Nano Filter: We have found the original lab reports.
The laboratory reports cited by Nano Filter are available online. We have read them and compared them with the published figures. Documents, addresses and method.
In May, we published a comparison between the Nano Filter cartridge and the Coldstream FTO+, based on the only documents available at the time. We wrote that the results highlighted by the brand came from a layout made by Nano Filter SA, reproducing figures attributed to SGS, and that we had not been able to obtain the original reports.
We have since consulted them. They are accessible online, on the SGS portal, without authentication. Here are the two addresses. Anyone can open them and check the following for themselves.
https://qrcode.sgsonline.com.cn/openreport/viewreport/d64cbea8-5cd1-4781-b143-80982f0a6121
https://qrcode.sgsonline.com.cn/openreport/viewreport/6b0b020d-db26-4ebe-b7d5-b7537c143460
The initial comparison, with its May and September updates, can still be found here: Nano Filter vs Coldstream FTO+: Promises, Tests and Evidence
To go further, you can also consult our comparison "Nano Filter vs Coldstream FTO+: Promises, Tests and Proofs", with its May and September updates.
We had also submitted a request for verification to SGS. The group's Corporate Security department replied to us on 5 September 2026 that the document submitted was not a document issued by SGS, and invited us to request the reports from the supplier.
What the two original reports contain
Report NBE24-0013449-01, dated October 17, 2024, six pages, issued by SGS-CSTC Standards Technical Services, Ningbo Branch.
Customer name: EVODROP AG, Birkenstrasse 21, 8306 Brüttisellen, Switzerland.
Sample Name: Reinguard filter, quantity four sets.
Measuring points: at start-up and at 5,000 litres.
Flow rate indicated at the end of the report: 2 litres per minute.
Report NBF25-0023352-02, dated November 25, 2025, five pages, same laboratory.
Customer name: EVODROP AG, same address.
Sample Name: 10inch water filter, quantity eight sets.
Measuring points: 0, 2,500, 5,000 and 7,500 liters.
Flow rate indicated at the end of the ratio: 4 litres per minute.
These are the two reference numbers on the letterhead of the documents distributed by the brand.
The flow rate, and why it is enough to decide
A gravity filtration system operates under the sole pressure of a water column of a few tens of centimetres. The cartridge concerned is announced at 18 litres per hour, or 0.3 litres per minute.
Both reports mention flow rates of 2 and 4 liters per minute. These values correspond to pressurized fed cartridges. A ten-inch filter, moreover, is a standard size of cartridge under sink or in-line.
You don't have to be a chemist to do this calculation.
The last page
Each SGS report ends with a page titled Sample photo, which shows the sample actually tested.
In the 2024 report, this page 6 of 6 shows a cylindrical stainless steel filter housing with threaded connections, accompanied by a pleated cartridge with blue tips. In the 2025 report, this page 5 of 5 shows a ten-inch cylindrical cartridge with blue tips.
Neither of these two photographs depicts a gravity-fed filter candle.
This same page bears, just below the photograph, the following mention, in English in the text:
Attention: Unless otherwise stated the results shown in this report refer only to the items tested.
Unless otherwise stated, the results presented in this report relate only to the items tested.
The page missing from the distributed version is therefore the one that identifies the device tested, and the one that reminds us that the results are valid only for it. The two pieces of information appear on the same sheet.
The figures match, except for one
The one-page document entitled Elimination of microorganisms presents a table with two columns, at start-up and after 5,000 liters. It contains twelve values, four per microorganism. Eleven correspond exactly to the two EVODROP reports.
Escherichia coli: 65,000 then 13 at start-up, 32,000 then 98 after 5,000 litres, i.e. 99.98% and 99.69%. These values are shown identically in the 2024 EVODROP report, tables 1 and 2.
Staphylococcus aureus: 46,000 then less than 1, and 30,000 then 1,100, i.e. more than 99.99% and 96.33%. Same tables, same ratio.
Pseudomonas aeruginosa after 5,000 liters: 750,000 then 4,500, i.e. 99.40%. Table 3 of the 2025 EVODROP report.
The table published is therefore composed of two separate reports, drawn up in the name of another company, relating to two different devices.
The twelfth value does not match. For Pseudomonas aeruginosa at start-up, the document released indicates an input concentration of 1,500,000 CFU/250 ml. The 2025 EVODROP report indicates, on page 2, table 1, point 0 litre, a concentration of 1.5×10⁵, i.e. 150,000 CFU/250 ml.
A factor of ten.
The deviation has an arithmetic consequence. With 150,000 at the input and less than 1 at the output, the reduction is greater than 5 log. With 1,500,000, it becomes greater than 6 log. Six log on bacteria is the threshold retained by the NSF/ANSI P231 protocol for microbiological purifiers.
