Corrections.

When we publish something that is wrong, we fix it, we date the fix, and we write it down here. That includes the errors nobody else caught — most of the entries below came out of our own review process, and they are listed exactly the same way a reader-reported error would be. A correction log that only records the mistakes someone else noticed is not a correction log. Newest first.

October 17, 2026 — FHO recovery: a plateau we called a milestone, a five-dog figure quoted alone, and a hip-replacement price we could not source

Re-reading every source behind our FHO recovery timeline found these errors. We described the 30-to-90-day figure as when a dog uses the leg “consistently.” The trial defines it as the point where weight-bearing is full or the remaining limp stops changing, and 13.3% of its dogs never reached full weight-bearing. We quoted 1.8 days to first weight-bearing as the effect of pain control without saying it came from five dogs. The rewritten section now gives all six regimens and the follow-up trial. We called a 27-dog force-plate study evidence that size does not matter, without saying it excluded dogs under 10 kg, grouped dogs by weight rather than by treatment, and appears to share its dogs with the same team's swimming trial, which the paper does not mention. We printed $5,500 to $8,000 per hip for a hip replacement with no source; it is replaced by the one hospital fee page we could find. We wrote that the difference between a good and a poor long-term result “is largely made” in rehab, which no study we cite shows. And a source entry carried a PubMed number belonging to a letter about the paper rather than the paper itself. The page also gains a section on the question search data shows readers bringing to it, a limp in week three.

October 17, 2026 — A ramp width we borrowed from its bigger sibling, then used to “correct” a page that had it right

Our Solvit telescoping ramppage gave the standard model's width as 20 inches. PetSafe's own page for that exact ramp (SKU PTV17-16898) gives 43 cm, or 17 inches. Twenty inches is the width of the longer Extra Long model. In August we had already caught the same mix-up with the ramp's weight, where an 18-pound figure turned out to belong to the Extra Long. This time it did more damage. On September 3 our Pet Gear tri-foldpage said the Solvit was “roughly 17 inches,” which was correct. We “corrected” it to 20 to match our own Solvit page, and logged that as fixing a contradiction. One article repeated the wrong figure as well. All three pages now carry the maker's figure. The Pet Gear is two and a half inches wider, not half an inch narrower. We also changed a sentence that said the Compact model was four inches narrower than the standard ramp. It is one inch narrower. The same check settled the Extra Long's length, which our table gave as 89 inches. PetSafe says 47 to 87. The lesson we take from it: when two of our own pages disagree, the tiebreak is the maker's page, never the other page. The check we run before every publish now fails if the 20-inch figure reappears.

October 10, 2026 — The limping guide: “Not a hip. A knee.” and five other sentences that said more than their studies

Our guide to telling a limp from weakness opened on a study of 369 dogs whose lameness had been put down to hip dysplasia, and in 32% of which a torn cruciate ligament was found. We summed it up as “Not a hip. A knee.” The same abstract says 94% of those dogs also had hip dysplasia on their X-rays. The hips were real, and the study judged the knee to be the main cause of the lameness. The article now says so. Re-reading every source on October 10 also found these errors. We cited a study of an induced swinging-limbproblem for the head-bob and hip-hike signs, which are weight-bearing signs; it is replaced by the same group's weight-bearing study, whose numbers show the hip on the sore side dropping by about as many millimetres as the head moves, not less (they are different measurements). We described a weight-distribution measure as a share of body weight, and said earlier studies agreed with it when one did not. We described an arthritis screening population as routine visits only. We gave a surgery-timing difference as an adjusted result when the percentages were the raw ones. And we stated that neurologic weakness is indifferent to rest, when the source says a constant gait change can be either. We also added the only randomised nail-grip trial, which was null, beside the traction product the guide recommends, and told readers to photograph the nails before the grips cover them.

