Corrections and changes log
We make mistakes, and this is the record. It lists every material error we have found in our own pages, data or grades, and every change to how grades are calculated, newest first. Errors that made a product look better are logged the same as ones that made it look worse. Our editorial policy explains how we source claims and handle brand requests.
Found something wrong? Email hello@shilajitdb.com with the page and what is wrong. If you hold a lab report for a product we have graded low or left out, send it: reports are read against the same rubric as every other, and a changed grade is logged here.
17 entries. Reconstructed from our records back to 16 September 2026; the method before that date is not logged.
- Method
Editorial policy, this log and downloadable data published
Added the editorial policy, this corrections and changes log, a write-up of our citation audit, and CSV/JSON downloads of every grade. Entries below are reconstructed from our records back to 16 September 2026; the grading method before that date is not logged here.
- Claim
Unsupported figures from Pandit 2016 removed from six articles
Six articles stated that Pandit 2016 enrolled 96 men and raised total testosterone by 20.45% and free testosterone by 19.22%, with no adverse effects. The published abstract (re-checked on Europe PMC) gives none of those figures, and the full text was not available to us. We removed them rather than leave unverified numbers on the page. The paper is marked unverified on /research.
- Grade
Lotus Blooming Herbs Authentic Shilajit: E to A+
A Certified Laboratories (A2LA) chemical COA dated 13 August 2025, with a separate microbial report and a lot number, was reviewed and scored 12 of 13 points (A). The same day the missing point, a stated country of manufacture, was added: India, taken from the brand's FAQ, which describes sourcing and purification in the Himalaya but does not use the words 'made in India'. The product page records exactly what the FAQ says. That took the product to A+.
- Disclosure
All affiliate links removed
The five Pürblack product links that had been affiliate tracking links now go to the brand's own pages, and every page that mentioned an affiliate relationship was rewritten. ShilajitDB earns no affiliate revenue.
- Data
Duplicate evidence rows and mismatched BBB profiles removed
2,154 of 2,345 evidence rows were exact copies of one another (a re-import had re-added them), so our rule of two evidence sources per page had mostly been counting duplicates. We also found that Better Business Bureau profile links for five brands pointed to a different company. 2,159 rows were deleted, the rule now counts distinct sources, and pages that no longer met it stopped being indexed.
- Grade
Essencraft resin: C to A, and capsules added
The brand's lab-testing page was read and its resin report (Eurofins Analytical Services India, 17 August 2026) replaced an older screenshot. The brand's capsule report was also recorded: its arsenic, lead and mercury values are well above the resin's.
- Grade
Seven Pürblack products: A to A+
Their stored COA links pointed to Pürblack-letterhead documents that named no laboratory. The Cambium Analytica reports (ISO/IEC 17025, published 19 March 2026) were located and read; each carries a batch code, numeric heavy metals on the finished product and a microbial panel.
- Method
GMP claims no longer score
201 of 243 products claimed GMP and earned a point on the claim alone. Checking 27 brands that were graded B or higher, or on a ranking page, found no verifiable certification (a badge, or a certificate with no facility named). GMP is still recorded as a brand claim but no longer counts. The overall maximum went from 14 to 13 points with thresholds unchanged, so a number of grades moved down a band.
- Citation
13 of 37 PubMed identifiers pointed at unrelated papers
The citation text was right and the identifier was wrong, so nothing looked broken. 24 citations across 11 articles were affected; one safety-review identifier was wrong in 24 articles. All were corrected, and a script now checks every cited paper's author and year against PubMed.
- Claim
12 of 41 claims misstated the paper they cited
Examples: a rat study was presented as a human trial in six articles; a trial result that went down was described as having risen; an open-label single-arm study was described as randomised and placebo-controlled; a seaweed iodine analysis was quoted as warning of a risk that the paper says does not arise at normal servings. All were corrected.
- Claim
Iron-deficiency anaemia claim removed
Three pages said a clinical study found improved haemoglobin in women with iron-deficiency anaemia after 12 weeks. We could not find any such study in PubMed or Europe PMC. The claim was removed, the pages now say no human trial has tested shilajit for iron status, and the evidence rating for that topic was lowered from Moderate to Preliminary.
- Claim
Wrong pharmacopoeia chapter and limits in the heavy-metals guide
The heavy-metals article labelled its limits column with the chapter for drugs rather than the one for dietary supplements, and its lead limit matched neither. The chapter and limits were corrected.
- Method
Brand and contract-manufacturer names no longer count as an independent lab
Ten products had earned the 'independent testing lab named' points for a value that was the brand's own name, a contract manufacturer or a placeholder. Those values were cleared. Ten transparency grades fell by one band; no overall grade changed.
- Data
Misspelt laboratory names corrected
'Laboratories' was misspelt in 260 evidence entries and 16 product records (one spelling appeared 168 times). No grade moved.
- Grade
Mars by GHC: A+ to B; Pure Indian Foods: A to B
We read both products' COAs against the fields our database held. Each was issued by the brand or its manufacturer and names no independent laboratory (Mars reports heavy metals as 'complies' against a limit rather than as measured values). Both had been recorded as independent, named-lab reports, and both were regraded. Mars also left four ranking pages.
- Disclosure
'Unaffiliated' claims removed while an affiliate link existed
The homepage, footer and an indexed methodology answer described the site as unaffiliated while some Pürblack links were affiliate tracking links. Disclosure labels were added on 5 September; the last of the contradicting wording was removed on 16 September. All affiliate links were removed on 9 October (above).
- Method
Grades rebuilt around what the lab report documents
The earlier formula gave 9 of 14 points to signals that are not testing (resin form, US manufacturing, a patent claim), ignored heavy-metal results, and made A+ impossible without a patent claim. The new formula scores a verified independent report, numeric heavy metals on the finished product, a laboratory named on the report, a microbial panel, a batch code, a recent date and a stated country. Product form now caps the top grade a product can reach instead of adding points. 194 grades or tiers changed.
Known open items
What we know is unresolved or unchecked today.
- Pandit 2016 (testosterone in men) is cited from its abstract only. Its sample size and effect sizes are not in the abstract and we have not read the full text, so those figures are left out. /research flags it as unverified.
- A dose range attributed to the Stohs 2014 safety review has not been checked against the full text. /research flags it as unverified.
- Our citation script checks that an identifier matches the paper's author and year. It cannot tell whether a sentence matches what the paper found; that check is done by reading, and so far only on the claims described in the audit write-up.
- Product data is gathered from brand websites, Amazon and UK retailers. A brand may hold a lab report we could not find; if you have one, send it to hello@shilajitdb.com.