The same document bears, at the bottom of the page, the following statement: all the values correspond exactly to this ratio and have not been modified.
We are not adding anything to that. The report is online, the page is indicated, everyone can compare for themselves.
This document also indicates Nano Filter SA as a sponsor, when the two reports cited in its header bear EVODROP AG. It announces an ISO 6222 and ISO 9308 method, when the Pseudomonas of the 2025 report was measured according to the Chinese standard GB 8538-2022. And it indicates that the counts were carried out before and after filtration through the Nano Filter gravity filter.
The same figure appears on their website
This value is not confined to a document sent on request.
The site's Tests and Certifications page, as of the morning of September 24, 2026, publishes the same table and value: Pseudomonas aeruginosa, 1.5×10⁶ CFU/250 ml input, less than 1 output, more than 99.99%. The page also indicates, further down, that the microbiological results come from the reports SGS NBF25-0023352-02 and NBF24-0013449-01.
The first of these reports is available online. It indicates 1.5×10⁵.
https://nanofilter.shop/pages/tests-et-certifications
The document on chlorine
A second document, entitled Chlorine and Disinfection By-Products, has the same two references in the header (NBF25-0023352-02 and NBF24-0013449-01). It goes further than the previous one, because it does not content itself with omitting the identity of the device tested: it affirms it.
It reads: residual chlorine and chloroform were added to enriched water and then measured before and after filtration through the Nano Filter gravity filter.
The values presented are, for residual chlorine, 2.08 then less than 0.01 at start-up and 2.04 then less than 0.01 after 5,000 liters. These are exactly the 0 and 5,000 liters points of the 2025 EVODROP report, carried out on the 10inch water filter.
For chloroform, 0.2935 then less than 0.0002 and 0.3071 then 0.0026. These are exactly tables 1 and 2 of the 2024 EVODROP report, carried out on the Reinguard filter.
A point must be added. The two reports cited in the header contain data on chlorine. The 2025 report on the 10inch water filter indicates more than 99.50% at 5,000 litres: this is the series that is used. The 2024 report on the Reinguard filter indicates 98.38% for dissociated residual chlorine at start-up and 85.79% after 5,000 litres: this series is nowhere to be found.
Of the two series available, the one that appears is the most favourable.
The endurance test
As of September 1, 2026, the Tests and Certifications page on the brand's website had a section titled Performance after 5,000 liters, introduced by this sentence: an additional test measured the cartridge's performance after 5,000 liters filtered, which is 1,000 liters beyond the guaranteed capacity of 4,000 liters.
This page is archived and available for consultation in its state as of September 1, 2026 at the following address:https://archive.ph/OLnMG
However, the only data at 5,000 liters that we have found are those of the two EVODORG reports, obtained on the Reinguard filter and on the 10inch water filter.
Mineral conservation
The same archived page has a section entitled Conservation of Essential Minerals, introduced by the phrase Nano Filter removes contaminants without demineralizing the water, and a table showing a preserved share of 104% for calcium, 130% for potassium and 115% for sodium.
You don't preserve 130% of a mineral. These values mean that the water comes out more charged than it entered. The corresponding test report also calls this table removal / leeching, i.e. removal and release.
This point also deserves to be compared with the technology claimed. The rejection of divalent ions, calcium and magnesium in particular, is the defining characteristic of nanofiltration. A table showing 104% preserved calcium describes something other than a nanofiltration membrane.
We add that retaining useful minerals is in itself a quality, and that our own cartridges do not demineralize either. It is not the result that we are discussing, it is the name given to it.
The point at 7,500 liters
The 2025 report has four measurement points. The documents released included only two.
At the point of 7 500 litres, which is not included, the report indicates an effluent of 6.2×10³ for Pseudomonas aeruginosa compared to 4.5×10³ at 5 000 litres, and for chlorine a reduction of 97.51% compared to more than 99.50% at the previous points.
End-of-life data existed. They were not published.
The new September 2026 report
Since our previous update, the brand has published an additional report, this time in full and without reformatting.
Report NBF26-0024602-01, dated September 10, 2026, six pages, SGS-CSTC Standards Technical Services (Ningbo).
Customer name: Nano Filter SA, c/o DYN SA, Rue de la Grotte 6, 1003 Lausanne.
Sample name: Gravity-fed water filter.
Trial period: from August 6 to September 10, 2026.
In form, this report is much superior to the previous ones. It is established in the name of the brand, it relates to the right product, and it is based on recognized protocols: NSF/ANSI 53-2024, NSF/ANSI 401-2023, NSF/ANSI 42, USEPA 537.1, EPA 1602, ASTM D5127. The microbiological results are expressed in logarithmic reduction, which is the correct unit.