October 10, 2026 — Cosequin: a xylitol all-clear that rested on a list we had withdrawn

On September 1 we withdrew an inactive-ingredient list for the Cosequin tabletthat we could no longer trace to a source. One sentence further down the page still read “No xylitol appears on any label reproduction we read,” which was an all-clear resting on that withdrawn list. Nutramax publishes no inactive-ingredient list for this tablet, so we cannot clear xylitol from the maker's own text, and the page now says that. The same page said the glucosamine “is shellfish-derived” as fact, but Nutramax does not state the source. And it said the tablet costs “well under a dollar a day even in a large dog,” when during a large dog's loading weeks on the bottle we link it costs $1.00 a day.

October 4, 2026 — 27 product pages told machines a price band like $1 to $2,026

Each product page carries a block of structured data that search engines and AI assistants read in place of the page: name, brand, our score, and a price band. Ours took that band from every number in the product's price note. Since August those notes have carried dated receipts, so the year became the top of the range. On October 4, 27 of our 46 product pages were wrong. The lift harness we recommend, $95 to $155 on Amazon, was described to machines as costing $3 to $2,026. A person reading the page saw the right prices. The machine copy, which is the one an assistant quotes, was wrong.

The band now counts only dollar amounts, and it skips amounts that are not prices (“$6.25 a pad”, “$49.05 under”, “$11 cheaper”). Article pages and product pages now share one parser. The weekly price check now reads the band through that same parser, so it tests the number machines see, not a second reading of the prose.

October 4, 2026 — Coolaroo does publish the fabric size of its elevated bed, for every size

From September 26 our bed sizing audit said that no sleeping-surface figure was published for the Coolaroo elevated bed, on either the Amazon listing or coolaroousa.com. That was wrong, and our own Coolaroo product pagehad printed the figure since August 20. Each size page on the maker's site has a “Fabric Dimension” line: Small 27.5 × 21.6 inches, Medium 35.4 × 25.5, Large 43.3 × 31.5, X-Large 52 × 37, all read October 4, 2026. The audit had not re-read the maker's pages when it made the claim, and it should have checked against our own page.

The audit now gives the figures with their source, and it counts Coolaroo with Casper and Orvis among the makers that print an interior for every size. The elevated beds articlenow gives the fabric size for every size too, not just the Large. The same day we found that the Large in grey we had linked on Amazon was no longer for sale. The card now links the same Large in Brunswick Green, which Amazon quoted at one to four weeks to ship, and it puts the maker's own store first, because that is the one shipping today.

October 4, 2026 — The elevated beds article: a cold-weather claim our own guide contradicts, and a reason to buy direct that did not exist

Our elevated beds articlesaid that “cold is when stiffness peaks” and that arthritic dogs “tend to do worse in cold,” and linked to our own cold-weather guide. That guide finds the canine evidence for this close to a null result: in the best study, temperature and daylight together explained under 1% of the variation in how much arthritic dogs moved. The article now rests its winter advice on what we can defend, which is that a cot keeps air moving under a still dog for hours. It also says plainly that this is reasoning, not a measurement.

The same article said “we would still rather you paid less direct” about a Big Barker that costs the same on both channels ($249.95 flat, $289.95 Headrest, re-read October 4). The only thing buying direct adds is the 180-night trial, and the article now says that. The same review also corrected an invented “month fourteen” for when cot mesh sags, a summary of customer complaints we could not source, a shelter study stretched to cover mesh beds and arthritic dogs (now removed), and overstatements of the heat-loss, thermal-comfort and pressure papers. It had also called Big Barker's a warranty “someone is prepared to honour,” when the published terms are a few lines long, with no claim procedure and nothing about Amazon purchases. The article now says exactly that.

October 3, 2026 — A harness price we retired in August was still printed as current on ten pages

In early August we verified, twice, that Help 'Em Up sold its harness direct at a flat $80 in every size, and we printed that on many pages. By August 14 the maker had moved to pricing by size — $80 X-Small, $95 Small, $110 Medium, $125 Large, $140 X-Large — and we corrected the product pageand some of the articles. On August 18 we also found that two follow-on figures were wrong: the Amazon premium is a uniform $15 at every size, not “$15 to $75,” and ordering a smaller rear half carries no premium at all, because each half is sold at half the matched price. The “$95” we had quoted for split sizing was a headline figure on the maker's ordering page, and reading it as a surcharge was our mistake.

