming microplastics: 0 |
| NSF | company id C0536866, NSF/ANSI 42 | 2026-10-02 | 200 | 6,827 | No Matching Products Found |
| NSF | company id C0536866, NSF/ANSI 53 | 2026-10-02 | 200 | 6,827 | No Matching Products Found |
| NSF | company id C0536866, NSF/ANSI 401 | 2026-10-02 | 200 | 6,827 | No Matching Products Found |
| NSF | manufacturer name "iSpring", every standard | 2026-10-02 | 200 | 20,566 | standards returned: 1; model rows returned: 14 |
The byte count is the control. A search that matched nothing and a search that never ran look identical in prose; they do not look identical on the wire, and every empty read above sits beside a read on the same host and script that did return rows.
Clearly Filtered — Water Pitcher
Evidence grade: listing-read. Names queried: Clearly Filtered, Inc. (company id C0872271); Clearly Filtered.
Both directories that list this pitcher were read today and both carry the same two-claim scope: NSF/ANSI 42 for chlorine and for taste and odour. There is no NSF/ANSI 53 listing for the company. The product page the corpus read markets chlorine, fluoride, lead, arsenic and PFAS, so four of those five are outside every listing this study read.
What this does not establish: Clearly Filtered publishes independent laboratory results for the contaminants outside its listings. A claim with no listing is not automatically an unsubstantiated claim — the FTC's bar is competent and reliable scientific evidence, and third-party certification is one route to it rather than the only one. This study measures certification coverage of claims, not truthfulness.
Rows read today, in the certifier's own wording
- Filtered Water Pitcher — Chlorine Reduction, Taste and Odor Reduction
- Clearly Filtered — Filtered Water Pitcher — Aesthetic Chlorine, Taste and Odor
- Affinity Filtration — Under Sink Filter — Aesthetic Chlorine, Taste and Odor
The byte count is the control. A search that matched nothing and a search that never ran look identical in prose; they do not look identical on the wire, and every empty read above sits beside a read on the same host and script that did return rows.
APEC — ROES-50
Evidence grade: listing-read. Names queried: APEC Water Systems; APEC.
WQA's own detail page for APEC Water Systems, read today, lists the ROES-50 under one reduction claim: TDS. NSF returns no matching product for APEC at all. The ROES-50 product page the corpus read publishes a contaminant table running to dozens of entries, from aluminium and arsenic to bacteria and barium.
What this does not establish: A reverse-osmosis membrane physically rejects far more than TDS; what is scoped here is what a certifier has verified and published, not what the unit does. APEC is certified — by WQA, to NSF/ANSI 58 — and the finding is about the breadth of the listing, not its existence.
Rows read today, in the certifier's own wording
The byte count is the control. A search that matched nothing and a search that never ran look identical in prose; they do not look identical on the wire, and every empty read above sits beside a read on the same host and script that did return rows.
Waterdrop — X16
Evidence grade: listing-read. Names queried: Waterdrop; Qingdao Ecopure Filter Co., Ltd.
Waterdrop is named here as a correction to this study's own method, not as a finding against the company. Searching NSF for the brand name returns no matching product; searching for its manufacturer, Qingdao Ecopure Filter Co., Ltd., returns a listing under NSF/ANSI 42, 53 and 401 in which hundreds of rows are Waterdrop-branded. A brand-level sentence reading "Waterdrop has no NSF listing" would therefore be false. The model-exact verdict survives it: no row in that listing is an X16, and the corpus's own evidence for the X16 is an IAPMO certificate carrying TDS, chlorine, taste and odour, and a lead-free materials row.
What this does not establish: An absence found under one name is not an absence. This is the second-query rule, and it is in this table because it changed an answer: the first query would have supported a brand-level claim the second query falsifies.
Rows read today, in the certifier's own wording
- Waterdrop ECF-7008S (one of the 555 rows) — Chlorine Reduction, Taste and Odor Reduction
- Waterdrop EFF-6002S (one of the 555 rows) — Cyst Reduction, Lead Reduction, Benzene Reduction, Carbofuran Reduction, Endrin Reduction, Mercury Reduction
The byte count is the control. A search that matched nothing and a search that never ran look identical in prose; they do not look identical on the wire, and every empty read above sits beside a read on the same host and script that did return rows.
Aquagear — Water Filter Pitcher
Evidence grade: searched-no-listing. Names queried: Aquagear (trade name); Aquagear (manufacturer name).
Two NSF queries today — trade name and manufacturer name — both return no matching product, against a positive control on the same host and script in the same run. The corpus holds no listed claim for this pitcher from any indexed body. Every claim on its record is written as an independent-laboratory measurement with a challenge concentration and a test method, which is the pattern this study counts as marketed-but-unlisted rather than as an unsupported assertion.
What this does not establish: Not listed is not untested and it is not false. Teymoorian and colleagues recorded the same position in a peer-reviewed comparison of pitcher filters: Aquagear is not NSF-certified for PFAS removal, and the manufacturer provides third-party performance data on its website.
The byte count is the control. A search that matched nothing and a search that never ran look identical in prose; they do not look identical on the wire, and every empty read above sits beside a read on the same host and script that did return rows.