It is also published in its entirety, with its verification QR code. It also bears the standard mention that the results relate only to the elements tested, and here this mention poses no difficulty: the element tested is indeed the product sold, and the document is accessible in full. This is why a laboratory report must be communicated, and we have nothing to say about it.
One question remains, and it is a big one. In all five result tables, without exception, the Testing Point column has the same mention: Start-up. At the start of the lifespan.
No measurements at 1,000, 2,000, 3,000 or 4,000 litres. No data on what the cartridge does at the end of its life, even though the announced capacity is 4,000 litres.
A second, more technical observation. The report indicates that the test protocol applied to the bacterial count, the MS2 bacteriophage and the nine antibiotics in Table 6 is the one provided by the customer. The method of analysis for MS2 remains EPA 1602, which is a recognized method, but the conduct of the test itself does not refer to any published standard. Table 6 is the only one of the six to have no reference value columns.
This is precisely the question that this article has been asking since its first version.
Who carried out the particle measurements
SMS reports mark tests that are assigned to other laboratories, and list them at the end of the document. This is a common practice, and they are credited with reporting it.
In the September 2026 report, the single marker refers to SGS Shanghai and covers three pesticides. The dual marker refers to SGS Xiamen and covers all twelve parameters of the NSF/ANSI 401 protocol. Both of these entities belong to the SGS Group.
The triple marker refers to SAE Technologies Development (Dongguan) Co., Ltd., Material Science Lab, which the report distinguishes from the two SGS entities it otherwise names. It covers exactly two lines: 3 μm polystyrene microspheres, and 0.5 to 1 μm microplastic particles.
The same is true in the two EVODROP reports. The 2024 report counted particles larger than 10 μm was carried out by the Shanghai Institute of Measurement and Testing Technology. The 5 μm count in the 2025 report was carried out by the same laboratory in Dongguan.
Over the three campaigns, no particle measurements were therefore carried out by SGS.
Not all tests are carried out by SGS
The Tests and Certifications page has a certifications section that indicates that all tests are performed by SGS, an independent ISO/IEC 17025 accredited laboratory.
This sentence is contradicted twice.
First by the page itself. Its own Sources section first cites the Measurlabs report number 29505, which covers PFAS, pesticides, pharmaceutical residues, metals, minerals and free chlorine. Measurlabs is a Finnish laboratory, separate from SGS. It is this report that provides the majority of the figures displayed on the page.
Then by the SGS reports themselves, which state that the particle counts were entrusted to SAE Technologies Development (Dongguan) and the Shanghai Institute of Measurement and Testing Technology.
The accreditation of these laboratories is not in question and we have nothing to reproach them with. What does not correspond is the assertion with the documents cited a few centimetres further down on the same page.
Microplastics and nanoplastics
The homepage, cartridge sheet, and system sheet announce a microplastic and nanoplastic elimination of between 98.96 and more than 99.99%.
The 98.96% is identifiable. This is the Microplastics particles line, from 0.5 to less than 1 μm, from the September 2026 report: 270,957 particles at the input, 2,811 at the output.
Three elements accompany this line in the report. It has been doped with ISO fine test dust 12103-1, A2, a standardized mineral dust used as a reference test dust in the industry, which is not plastic. It was measured according to ASTM D5127-13, which is a guide for ultrapure water in the electronics and semiconductor industries, not a performance standard for drinking water. And it was carried out by the Dongguan laboratory mentioned above.
You have to recognize a point, and it counts. The Tests and certifications page itself declares this substitution, in a method note under the table. It's not hidden. It's the home page and the product sheets that include the number without the score.
The fundamental question remains. The nanoplastics range is below the micron, and generally much lower. The finest measurement published is on the 0.5 to 1 μm wafer, and it was obtained with mineral dust. No tests on plastic particles below 3 μm are included in the available documents.
We add, because it would be dishonest to omit it, that the line of polystyrene microspheres of 3 μm is correctly used. It documents the retention of protozoan cysts, for which these microspheres are the recognized substitute. On this precise point, there is nothing to complain about.
What volumes are the figures displayed on
There are three sources that feed into the tables on the Tests and Certifications page. It's helpful to know what each one measures.
The Measurlabs Report 29505 provides PFAS, pesticides and their metabolites, pharmaceutical residues, metals, minerals, and free chlorine. It indicates, in each of its sections, a filtered volume of five liters. Five livery for the twenty PFAS. Five liters for the one hundred and fifty-seven pesticides and metabolites. Five liters for metals.
