Skip to content

We test what gets advertised at you

We buy the viral product with our own money. We cut it open. We publish the evidence.

TearDown Labs is a product teardown lab. We take heavily advertised products apart — physically, chemically, and down the supply chain to whoever actually made them — and publish what we find, including the parts that don't fit a headline.

No brand money. No free samples. No affiliate cuts. No advance sight for the brand. Ever.

The pledge

Zero sponsorship.

This is not a slogan. It is the rule that decides what we are allowed to do, and it costs us money every single time we apply it.

We test products that brands pay real money to advertise. The moment any of that money reaches us, everything we publish becomes worthless — so none of it is allowed to. Four things we will never accept:

  1. No money or samples from what we test

    No sponsorship, no ad spend, no paid placements, no "partnerships" — and no free, discounted or gifted product. Not from the brand whose product we test, not from its competitors, and not from any company in a category we test or intend to test. Money to pay for the work itself — a grant, a prize, a funder who simply wants more of this to exist — is a different thing, and it is allowed on one condition: it buys no say in what we find, what we publish or when. Every funder we accept anything from is listed on this site in full, by name, with what they gave and what it paid for. If a funder turns out to fall inside this refusal, we turn the money down, and we say publicly that we did.

  2. No free or discounted samples

    We buy every product at full retail price, from the same listing a customer would use, on an ordinary account. No review units, no press samples, no influencer discounts, no "gifted" anything. A product that arrives as a gift is not the product a customer receives.

  3. No affiliate or referral cuts

    Every link out of this site is plain and unmonetised. If we ever recommend something, we earn nothing when you buy it. There is no commission, no referral code, no tracked link — so the recommendation cannot be bought.

  4. No pre-publication review by the brand

    Brands get no advance sight of our findings, no draft to comment on, no veto and no copy approval. They find out what we found when everyone else does. We will happily publish a correction or a right of reply — after publication, in the open, labelled as such.

This makes us slower and poorer than the version of this site that takes the money. That is the entire trade, and we are choosing it on purpose: our only asset is that you can believe what you read here.

Funding, on the record

Nothing to declare yet: no grant, prize or third-party funding has been received, and nothing has been accepted from anyone. When that changes it is recorded here in full — who gave, how much, and what it paid for — before the work it funded is published.

This list is the whole of it. If money ever arrives that is not written down here, treat everything on this site as compromised.

Methodology

How a teardown is done here.

Written so that a sceptic can hold us to it — and so that someone with the same product and the same tools could try to reproduce us.

  1. We buy at retail price, like any customer

    Full price, from the normal listing, on a normal account. We keep the receipt and the order reference. No samples, no early access, no factory-direct units that a customer could never get.

  2. The brand gets no notice and no editorial input

    No heads-up before we buy, no draft before we publish, no line we agreed to soften. If a brand disputes something, it happens in public after the fact, with their words printed alongside ours.

  3. Physical and chemical claims get physical and chemical tests

    A claim about what a product is made of, how much of it there is, or what it does, is tested by cutting it open, weighing it, measuring it or running it through the appropriate test — not by repeating what the packaging says.

  4. We publish the evidence, and label what we removed

    Photographs of what we found inside, instrument readings and measurements, the test conditions, and a bill-of-materials cost estimate with a named source and a price for every input. Enough for you to check our arithmetic. Where a capture holds someone else's personal data, or the identity of a supplier we promised not to name, what we publish is a redacted copy and a log recording every removal, file by file — not the unredacted original. Each published copy carries its own SHA-256, so you can check that the file you hold is the one we hashed.

  5. We state what we could not verify

    Every published teardown carries its own limits: what our methods cannot see, what we could not test, where the uncertainty is, and what we are still guessing at. That goes in the same document, not in a footnote.

  6. Corrections are published, never quietly edited

    If we get something wrong, the correction is published as a dated note saying what was wrong, what changed and why. We do not silently edit a published finding, and we do not delete one.

Said plainly · These are commitments, not a track record. No teardown result has been published yet — the first one is in preparation, and these rules apply from the first one onward.

Teardown #1 · next up

Category chosen — product not bought, no test run yet

Blue-light glasses, and the “99%” claim.

We are going after a number, not a vibe. Evening eyewear is sold on the promise that it blocks 99% of blue light — one brand in this category advertises 100% — and neither page says what band that number was measured across. That is what makes it testable.

The plan: buy the pair at full retail from the same listing a customer would use, measure how much light actually gets through from 380 to 780 nanometres, identify what the frame and the lens are really made of, and publish the curves, the method, and a bill-of-materials estimate with a source on every line. The same specification — acetate frame, CR39 amber lens, the buyer's own logo — is offered by a public factory listing at $6.39 a pair on a 12-pair minimum, against $129.99 and $174.00 at retail. We name no factory and no supply relationship, because we have not established one.

The claim may well hold if “blue light” is defined narrowly enough. If it does, we will say so — that is the point of measuring instead of guessing. Nothing here is a result: no pair has been bought yet, no test has been run, no brand has been contacted, and no brand has paid us anything.