We fixed the product page and did not sweep the rest of the site, and no automated check was added. On October 3, ten articles still printed one or more of the three dead figures as current, one of them under the words “we re-checked the pricing today”: tracking whether a limp is improving, getting a dog into a car, Great Dane joint problems, muscle wasting on one back leg, the lift harness comparison, senior dogs with weak back legs, the recovery gear guide, the TPLO gear checklist, how to tell if your dog is in pain and degenerative myelopathy. Every one now gives the prices we re-read on October 3 and says on the page what it used to say. All three figures are now in the check that fails the build when a retired claim comes back. We tested that check against the old versions of all ten pages, and it flags every one of them.

September 26, 2026 — A three-cadaver pressure study stated more firmly than its authors did, on fifteen pages; and a bed-sizing rule that grew back

Fifteen of our pages cited the 2019 Journal of Small Animal Practicepressure-mapping study by Caraty and colleagues as showing that “the thinnest dogs produced the highest peak pressures on every surface tested,” and several added “with the least muscle left to pad those points.” The study laid three cadaver dogs of similar size and different body condition on four surfaces. Its authors write that the contact pressures “appeared to depend on body condition score, with maximal pressures observed in thin dogs.” They do not say “every surface,” and body condition score measures fat cover, not muscle. We had softened this on our bed roundup on September 6 and on the bed replacement guideon September 19 without sweeping the rest of the site, which is how it survived. It is now corrected on every page in the authors' own terms; where a page still connects the finding to muscle loss over a sore hip, it says that step is our inference. The automated check that fails the build on retired claims now includes this one.

The same day we found our bed-sizing rule had grown back in the wrong shape. On August 18 we corrected eight pages that said to add four to six inches to a dog's measured length, when the size chartsays 8 to 12. By today “add four inches” or “add a few inches” was back on three: the Big Barker and Tempur-Pedic product pages and the Tempur-Pedic vs Sealycomparison. All three now give the chart's rule, and it is now enforced by the same check instead of depending on memory.

September 15, 2026 — A retired “direct-only” claim survived on one page for a month; the check that should have caught it was blind to a decimal point

On August 13 we retired the claim that Big Barker's flat Sleek edition was sold only on the maker's site: it appeared on Amazon that day at the same $249.95, and we corrected it on the product page and the pages that compared it. One sentence in the Big Barker vs Bully Bedscomparison kept saying the Sleek Large “is direct-only” from August 13 until today. The automated tombstone check that exists to stop retired claims coming back had a window that ended at any full stop, and the full stop inside “$249.95” ended it before the offending words. A reviewer caught the same claim being re-drafted in today's warranty article; we fixed the draft, widened the check so a decimal inside a price no longer hides a sentence, re-ran it across the corpus, and it found the month-old survivor. Both are corrected in place; the Sleek Large is $249.95 on both channels, verified 15 September 2026.

September 14, 2026 — A within-group result printed as a head-to-head: the 5.6% vs 0.4% fish-oil diet figure

On two pages — the fish-oil dosing article (August 25) and the Welactin product page— we described the 2010 JAVMA force-plate trial of a fish-oil therapeutic diet as showing peak vertical force “rose 5.6% against 0.4% on control food,” and called it the strongest omega-3 result in arthritic dogs. Both numbers are real. Printed side by side like that, they read as a head-to-head comparison, and they are not one. The paper's abstract reports that the 5.6% change was significantwithin the diet group and the 0.4% was not significant within the control group. It does not report a significant difference betweenthe two foods, and the independent Helsinki group that later reviewed it describes it as finding “no significant differences between treatment and placebo groups.”

The dosing article also said that “no equivalent trial exists” for a fish oil added to ordinary food. Two placebo-controlled trials of oil over food do exist (Hielm-Björkman 2012, 77 dogs, no separation from corn oil on a force plate; Mehler 2016, 78 dogs, exam scores improved against mineral oil). We had not read them before writing that sentence.

What changed: both pages now state the within-group nature of the 5.6% figure, name the two bottle trials, and link to the new reconciliation, where every omega-3 trial in either family is tabled with its instrument, its between-group result and its funder. The direction of the error ran in favour of the product category we earn from. We found it ourselves while reading the trials for the new article.

