Free Certificate of Analysis (COA) Verification Tool
Upload a Certificate of Analysis as a PDF or an image, or paste a link to one, and find out whether it holds up. It reads the file the way a document examiner would, recovers the fields from the page even when the certificate is a scan, checks whether the numbers survive arithmetic, compares the document against the issuing laboratory’s own format, and asks that laboratory directly wherever a public register exists. Free, no account, and it works on certificates from any vendor.
Drop the certificate here, or click to choose
PDF, PNG or JPEG, up to 20 MB
This tool reports what it can check about a document. A certificate that passes every check is still only a statement about the material a laboratory received, which is not necessarily the material in your vial. Nothing here is a substitute for testing your own sample, and none of it is medical advice. All Volta materials are for in-vitro laboratory research use only.
What a Certificate of Analysis actually is
A Certificate of Analysis is a laboratory’s written statement of what it measured in a sample it was given. It records the sample as it was identified to the laboratory, the tests performed, the results obtained, and the date the work was done. That is the whole of it.
The single most misread thing about a COA is its scope. A certificate is evidence about a sample that arrived at a laboratory on a particular day. It is not evidence about a vial in your hand, unless something independent connects the two: a batch or lot number printed on both, and a supply chain you have reason to trust. A genuine certificate attached to the wrong product is still a genuine certificate, and it tells you nothing useful.
A COA is also not a quality judgement. Laboratories report numbers. Deciding whether 97.9% purity is acceptable is the buyer’s judgement, or the specification’s, and a certificate without a stated specification is reporting results against no acceptance criteria at all.
- Certificate of Analysis
- Reports measured values from named test methods on an identified sample. The document type this tool is built to read.
- Certificate of Conformance
- Asserts that material meets a specification without necessarily reporting the underlying numbers. Weaker evidence: you are taking the assertion on trust.
- Test report
- Often used interchangeably with a COA. Many third-party peptide laboratories head their documents "Test Report" rather than "Certificate of Analysis"; the wording carries no significance on its own.
- In-house vs third-party
- A certificate produced by the seller reports the seller’s own measurements. A third-party certificate comes from a laboratory with no stake in the result. Only the second is independent evidence, and only if the laboratory can be shown to exist.
What to look for on a certificate
Most certificates that turn out to be worthless are not elaborate forgeries. They are thin: a compound name, a number, a logo, and nothing that could ever be checked. The questions below are the ones that separate a document carrying information from a document carrying reassurance.
- A named laboratory you can reach
- A real address, a real domain, a real contact route. A laboratory name that appears nowhere except on this document is the most common feature of a fabricated certificate.
- A report or task identifier
- The number that ties the document to a record at the laboratory. Without one, there is nothing to verify against, even in principle.
- A batch or lot number
- The only thing connecting the certificate to a specific production run. Check it against the vial label. Many genuine reports leave this blank because the client did not supply it, so its absence is a limit on what the document proves rather than a sign of fraud.
- Dates that make sense together
- Ordered, received, analysed, reported. They should run forwards, and none should be in the future. A certificate dated before the sample arrived is describing work that had not happened.
- The method, not just the answer
- For a chromatographic purity figure: the column, the gradient, the flow rate, the detection wavelength. A percentage with none of that behind it cannot be reproduced or assessed by anyone. Some laboratories issue deliberately short summary reports, so this is a limit on the evidence rather than proof of anything.
- Identity evidence, separate from purity
- Purity says how much of the main component is present. It says nothing about what that component is. Identity comes from mass spectrometry, or from a sequence determination. A certificate with a purity figure and no mass has confirmed nothing about the molecule.
- Numbers reported with their measurement
- A quantitative certificate states what was measured. "Pass", "conforms" and "complies" with no value behind them are assertions, not measurements.
- A verification route
- Some laboratories print a key or a QR code that resolves to their own copy of the report. That is the strongest check available to a buyer, because it does not depend on trusting the seller at all.
Understanding the numbers on a peptide COA
The figures on a certificate are more specific, and more limited, than they look. Each one answers a narrow question, and most disputes about peptide quality come from reading a number as an answer to a question it was never asked.
- Purity by HPLC area percent
- The main peak’s share of the total integrated peak area at a stated detection wavelength. It measures a share of ultraviolet absorbance, not a share of mass. Two compounds present in equal amounts do not necessarily produce equal peak areas, because they absorb differently. Area percent is a good relative measure and a poor absolute one, and it can only see what elutes from the column and absorbs at that wavelength.