Product category
Blue-light-blocking evening/sleep eyewear — amber-tinted glasses sold on a percentage-of-blue-light-blocked claim. Retail for the pairs we are looking at: $129.99 and $174.00.
Marketed claims we intend to test
1) “Blocks 99% of the blue light” — measured by spectral transmittance across a band we define and publish up front. 2) “Blocks 100% of blue light and most green light” — the same measurement. 3) The frame material, described in the page's own product data as “premium acetate” — FTIR identification. 4) The filtering is inside the lens, not a coating that rubs off — transmittance before and after 100 standardised wipes.
Evidence published so far
One published page, and the evidence pack behind it: the claim-verification record for this teardown — a claim verification, not a test result. It holds the “99%” and “100%” claims as each brand's own page carried them on 28 September 2026, with the page address, the capture date and a SHA-256 fingerprint of the redacted public copy we publish of each. The five redacted public copies it rests on are published beside it, with the log that records every removal made in them and a readme explaining how to check a copy you download. No measurement exists: no pair has been bought, no lens has been measured, and no laboratory has run anything.
What the measurement would cost
No bench work is funded, and no money has been spent. The one measurement this teardown needs — spectral transmittance across a band we define and publish first — would cost about US$75–200 for a single instrument session, plus US$20–40 to ship the lens and have it returned. Those are our own figures, not quotes: the US$75–200 is our reading of five published university hourly cards (CAMCOR at the University of Oregon, Texas A&M, Duke, Georgia Tech and WVU), which price instrument time and not sessions, and the shipping band is our own estimate of a flat-rate postal box each way. No lab has been booked and nothing has been measured — but a price has now been quoted: €1,417 for the full two-technique scope (Measurlabs offer M40708, 1 October 2026), the final price with no tax added and shipping the only extra (≈US$1,601 at the ECB reference rate of 1.1298 USD/EUR published for 1 October 2026 — our conversion). Those cards price instrument time to academic users; that quotation buys a managed commercial service, and it is materially higher.
Read the claim-verification record

Published: “CLAIM VERIFICATION — NOT A TEST RESULT.” The record of what the two pages say — no measurement exists, and none is scheduled.

The evidence packThe five redacted public copies the record rests on, the log of every removal made in them, and the hashes to check a copy you download.

This is a plan, not a finding. The product has not been bought and no test has been run — everything above describes what we intend to do and how, plus prices taken from public listings.

For brands and retailers

Find out how your product fails before your customers do.

We do pre-launch vulnerability audits. We collect every claim a product is about to be sold on — product page, packaging, label, listing, ad creative, influencer script — record each one verbatim with its source and capture date, and sort them claim by claim into three piles: the ones a buyer can check; the ones nobody can check as written; and the ones that are undefined, meaning a number with no band, no test and no conditions attached. Where a number is undefined, we say so, and what would make it checkable.

The audit is bought by the party that has to defend the claim — the founder or product lead about to spend on launch claims, the agency about to put a client's claim live, the retailer's own-brand sourcing or QA function — not by the customer who is eventually told about it. One party pays for the check; nobody buys the finding. We take no sponsorship from the brands we audit and the audit buys no silence: if a claim will not survive an evidence-based teardown, telling you that first is the whole point of buying the check.

The three tiers

  • Available now

    Pre-launch Product Vulnerability Audit — Tier 1 (desk claim-check)

    • Your complete claim inventory, every claim recorded verbatim with its source and capture date, and the SHA-256 of each capture we froze.
    • The testability screen, claim by claim: checkable, uncheckable as written, or undefined.
    • A review of the specification you publish against the documents you hold.
    • A findings memo and a remediation list, in the format of the worked example.
    • A page stating what this audit does not establish — in the same words in every report, and not removable by the client.

    US$900, one-time

    Flat, whatever we find. Direct laboratory cost: $0 — desk work only: no laboratory, no unit shipped, no measurement, and the memo says so on its face. Turnaround: 5 working days from acceptance and receipt of your claim set — weekday working days, weekends not counted.

  • In development — not for sale

    Tier 2 — Single-measurement audit

    • Everything in Tier 1.
    • One unit and one instrument session: spectral transmittance, or FTIR material identification.
    • The raw instrument data and a method statement.

    Indicative US$2,000–3,000

    Indicative only — nothing buys this tier today. No laboratory has been engaged or booked, and the one quotation we hold (Measurlabs offer M40708, 1 October 2026 — €1,417, the final price with no tax added and shipping the only extra) is for teardown #1's own measurements, not a client audit at this tier.

  • In development — not for sale

    Tier 3 — Full pre-launch audit

    • Everything in Tier 1.
    • Up to three units, both techniques, and a consistency check across units.

    Indicative US$5,000–7,500

    Indicative only — nothing buys this tier today. No laboratory has been engaged or booked, and the one quotation we hold (Measurlabs offer M40708, 1 October 2026 — €1,417, the final price with no tax added and shipping the only extra) is for teardown #1's own measurements, not a client audit at this tier.