September 13, 2026 — We said the Comfy Cone was not being wound down; a week later it is unbuyable above X-Small

On September 6 our Comfy Cone pagereported thin Amazon stock on the soft e-collar, moved our link to the one long size still buyable, and added: “This is not a discontinuation notice … the pattern here looks like ordinary supply rather than a product being wound down.” That was a guess dressed as a finding, and we should not have printed it.

On September 13 we pulled the entire sixteen-variant family through the Amazon Creators API instead of checking the sizes one at a time. Two X-Small listings had 6 and 16 units. Every other size — Small, Small-Long, Medium, Medium-Extra Long, Large, X-Large, XX-Large, in both colours — was “Currently unavailable,” “Temporarily out of stock,” or returned no offer at all. The size we had linked seven days earlier was dead. The maker's own site has no cart and points buyers at that same Amazon storefront.

What changed: the Amazon button is switched off on the product page and on the four articles that card the cone, the variant-by-variant table is published with both dates, and the sentence predicting ordinary supply is gone. What we still do not know: whether this is a gap or the end of the product. Chewy lists every size, but its pages rate-limited every check we made, so we are not claiming Chewy has stock and we are not linking it. We will re-check the full family weekly and record the result on the page rather than predict it again.

A second, internal error from the same pass belongs here because it nearly reached readers. The --writemode of our weekly stock check cleared the “unavailable” flag on three products we had deliberately retired — two whose Amazon listings now sell a different brand's product entirely, and one discontinued coat — because those listings are technically in stock. We caught it in the diff before deploying, restored the flags, and changed the script so it can set a flag but never clear one. “In stock” and “the right product” are different checks, and the tool now knows it.

September 6, 2026 — The site footer claimed a vet review that does not exist

Our vet reviewers pagesays plainly that no veterinarian reviews this site, that every article carries a “vet review pending” tag, and that we would rather name the credential we lack than imply one. All of that was true. It was also being contradicted, on the same page, by the footer underneath it, which read “Reviewed by licensed veterinarians.”

A weekly maintenance pass found the claim alive on five surfaces, all of them site-wide or near it: the footer on every page; the site's root meta description (“Vet-reviewed guides…”), which is what a search engine or an AI crawler reads first; the OpenGraph card every shared link renders, which said “Tested · Vet-Reviewed · No Paid Placements”; the beds hub title and its page metadata; and a line on our about page. Every one is corrected. The OpenGraph card now says “Sourced · Corrected in Public · No Paid Placements,” all three of which we can evidence.

The same pass found a second survivor of an older correction. Our affiliate disclosurestill described “testing notes from our test home” and photographs of “products we test in-house.” The invented-testing family of claims was deleted from the articles in August 2026; it had survived here. The only hands-on testing we do is the bed protocol at how we test beds, and the page now says so.

The reason all six survived a system built to prevent exactly this is worth recording, because it is the actual defect. We run an automated check that scans the corpus for claims we have publicly retired and fails the build if one reappears. It was passing every day. It only ever read the articles and product pages — never the footer, the layout metadata, the share card, or the hub pages. The most-published text on the site was the least-checked. The scanner now walks the site chrome as well, and “vet-reviewed” is a tombstoned phrase that will fail the build anywhere it reappears outside an explicit statement that no reviewer exists.

September 6, 2026 — Paw.com does publish the PupRug's foam depth, in its FAQ

Since August this site has said, on our orthopedic bed roundup, our floor traction guideand the PupRug product page, that Paw.com publishes no foam depth for the PupRug anywhere. That replaced an older and worse error — a “3 inches of memory foam over 1 inch of support foam” figure we could not source. The replacement was wrong too.

The depth is published, in Paw.com's own FAQ: “Our PupRug™ Beds, including the PupRug™ Faux Fur Orthopedic Beds, PupRug™ Runners, and PupRug™ Faux Animal Print Beds have a memory foam base that measures approximately 2 inches thick.” Read September 6, 2026. It appears on none of the ten product pages and on no Amazon listing, which is why we missed it twice, but a figure in a company's help centre is published and we should have found it.