- Why the peak table must total 100%
- Area percentages are shares of one chromatogram. A printed peak table summing to more than 100% is arithmetically impossible. Summing to less usually just means small peaks were omitted from the printed table, which is common and not in itself a problem.
- Identity by mass spectrometry
- Electrospray ionisation produces a series of multiply charged ions rather than a single mass. A reported m/z at charge state z converts to a neutral mass of roughly (m/z x z) minus z proton masses. Certificates quote either the monoisotopic mass or the average mass, and the two differ measurably on a large peptide, so a mismatch of a few daltons is often a difference of convention rather than a different molecule.
- Net peptide content is not purity
- Chromatographic purity describes the peptide fraction only. A lyophilised vial also contains bound water and a counter-ion, typically acetate or trifluoroacetate left over from synthesis and purification. Peptide content, water and counter-ion are shares of the same powder and should account for the whole of it. This is why a vial labelled with a given mass contains less actual peptide than that figure suggests, and why net peptide content is always lower than the purity percentage.
- Amino acid analysis
- The reference approach for establishing how much peptide is actually present, rather than inferring it from ultraviolet absorbance or from the weight of the powder. Rarely seen on certificates in this market, and a strong signal of rigour when it appears.
- "Not detected" is a statement about the method
- It means the analysis did not see the substance above its detection limit. It does not mean zero. A result reported as not detected is only as meaningful as the limit of detection behind it, and a certificate that reports the phrase without the limit has left out the informative half.
- Limit of detection and limit of quantitation
- Detection is the level at which a method can tell that something is present. Quantitation is the higher level at which it can say how much. A value between the two is real but not reliably measurable, which is why careful reports say "below the limit of quantitation" rather than giving a number.
- Significant figures and false precision
- A purity reported as 97.945% claims precision to the thousandth of a percent. Chromatographic integration does not support that, and the extra digits are an artefact of the software rather than a measurement. This is normal on real certificates and is not a sign of forgery. What is worth noticing is the opposite pattern: a figure of exactly 100.00%, or the same suspiciously round number on every batch a seller publishes.
- Dimer and higher molecular weight species
- For recombinant proteins, individual molecules can associate into dimers and larger aggregates. These are usually measured by size exclusion chromatography, which separates by size rather than by hydrophobicity, and they are reported separately from chromatographic purity because a reversed-phase method may not resolve them at all.
- Milligrams and international units
- Some biologics are quantified by biological activity in international units as well as by mass. For recombinant human growth hormone the conventional conversion is three international units per milligram, so a vial reported as 3.86 mg is equivalently reported as about 11.58 IU. When both figures appear on a certificate, they should be consistent with each other; a pair that does not divide cleanly is worth asking about.
How fabricated certificates usually fail
Forged certificates are rarely built from nothing. The common cases are a genuine report with a figure edited, a genuine layout with a different laboratory’s name on it, or a real certificate for one batch presented as evidence for another. Each leaves different traces.
- An edited value on a genuine report
- The most common and the hardest to see by eye. In a PDF it often leaves a second text run underneath the new one, a figure set in a typeface used nowhere else on the page, or an annotation object drawn over the original. In an image it can leave a number whose weight, spacing or baseline differs subtly from its neighbours.
- Metadata that contradicts the document
- A PDF records which software wrote it and when. A laboratory report comes out of a reporting system. An image editor named in the producer field, an edit history left in the file, or a creation date earlier than the analysis date printed on the page are all things a genuine export does not have.
- A file saved more than once
- A PDF written once and sent has one end-of-file marker. Every later save appends another, along with a fresh cross-reference table, and the previous content usually survives inside the file.
- A lookalike verification address
- A certificate that invites you to verify it at a domain one character away from the laboratory’s real one is sending you to a site the seller controls, where of course it will verify. Read the printed address character by character.
- An identifier of the wrong shape
- Laboratories issue report numbers and verification keys in consistent formats. An invented identifier often does not match the pattern the named laboratory actually uses, which is checkable without contacting anyone.
- A recycled certificate
- The same report number, or the same batch number, appearing across products or across months. Nothing on the document itself reveals this; it is found by comparing certificates a seller has published over time.
- A flattened image where a document should be
- Re-exporting a PDF as a picture destroys the metadata and structure that would settle the question. That is not proof of anything on its own, because several laboratories publish their reports as images, but a seller who can only ever produce screenshots is worth a question.
Frequently asked questions
The questions people actually ask about certificates of analysis, answered without overstating what a document can prove.
