How we test and score mad honey brands
Every brand is scored out of 100 on six weighted criteria, with 65% of the weight on evidence a buyer can independently check: published lab results, traceable sourcing and measured potency. Missing evidence scores zero for that criterion — we do not give credit for claims. The same rubric is applied to the brand we operate.
The six criteria and their weights
Weights reflect what actually goes wrong for buyers in this category. Dilution and unverified potency are the dominant failure modes, so evidence-based criteria dominate.
Lab testing & purity
25%Does the seller publish an actual certificate of analysis with quantified grayanotoxin values, a named laboratory, a method and a batch identifier — before you pay? A downloadable summary without numbers scores in the middle. A testing claim with no document scores low. Mad honey is an ingestible with a genuinely active compound whose concentration varies by an order of magnitude between batches. Unverified potency is the largest risk a buyer takes, so it carries the most weight.
Sourcing transparency & ethics
20%Can the brand name the district, altitude band, harvest window, bee species and the hunting community — and are those claims internally consistent? Generic 'Himalayan origin' copy with no district and no season scores low. Traceability separates cliff-harvested honey from repackaged commodity honey, and inconsistent origin claims are the cheapest lie to catch.
Authenticity & potency
20%Measured grayanotoxin content where published, checkable specifications where not, and consistency between what the label claims and what the evidence shows. Buyers fear paying premium prices for diluted honey — and unmeasured honey that is stronger than expected is the more dangerous failure.
Customer reviews & reputation
15%Review volume, whether the rating distribution is shown (including the low ratings), and whether counts look organically accumulated rather than asserted. Social proof matters but is gameable, so it sits below the evidence-based criteria. A brand that shows its one-star reviews scores above one that only shows an average.
Price & value per gram
10%Best available cost per gram across the brand's range, including shipping where it is charged. Sellers who publish no price at all are scored down. Cheapest is not best where dilution is the main fraud, but price still has to be defensible. We score it honestly even when it costs our own brand the category.
Shipping, guarantee & support
10%Published delivery times, warehouse location, and the strength of the refund or satisfaction guarantee. A guarantee covering only non-delivery scores below one covering dissatisfaction. It decides whether a good product actually arrives, and it is the most common source of one-star reviews in this niche.
How each criterion is scored
Each criterion is scored 0–100 against a fixed rubric, then multiplied by its weight. The rubric for lab testing, as an example:
- 90–100 — certificate published pre-purchase with quantified grayanotoxin values, a named laboratory, a stated method and a batch identifier.
- 70–89 — per-batch reports published or downloadable, but graded (“high”, “medium”) rather than measured.
- 50–69 — testing claimed, or a certificate promised with delivery, with nothing checkable before payment.
- 0–49 — a testing badge with no document, no laboratory and no method, or claims that contradict the available evidence.
The other five criteria use equivalent four-band rubrics. Where a claim cannot be checked, it does not earn points — an unverifiable claim and no claim score the same.
What we actually do for each brand
- Buy or obtain the product where possible and record batch codes.
- Request the certificate of analysis and note whether it arrives, what it covers, and whether it matches the batch.
- Record every sourcing claim on the site and pack, and check whether it descends below country level.
- Normalise price to cost per gram including shipping to the US.
- Log review volume, distribution and how the seller answers criticism in public.
- Time a pre-purchase support question and record the response.
Our conflict of interest, stated plainly
Compare Mad Honey is published by the Honey Hunter Association, which holds a commercial contract with Magaranger — the brand ranked first here. Every brand is scored on the same published rubric. Two things keep that honest and checkable: the rubric is published above, and we name the categories our own brand loses. In the current build, Sherpa Honey beats us on value and Real Mad Honey beats us on service. If you think a score is wrong, tell us — corrections are published with the date.
What we have and have not verified
The current scores come from a desk review carried out on 2026-07-21: we read every brand's storefront, product pages and lab-report pages, and recorded what a buyer can see before paying. That is deliberately the same thing the rubric measures — but it is not the same as holding the jar.
Checked on 2026-07-21
- Published jar sizes and prices, converted to best available cost per gram
- Presence of a certificate of analysis, and whether it carries quantified values, a named laboratory, a method and a batch identifier
- Origin claims below country level: district, altitude, species, community, season
- Stated shipping origin, delivery times and refund or satisfaction guarantee
- Visible review counts and whether the rating distribution is shown
- Health and medical claims made in marketing copy
Not yet done
- Physical purchase of each brand's entry jar and recording of batch codes
- Emailing each seller to request a COA and timing the reply
- Independent assay of any purchased jar
- Verification that Best Mad Honey's downloadable batch PDFs contain quantified values
- Verification that Medicinal Mad Honey's QR codes resolve to real batch records
Until the right-hand column is cleared, every page carries a note saying so. We would rather show you the edge of what we know than imply we have done work we have not.
- We score what a buyer can verify. A brand may run excellent internal testing and score poorly here simply because none of it is public.
- Grayanotoxin content varies between batches even from the same producer — published analyses of mad honey report grayanotoxin I concentrations spanning roughly a 90-fold range. A single lab result describes one lot, not a brand's permanent potency.
- Prices and stock move. Per-gram figures use the best available size and are accurate as of the last-updated date.
Brands we did not score
Silently dropping a brand looks the same as never finding it, so here is what was left out and why:
- Rare Mad Honey (raremadhoney.com) — The site returned an HTTP error throughout our research window on 2026-07-21, so no listing, price or claim could be recorded.
- Strong Mad Honey (strongmadhoney.com) — Identified too late in this scoring pass to review properly. Queued for the next quarterly update rather than scored on a glance.
- The Mad Honey (themadhoney.net) — Queued for the next quarterly update.
Frequently asked questions
How do you handle a brand that tests but does not publish the numbers?
It lands in the middle bands, not the top one. A per-batch report that grades potency as “high” scores in the 70s; a certificate promised with delivery scores in the 60s. Both are worth more than a bare badge and less than a published measurement, because only a measurement can inform the decision to buy.
Do brands pay to be included or ranked?
No. No brand can pay for inclusion, position or a score change. We link out to one brand — Magaranger, which we operate — and that link is marked as sponsored.
How often are scores updated?
Quarterly, plus immediately whenever a brand publishes new documentation or changes something material. Every page carries a visible last-updated date.
What would change Magaranger's score?
The same things that would change anyone's, in both directions. Publishing a current certificate for each harvest instead of the April 2025 one on display would move it up; so would a larger jar size, since it currently scores 44/100 on value against Sherpa Honey’s 96. Withdrawing the published values would cost it 25% of the total and second place with it.