Two inches changes the recommendation, so we have changed it. On a page whose whole argument is that depth is the number to buy on, this is the shallowest bed we rank — a Big Barker's published build is 7 inches in three layers. Two inches of memory foam is a real cushion for a small or middleweight dog on a hard floor and is not an orthopedic mattress for a heavy one, because what matters is what is left under the hip after the dog settles. That is precisely the complaint owners of medium and large dogs have been making in reviews for years, and it was dismissible only for as long as there was no published number to test it against. Our score on the bed drops from 8.6 to 7.6, it is now explicitly not our recommendation for a dog over roughly 50 lb with sore joints in any size, and it keeps its place as the pick for a room that has to look good. The bed did not change. We finally found the number.

The same FAQ contradicts the product pages on the warranty. The product pages advertise a “LIFETIME no-flat-foam guarantee.” The FAQ says “We offer a ten-year guarantee against flattening for all of our memory foam mattresses,” claimed by emailing a photo of the flattened foam to support@paw.com. Both were live on the same day. We now publish the disagreement rather than the flattering half, and we say which one we would plan around: ten years, because it is the one with a procedure attached.

August 19, 2026 — Manganese, wrong on two product pages in opposite directions

Writing up the question of whether Dasuquin and Cosequin can be given together meant adding up every ingredient a stacked dog would receive, and the addition did not reconcile with our own pages. Two errors, made independently, pointing opposite ways. Our Dasuquin with MSM page listed “manganese ascorbate” in its materials line. That tier does not contain manganese: read on dasuquin.com on August 19, the Dasuquin with MSM soft chew panel has eight rows — glucosamine, MSM, chondroitin, total omega-3s, EPA+DHA, ASU, Boswellia, green tea — with no manganese row and none in the other-ingredients list. Manganese is printed and quantified one tier up, in Dasuquin Advanced, at 1 mg per small/medium chew and 3 mg per large chew. And our Cosequin page said the linked MSM & HA tablet adds 3 mg of manganese over the classic Plus MSM tablet. It does not — both tablet lines carry the same 3 mg of manganese ascorbate, and hyaluronic acid is the only difference between them.

The practical effect of the first error was that anyone computing a stacked dose from our pages would have counted manganese twice. Both lines are fixed, both product pages now carry the correction in the body rather than only in a changelog, and the article that found it publishes the corrected arithmetic in full: a dog over 60 lb on Cosequin plus Dasuquin with MSM receives 9 mg of manganese a day at loading, all of it from the Cosequin. We do not know whether that number matters clinically in a dog — no regulator publishes an upper limit — and the article says so.

August 13, 2026 — Dasuquin was the product in the trial, and we had not said so

Our Dasuquin with MSM page said the evidence for the category is “strongest in the laboratory and weakest in blinded canine trials.” That was true as far as it went, and it left out the specific fact that matters most. Reading the full text of Kampa et al. 2023 rather than our own summary of it, the trial names its glucosamine arm outright: “DASUQUIN® with MSM Soft Chews, Nutramax Laboratories Consumer Care, Inc.” — the large-dog formula we link, at about 30 mg of glucosamine per kilogram daily. Its change in peak vertical force at week four was −0.05, against +0.08 on placebo, while two green-lipped mussel lipid extracts and carprofen all improved significantly. A second blinded trial by a different group, using activity monitors over 97 days in 60 dogs, reached the same answer.

Dasuquin is our second-best-earning recommendation, so to be explicit about what we have and have not concluded: the trial used 15 dogs per arm and was powered to detect the mussel extract's effect, not to exclude a smaller one, and it ran maintenance dosing through the four-to-six-week window Nutramax's own label says to double. Those are real defences and we have published them alongside the finding. What we have stopped doing is describing the product as well evidenced. Its score comes down from 9.2 to 8.8 — the product has not changed, our reading of it has. In the same pass we fixed a stale YuMOVE claim in our best-joint-supplements article, which still cited the unpublished UK trial we retired on August 12; the earlier pass corrected two other articles and missed that one.