Certificates of analysis
A Certificate of Analysis is a laboratory’s record of what it measured in a specific sample: the sample as identified to the laboratory, the tests run, the results, and the dates. It is evidence about the material the laboratory received on a particular day. It is not, by itself, evidence about any other material, including the vial you are holding.
Yes. Upload a certificate as a PDF or an image, or paste a direct link to one, and the full analysis runs at no cost. There is no account, and the tool works on certificates from any vendor and any laboratory, not only ours.
No, and this is the most important limitation to understand. A clean result means nothing checkable is wrong with the document. The document describes a sample a laboratory received. Connecting that sample to the vial in your hand requires a batch number that matches your label and a supply chain you trust. Only testing your own material answers the question directly.
Purity says how much of the main component is present relative to everything else the method can see. Identity says what that component actually is. They are measured differently: purity typically by chromatography, identity by mass spectrometry or sequencing. A certificate reporting 99% purity with no mass data has established that the sample is 99% something, without establishing what.
It measures the main peak’s share of total integrated peak area at a stated detection wavelength. That is a share of ultraviolet absorbance, not a share of mass. Because different compounds absorb differently, area percent is a reliable relative measure and an unreliable absolute one. It also cannot see anything that fails to elute from the column or fails to absorb at the wavelength used.
Because they measure different things. Purity describes the peptide fraction only. A lyophilised vial also contains bound water and a counter-ion such as acetate or trifluoroacetate left from synthesis. Peptide, water and counter-ion are shares of the same powder, so a vial can be 98% pure by chromatography while the actual peptide is a noticeably smaller share of the total mass.
It means the method did not see the substance above its detection limit. It does not mean zero, and it is only as informative as the limit behind it. A certificate that reports "not detected" without stating the detection limit has omitted the half of the result that gives it meaning.
It is worth a second look. Real chromatographic integration produces varied figures with a tail of minor peaks underneath. Exactly 100%, or the identical near-perfect figure across every batch a seller publishes, reads as a claim rather than a measurement. Synthetic research peptides realistically land somewhat below that.
It is the only thing tying a certificate to a specific production run, so it should match the label on your vial. Many genuine reports leave it blank because the client did not supply it when submitting the sample. A blank batch field is therefore a limit on what the certificate can prove rather than evidence that it is fake.
Janoshik reports carry a task number at the top and a printed unique key at the bottom, with the line inviting you to verify the test on their site. Entering both on the laboratory’s own verification page returns their copy of the report. Compare the task number, sample name, purity and quantity against the document you were given: a genuine number with edited figures is the case this catches.
Laboratory sites commonly sit behind bot protection that challenges automated requests coming from server addresses. When that happens the lookup returns nothing, which says something about us and nothing about your document. The result is reported as blocked, it never counts against the certificate, and the tool gives you the verification link and the values to compare so you can complete the check from your own browser.
Easily, and this is more common than outright forgery. A real certificate for one batch presented as evidence for a different batch is genuine and irrelevant. So is a real certificate for a sample the seller hand-picked. So is a real certificate that tests only purity when the question was identity. The document being authentic and the document being informative are separate questions.
Third-party means the testing laboratory has no commercial interest in the outcome. A vendor testing its own product reports its own measurements, which requires you to trust the vendor about the very thing you were trying to check independently. Third-party testing only counts if the laboratory can be shown to exist outside the certificate.
It indicates the laboratory has been assessed as competent to perform particular tests under a recognised standard. Two cautions: accreditation is granted for a defined scope, so a laboratory can be accredited for some methods and not the one on your report, and a claim of accreditation printed on a document is only as good as the accreditation body’s own register, which is where it should be checked.
Only if it resolves to the laboratory’s own domain. A QR code is just an encoded link, and whoever made the document chose where it points. A code leading to a seller-controlled page, or to a domain a character away from the laboratory’s real one, is worse than no code at all because it manufactures the appearance of verification.
A PDF carries evidence a picture does not: which software wrote it, when it was created and modified, how many times it has been saved, whether objects were replaced, and whether text has been drawn over other text. Flattening to an image discards all of it. The tool reads images perfectly well, but a document supports more checks than a picture of one.
For recombinant proteins, molecules can associate into dimers and larger aggregates. These are measured by size exclusion chromatography, which separates by molecular size, and reported separately because a reversed-phase purity method may not resolve them at all. A protein certificate that reports purity but says nothing about aggregates has left a relevant question unanswered.
Some biologics are quantified by biological activity in international units alongside mass. For recombinant human growth hormone the conventional conversion is three international units per milligram, so a reported 3.86 mg corresponds to about 11.58 IU. Where a certificate prints both, they should be consistent with one another.