Tier 1 is the only tier you can buy today: US$900, one-time, flat, whatever we find. It is desk work — no laboratory, no unit shipped, no measurement — and the memo says so on its face. The turnaround is published too: the report comes back within 5 working days of your acceptance and of our receiving your claim set — and if you send no claim set, the clock runs from acceptance, on the claims we capture ourselves. A working day is a weekday, Monday to Friday; weekends are not counted.

Tiers 2 and 3 are in development and are not on sale. Their ranges are our own estimate, read from published university facility hourly rate cards. For scale, one UV-Vis transmittance session reads at US$75–200 and round-trip shipping at about US$20–40. Both are our own figures: the band is our reading of five published hourly cards (CAMCOR at the University of Oregon, Texas A&M, Duke, Georgia Tech and WVU), which price instrument time and not sessions, and the shipping is our own estimate of a flat-rate postal box each way. Those cards price instrument time to academic users; they are not what a commercial mail-in order costs. Measurlabs has now quoted €1,417 for teardown #1's two lenses and two materials — the final price, with no tax added and shipping the only extra — (offer M40708, 1 October 2026); a single transmittance sample there is €490, or €587 with the once-per-order fee, nearly three times the top of the band above, because the price buys sample handling, reporting and project management as well as instrument time.

Direct laboratory cost is passed through at cost and never marked up, with the facility's own figure in your quote, re-quoted before anything is booked. The term we recommend is that you pay the facility directly, so “nothing is passed through” is checkable against the facility's own invoice.

What you supply

  • Your claim set, if you have one — every performance, materials, safety or comparative claim you publish or intend to publish, wherever it appears or will appear. Supply nothing and the audit still runs, on the claims we can capture from what you have already published.
  • The product or listing as a customer sees it, so the screen reads what a buyer reads.
  • The specification the product is supposed to meet, and any report, study or certificate you already hold behind a claim.
  • The launch date you are working to — it decides whether there is time for anything beyond the desk tier.

Tier 1 is desk work: nothing is shipped and no unit leaves your hands. Units and shipping apply only to the measurement tiers, which are not on sale.

What you get

  • A complete desk claim-check in the format of the worked example — the same sections, filled in with your product.
  • The claim inventory: every claim verbatim, with its source and capture date, and the SHA-256 of each capture we froze.
  • The testability screen, claim by claim.
  • The specification review, read against the documents you hold.
  • A findings memo and a remediation list: what to change, and what is already defensible.
  • A page stating what this audit does not establish, in the same words in every report and not removable by the client.

No laboratory, no measurement, and the memo says so on its face. Nothing we sell is a legal, regulatory or safety conclusion.

The conflict rule, in your terms

We do not sell a claim check to a company in a category we test or intend to test. Our published refusal is the reason: we accept no money, samples, sponsorship or free or discounted product from the brand whose product we test, its competitors, or any company in a category we test or intend to test. A fee from such a company would be money from inside a category we cover, so the rule binds us.

The category check runs before money and before scope. An enquiry gets a one-line answer — we can take this, or we cannot and here is the category reason. If we cannot, we take no fee and no sample.

The billing rule is part of the same fence, and there is no way around it: every engagement is billed on its own — a single-use payment link or an invoice created for you after the category check and the written scope. There is no standing public payment page on this site and there never will be, because a link anyone could pay would let a company in a category we test pay before the ring-fence check runs.

This is forewarning, not approval: nothing we sell says a claim will survive a later teardown, a regulator or a court. A paid claim check buys no protection from a future teardown — if we ever tear down a category we have audited in, we tell you before we publish anything, take no further or renewal work from you, and disclose the relationship inside that teardown.

Clients, on the record

Nothing to declare yet: no audit has been commissioned and no brand has paid us for anything. When that changes, every client and its product category is named here in full, by name, with what was paid and what it paid for — that list is the only way anyone can check the ring-fence instead of taking our word for it.

What you cannot buy, at any tier

No editorial review. No preview. No veto. No delay. No softer finding.

You do not see the memo before it is final, and you cannot require us to change, delay, withdraw or soften a finding. There is no price, at any tier, at which a claim we found unsupported becomes supported — there is no “clean report” product. A result that could be edited could not be believed by anyone else.

If we publish anything naming your company, your reply is published too — unedited, labelled as a reply and printed in full wherever the document appears. If we misstate a fact about your own product, we correct it on the record, with a note that we did.

How to start

Email the enquiry address with the product, the category, the launch date and the claims you intend to publish — that is all the category check needs, and nothing is owed for it. You get a one-line answer on the category within 2–3 working days, then a written scope and fee if we can take the work. Once you accept and we have your claim set, the desk check runs and the report comes back within 5 working days — that window is published, and it is the same on every Tier 1 engagement.

A serious prospect can also ask to see the worked example at no charge: a complete desk claim-check of a real advertised product, in the format every paid report follows. It is a demonstration of the format, not a commissioned audit and not a test result — nothing in it was bought, tested or measured.

Payment is by invoice or card, on the terms set out in the engagement letter sent with the written scope, so nothing is paid against a scope that has not been written down.

Email the lab

Opens in your own email client — [email protected].