August 14, 2026 — A harness claim reversed by the gait literature; a coat recommendation withdrawn

We had written, on our Easy Walk harness page and in our French Bulldog guide, that non-Y-front strap layouts sit across the shoulder “more than the best-fitting harnesses do.” The one published gait comparison found the opposite: Y-front designs restricted shoulder extension more than strap-over-shoulder designs (Lafuente et al., Veterinary Record2019;184(2):64). Both pages now state the reversal and cite the trial. The same pass corrected “five adjustment points” to the manufacturer's own figure of four, and renamed the page to the Deluxe variant the listing actually sells.

Our Gentle Leader page said the head collar “includes a training DVD” (the current listing ships none) and was “developed by a board-certified veterinary behaviorist” — a rank we could not verify, so we now print what is checkable instead: R.K. Anderson's 1980s head-collar patents. Its score fell from 8.4 to 7.8. Our training-methods guide repeated the board-certified line and two defaults (“the harness most R+ trainers reach for,” a claimed KPA/CCPDT curriculum standard) that no survey substantiates; all are removed, and a long-line “whiplash risk” warning is now labeled what it is — a commonsense caution with no published injury data either way.

The Back on Track mesh coat is no longer recommended at all. We had recommended it on “anecdotal results from rehab vets,” which is not a source. There is no controlled canine evidence for the far-infrared mechanism, the one live listing ships in six to seven months from another continent, and the maker's current 1,690-variant catalog no longer contains the product. The page stays up as a dated record of the mistake, scored accordingly.

Two smaller ones, same day. Our ZIWI Peak page called full-feeding “comparable to fresh-cooked subscriptions”; at the maker's published 4,900 kcal/kg and the verified $52.76/kg price, a 70 lb dog costs about $16 a day — above most subscriptions, and the page now shows the arithmetic. And a morning hedge about the Solvit ramp resolved the same afternoon: we wrote that petsafe.com returned 404 on every URL we tried — the site is live (our URLs were wrong), the 400 lb rating is re-verified on the maker's Canadian page, and we learned the standard model has quietly left PetSafe's US catalog, which the product page now discloses.

And two from the same day's gear pass. Our Walkin' rear-support harness page (score now 6.8, from 7.9) had recommended it for degenerative myelopathy; the maker's own catalog no longer sells the harness as of 14 August 2026, and for a progressive disease the maker itself points buyers to its front-companion system — our page now says both. Our Comfy Cone page said the liner was washable; All Four Paws' own product page sells disposableliners and the care instruction is “wipe clean” — corrected here and, earlier, in our TPLO recovery guide.

August 14, 2026 — Invented testing and a fabricated journal name, removed

A citation-repair pass over our oldest pages found our flagship bed guide saying the Big Barker pilot study was “published in The Canine Health Foundation.” It was not published in any journal; it is a manufacturer-hosted study report, and our page now links the primary PDF directly with the design caveats we apply everywhere else (open-label, single-arm, owner-scored). The same pass removed an unsourced claim that a good bed “pays back in fewer vet visits and less NSAID use,” an invented foam-density shopping rule, and a “consulted two veterinary orthopedists” line we could not substantiate.

Separately, our PupRug page claimed testing “under three Labradors and a Pyrenees mix.” That testing did not happen — the same defect we removed from our Dasuquin page on August 7 — and the sentence is gone. The page had also misstated the product in both directions: it said spot-clean-only when the maker's own care copy says the cover machine-washes, and it said “1 year” where the maker advertises a lifetime no-flat guarantee — which we now note carries no measurable threshold at all. A foam-depth spec we printed appears nowhere the product is sold and has been removed. Our PetFusion warranty line said three years; the live listing says 24 months.

August 13, 2026 — Furhaven core prices; the Big Barker “Amazon markup” retired

We had written that Furhaven sells replacement foam cores at “$13.99 in every size.” That is the Small's price. The cores are priced by size — $13.99 Small up to $69.99 Giant, $25.99 for the Large most readers of these pages would buy — and the error came from reading a “from” price on the parts page instead of the per-variant feed. It appeared in four articles, the product page and our internal planning documents; all are corrected, and the ownership arithmetic survives with the right numbers (a Large plus a core every 18 months still runs about $26 a year).