Read which check failed, because the findings are separated deliberately. A contradiction, such as a register that disagrees or arithmetic that cannot hold, is decisive. A warning about a thin report is a limit on the evidence. A check that could not run is neither. Take the specific finding to the seller and ask for the original file, the batch number, and the laboratory’s own copy of the report.
The file is processed to produce your report and is not published or added to any public listing. Treat any document you upload anywhere as something you are willing to share, and redact anything on it you would not want read.
Methodology: how this tool reaches its answer
The verification runs in stages. The first four involve no artificial intelligence at all: they are arithmetic, date ordering, byte inspection and pattern matching, and they produce the same result every time for the same file. AI systems are used for two specific jobs that rule-based code cannot do, and their output is checked against the deterministic results rather than trusted on its own.
- 1. Reading the file as bytes
- For a PDF, the container is examined directly rather than through a viewer: which software wrote it, when it was created and last modified, how many times it has been saved, whether internal objects were replaced, whether annotations have been drawn over the page, and which fonts the text uses. A viewer is designed to show you the document as intended, which means hiding exactly this.
- 2. Recovering the text and its layout
- Where the document has a text layer, every run of text is extracted with its position and typeface. That makes two tampering signals findable that a plain text dump loses: a figure set in a typeface barely used elsewhere, and two pieces of text occupying the same position, which is what covering a value rather than removing it leaves behind.
- 3. Reading a certificate supplied as an image
- Many certificates arrive as pictures, and several laboratories publish them that way. An AI vision system transcribes the page into the same set of fields the text path produces, reading identifiers character by character. It is asked to transcribe and explicitly not to judge, and its transcription is then subject to every arithmetic and consistency check below, so a misreading still has to survive the numbers.
- 4. Checking the numbers
- Percentages must fall in a possible range. A printed peak table must sum to 100%. Peptide content, water and counter-ion cannot together exceed the whole. Dates must run forwards and cannot be in the future. A reported mass is compared against the accepted mass for the compound named, computed from its sequence where that is known, and cross-checked against other compounds in case the mass belongs to a different molecule than the one on the letterhead.
- 5. Comparing against the laboratory’s own format
- Where the issuing laboratory is recognised, the document is compared against the layout that laboratory actually issues, and its identifiers against the formats it actually uses. A certificate invented from scratch, or one wearing a laboratory’s name over a different layout, usually fails this. A genuine report with one edited figure passes it, which is why it is treated as corroboration and never as proof.
- 6. Asking the issuing laboratory
- A small number of laboratories publish a register that can be queried with the identifiers printed on the report. Where that is possible it outranks everything else, because it does not depend on the document at all. Where it is not possible, the tool says so and never treats the silence as evidence.
- 7. Weighing it up
- An AI system receives the results of all the stages above, together with the document text, and is asked to weigh them and explain them. It cannot overturn the arithmetic: a finding that is a contradiction stays a contradiction regardless of what it concludes, and only a successful register lookup can produce the highest verdict. Its own visual observations are recorded as advisory and are deliberately prevented from outranking a confirmed laboratory record.
What this tool cannot tell you
A verification tool that overstates its reach is worse than none, because it converts uncertainty into false confidence. These are the boundaries, stated plainly.
- It cannot tell you what is in your vial
- It examines a document. Even a certificate confirmed by the issuing laboratory describes the material that laboratory received, which is not necessarily the material you were shipped. Only testing your own sample answers that question.
- A clean result is not a guarantee
- It means nothing checkable is wrong. A competent forgery of a laboratory that publishes no register, on a document with no internal contradictions, can pass. That is why the tool distinguishes "consistent" from "verified" and will not award the latter without an independent confirmation.
- Absence of a check is not a failure
- A laboratory with no public register, a query blocked by bot protection, or a document that carries no identifier all mean a check could not run. None of them counts against the certificate, and they are reported separately from actual defects for exactly that reason.
- An image carries less evidence than a document
- A picture of a certificate has no metadata, no save history and no internal structure. The fields can still be read and every arithmetic check still runs, but the forensic layer is unavailable. The original file is always the better input.
- It does not assess the laboratory
- Recognising that a laboratory issued a report is not an endorsement of its competence, its accreditation scope, or its methods. Those are separate questions, and a genuine certificate from a weak laboratory is still weak evidence.
All Volta Peptides materials are supplied for in-vitro laboratory research use only. This page explains how to read analytical documentation and is not medical advice, not a recommendation to acquire any substance, and not a statement about the safety of any material. Nothing here describes administration to humans or animals.