The September 2026 SMS Report provides microbiological results, particle counts and some of the micropollutants. Its five tables of results all bear the same mention in the Testing Point column: Start-up. (Start-up means "on a new cartridge, in its very first liters.")
The two EVODROP reports supply the columns after 5,000 liters. They were carried out on the Reinguard filter and on the 10inch water filter, at flow rates of 2 and 4 liters per minute.
In other words, and this is the central point of this update: on the product actually sold, no published measurements exceed five filtered liters or the start of the service life. The only endurance data available comes from other devices.
A new filter media always works well. The question has never been what it does to the first liter, but what it still does to the two-thousandth.
A correction to our September 1st update
We wrote at the time that no virological trials appeared in the documents provided. This is no longer true.
The September 10, 2026 report includes an MS2 bacteriophage assay, with an influent of 1.0×10⁵ PFU/mL, an effluent of 3, or 4.52 log reduction. MS2 is the model virus used by microbiological reference protocols, which typically require 4 log on viruses. Therefore, this result satisfies the criterion.
We point it out because we correct our mistakes, and because we ask others to do the same.
Two clarifications, however, which do not detract from the above. It is a bacteriophage used as a substitute, not a human virus. And the measurement is made at start-up. For comparison, Coldstream FTO+ cartridges have data on the human pathogen rotavirus, measured in increments up to 3,000 litres.
What we don't say
We do not lay any charges. We describe documents, we give addresses, and we invite everyone to open them.
We do not know what agreements may exist between the companies mentioned, or in what capacity the results obtained by one may have been presented by the other. We do not comment on the intrinsic quality of the cartridge: it has never been tested by us, and the September 2026 report shows start-up performance that is not mediocre.
What we are seeing is simpler. For months, a commercial communication has relied on reports that have the name of another company and describe other devices.
And the first report made in the name of the brand on the right product only measures the start-up.
In accordance with our rule, we sent Nano Filter SA, before publication, a letter detailing the foregoing findings and asking the resulting questions. We indicated the date of publication of this update, and proposed an extension if the preparation of a response justified it.
As of the date of publication, we have not received any response.
The offer remains open. Any response that we receive will be published here, in full.
The documents cited
All documents mentioned in this update are public. Here's where to find them.
The two original SMS reports on the SMS Audit Portal:
https://qrcode.sgsonline.com.cn/openreport/viewreport/d64cbea8-5cd1-4781-b143-80982f0a6121
https://qrcode.sgsonline.com.cn/openreport/viewreport/6b0b020d-db26-4ebe-b7d5-b7537c143460
The documents published by the brand, served from the file space of its own store:
Elimination of microorganisms:
https://cdn.shopify.com/s/files/1/0983/0747/3790/files/SGS-Rapport-Pathogenes-Nano.pdf?v=1783799621
Chlorine and disinfection by-products:
Full test report, 224 parameters:
SMS Report NBF26-0024602-01 dated 10 September 2026:
https://cdn.shopify.com/s/files/1/0983/0747/3790/files/Rapport-SGS-NBF26-0024602-01.pdf?v=1789127745
The v parameter for these addresses is a Unix timestamp, which corresponds to the last update of the file. It gives July 11, 2026 at 9:53 p.m., July 12, 2026 at 3:28 p.m., August 11, 2026 at 5:28 p.m., and September 11, 2026 at 1:55 p.m. Paris time, respectively.
Two of these values are used to verify the method. The Measurlabs report is dated August 10 and its timestamp is August 11. The SGS report is dated September 10 and its timestamp is September 11. In both cases, the day after the issuance.
The Brand Tests and Certifications page, in its current state:https://nanofilter.shop/pages/tests-et-certifications
The same page, archived in its state as of September 1, 2026:https://archive.ph/OLnMG
We keep an archived copy of each of these pieces. If one of them becomes inaccessible, we will publish the corresponding archive link.
What this means for you
None of the above requires you to trust us. The addresses are on this page. The reports are with SGS. The archived page is with a third party. Open them.
And ask the same four questions everywhere. On how many liters was the measurement made, and at how many points. Who conducted the test, and who sampled. What water, with what characteristics. What performance standard, and not just what method of analysis.
Also look at the last page. This is often the one that says which product has actually been tested.
Our assessment
At the end of this verification, our assessment is as follows, and we assume it as such.
The documents we were able to consult do not support the announced performance for this cartridge over its lifetime. The tests on significant volumes were ordered by another company and carried out on other devices. The two tests carried out on the product actually sold concern, one on five filtered litres, the other on the start-up alone.
Everyone will judge what this says about a Nano page entitled "Evidence, not mere promises."