Separately, we retired one of our own most-repeated findings. We had reported Big Barker's Amazon listing running about $40 above the manufacturer's own site. Checked like-for-like on 13 August, the edition Amazon sells (Headrest, Large) costs $289.95 in both places — our comparison had been putting Amazon's Headrest price against the direct site's cheaper Sleek edition. The accurate version of the finding, as of that morning: the $249.95 Sleek edition was not sold on Amazon, so buying direct still saved money via edition choice. By the same evening a Sleek-priced flat Large had appeared on Amazon at the identical $249.95 — rebrand-era listings churn daily — so the durable version is simpler still: the channels match on price, and the 180-night trial only comes with a direct purchase. Corrected wherever we said it, including the summary file we publish for AI assistants, both times.

August 12, 2026 — ASU evidence overstated; an unverifiable YuMOVE trial claim

Our Dasuquin-versus-Cosequin comparison said ASU “has its own supporting trials for protecting cartilage” — in three places, including an FAQ answer that feeds the structured data assistants read. That overstated the evidence, and it contradicted our own Dasuquin product page, which had carried the more defensible line since August 7. The strongest ASU work is laboratory and animal-model evidence; the one randomized double-blind canine trial we can name tested ASU only inside a combination product with fish oil, so it cannot isolate ASU's contribution. All three sites now say that, with the trial cited by DOI.

In the same pass we stopped repeating YuMOVE's “independent UK clinical trial” claim, which appeared on our product page and in two articles. The study it refers to is unpublished and the college unnamed — nothing is indexed on Europe PMC — so it cannot be checked by us or by you. We also found the manufacturer's own dosing page (three chews a day for a 70-pound dog, about $426 a year) disagreeing with its Amazon listing (“just one chew a day”) by a factor of three, and a sentence on our page claiming palatability testing we never did. The product's score came down from 8.4 to 7.9 for the disclosure gaps; the price range and the money-back guarantee's real conditions (six weeks of use, then a 30-day claim window) are now stated.

August 11, 2026 — Misattributed journal in a neutering citation

Our guide on when to spay or neuter cited the Golden Retriever Lifetime Study gonadectomy analysis to the Journal of the American Veterinary Medical Association, with an approximate title and a vague “2019–2024 publication series” date. The paper is Simpson M, Albright S, Wolfe B, Searfoss E, Street K, Diehl K, Page R, “Age at gonadectomy and risk of overweight/obesity and orthopedic injury in a cohort of Golden Retrievers,” published in PLOS ONE in 2019, 14(7):e0209131. The finding we drew from it was correct; the reference pointing at it was not, which meant a reader trying to check us would have gone looking in the wrong journal.

Found while researching a new breed page, and worth saying how: a citation that is slightly wrong is harder to catch than one that is invented, because it survives every check except actually resolving it. We now verify references against the DOI record rather than against the shape of a plausible citation.

August 10, 2026 — Zesty Paws ingredients

We described Zesty Paws Mobility Bites as adding turmeric and hyaluronic acid to a glucosamine, chondroitin and MSM base. It contains neither. The product discloses a single 1,510 mg “Proprietary Hip & Joint Blend” holding four ingredients — glucosamine HCl from shellfish, OptiMSM, chondroitin sulfate and vitamin E. We had been crediting it with two ingredients it does not have, in a product page and in an article comparing it against Dasuquin, where the claim also sat inside the structured data that search engines and assistants read. Both are corrected.

Checking it turned up two things worth more than the original claim. Because the actives sit inside a proprietary blend, the glucosamine and chondroitin amounts are not published at all, so no one can check them against a studied dose — and the one active that does carry a number is contradicted by its own maker, with zestypaws.com stating 500 mg of OptiMSM per two chews and the Zesty Paws Amazon listing stating 400 mg, both read on the same day. We publish the disagreement rather than picking one. We also had the price stale: $32.97 verified, not the ~$37 we listed, which at the label serving works out to about $267 a year for a 70-pound dog — more than Dasuquin with MSM at maintenance, so our framing of this as the cheaper alternative was backwards.

August 10, 2026 — Unsupported satisfaction guarantees

Our Dasuquin and Zesty Paws pages both carried “Manufacturer satisfaction guarantee” in the warranty field. Neither was substantiated, and the Dasuquin one is close to the opposite of the truth. Nutramax's published policy asks you to report within 10 business days of receipt, and between 10 and 30 business days accepts a return only if the product is unopened, unmarked, undamaged and has a year left before expiry. The label prescribes a 4–6 week loading phase — so you cannot run the trial the product asks for and still be able to send it back. That is now stated on the page. For Zesty Paws we could not verify any guarantee, so we no longer claim one. This is the same defect we fixed on Cosequin earlier the same day and did not carry across at the time.

August 8, 2026 — Welactin dosing

We described Welactin as a pump bottle delivering 200 mg EPA and 130 mg DHA per pump. Current bottles are not dosed that way: they use a 6 mL scoop delivering 1,610 mg of total omega-3 fatty acids, of which 1,330 mg is EPA and DHA. Every number we had built on the old figure was therefore wrong, including how long a bottle lasts — roughly seven weeks for a 60-pound dog, not the month we had published. We corrected the product page and five articles, and reworked the arithmetic in each rather than only swapping the numbers. In the same pass we removed a claim that the product is third-party tested for heavy metals, which we could not verify, and a refrigeration instruction that does not appear on the manufacturer's label.

August 7, 2026 — Dasuquin

We had published a sentence saying we used Dasuquin on a test dog after TPLO surgery. That testing did not happen, and the sentence is gone. We also overstated the evidence: we wrote that ASU is clinically shown to slow cartilage degradation, when the support for it comes from laboratory and animal-model work, and that is how the page now reads. An unsourced claim that it is the most researched joint supplement has been removed. The score came down from 9.4 to 9.2.

August 6, 2026 — GingerLead pricing

We described the GingerLead sling as costing half what the Help 'Em Up harness costs. It does not. At $69.95 against $80 direct, the gap is about $10, and framing it as half price changed the decision we were asking readers to make. Corrected in five articles. While rechecking the sling we also found drawbacks we had never disclosed, and its score came down from 8.7 to 8.3.

August 5, 2026 — Help 'Em Up pricing

Our pages implied a single price across all configurations of the Help 'Em Up harness. The Mix & Match option — the configuration we specifically recommend for dogs with wasted hindquarters — costs $15 more, and we had never said so. Readers following our recommendation were budgeting from a price that did not apply to them. The premium is now disclosed wherever we make that recommendation. We also added a note that the Amazon buy box for this harness is held by a third-party reseller rather than the brand.

August 5, 2026 — K9 Ballistics bed guarantee

We wrote about the Tough Ripstop line in a way that implied its 120-day guarantee covers the whole bed against chewing. It covers the cover only. For a determined chewer, the Chew Proof Armored line is the one to buy, and the pages now say that. The score came down from 8.7 to 8.4. The same recheck turned up something in the brand's favor that we had missed and have now published: a 10-year warranty against the bed flattening.

August 3, 2026 — Misattributed research finding

A draft attributed force-plate correlations to the Helsinki Chronic Pain Index when those correlations belong to the CBPI questionnaire. The Helsinki figure was 0.128 and not statistically significant — close to the opposite of what the draft claimed. Our pre-publication review caught it and it never went live. We are listing it here anyway, because the review gate is part of what readers are trusting when they trust a number on this site, and a log that hides its near-misses tells you less about how careful we are than one that shows them.

July 30, 2026 — Pet Gear ramp weight capacity

Amazon carries multiple listings for the same Pet Gear ramp with weight specs that contradict each other. We had picked one and repeated it as fact. We now publish the disagreement itself and tell readers to check the capacity on the listing they are actually buying from — for a ramp holding an unsteady dog, that number has to be right for the unit in the box. We found the same pattern on Carhartt coat size charts on August 7 and handled it the same way.

Found something wrong?

Tell us. If you spot an error on any page, write to us through the contact page and point us at it — we would rather be corrected than be quoted wrongly for years. Corrections reported by readers are credited here by name when the reader is happy for us to do that.