How we test it · Monte-Carlo validation
How do we know the sachet actually works?
We don't check one perfect bowl on paper. We run each recipe across many thousands of simulated real-world bowls — each one slightly different, the way real kitchens and real groceries actually are — and confirm that every essential nutrient still clears the vet reference floor, and that nothing crosses a safe upper limit. This page shows what came back.
The claim, in one line
A complete diet has to do two things at once for every nutrient: never fall below the minimum an animal needs, and never rise above the amount that could harm it. The minimums come from the NRC 2006standard — the nutrient reference vets and nutritionists use. The ceilings are each nutrient's safe upper limit. Getting both right, on an ordinary day in an ordinary kitchen, is the whole job.
So our test asks a plain question: across a huge spread of realistic bowls, on what fraction of days does every nutrient clear its floor? The results below are read conservatively — we report the cautious lower estimate (a Wilson confidence bound), not the flattering one — for the full size range of dogs and cats, and cats and dogs are held to the same bar.
Figures are read as a cautious Wilson lower bound (C/R 95/99) against the ≥90% NRC floor, across the full size range (dogs 3–110 kg, cats 2.5–10 kg), each configuration validated across 1 Million+ simulations.
Plot 1 · Adult maintenance, across the full size range
The share of days every nutrient clears the floor
Each bar is the fraction of simulated feeding days on which the single weakest nutrient still cleared the NRC floor — not an average, the worst nutrient in each bowl. The copper stress-test quadruples the modelled copper load and the diet still holds, because copper is the nutrient we deliberately keep on the tightest leash.
Plot 2 · Every life stage · worst-case vs recommended
Even the leanest bowl still passes
The recommended bowl is diverse — chicken and mutton with a lean farmed fish like tilapia and egg (plus curd and veg for dogs). We also test the floor: a bare chicken-only bowl, the least varied diet an owner might realistically feed. It scores lower, as you'd expect, but it still clears the NRC floor. Variety buys margin; it is never the difference between adequate and deficient.
The method · what “Monte-Carlo” means
We don't test the perfect bowl. We test the messy one.
A single calculation on paper assumes every ingredient is exactly average and every scoop is exactly right. Real life isn't like that. A Monte-Carlo simulation runs the recipe over and over — many thousands of times — jiggling each input a realistic amount on every run, then scores all of them. It turns “does this work in theory?” into “on what fraction of ordinary days does this actually hold?”
Every simulated bowl carries four independent kinds of real-world variation at once — nudged up and down with no thumb on the scale — so the pass rate reflects the full spread of things that genuinely differ from one feeding to the next.
Real food is never identical batch to batch. The natural swing in nutrient content of fresh meat, fish, liver and egg across seasons, suppliers and cuts.
The day-to-day difference in a home portion — a 290 g cut weighed on a market scale versus the same cut on a kitchen scale.
The small potency spread in the purified vitamins and minerals inside the sachet, batch to batch.
The tiny variation in how much powder each 5 g sachet actually holds off the manufacturing line.
Gut-absorption differences between individual animals are left out on purpose: they apply equally to every complete diet, and the NRC floor already carries a safety margin for them.
The two boundaries · floor and ceiling
Enough is a window, not a target
For nutrients that can harm in excess — vitamins A and D, selenium, copper — more is not better. Once a nutrient clears its floor and sits below its safe upper limit, the extra is only margin. So the sachet is built to land every nutrient inside the safe window, and copper — the one we watch most closely — is held well above the floor and roughly six to seven times below its ceiling, even under the 4× stress-test.
Schematic, not to scale — it shows where nutrients are placed, not measured distances
Under the hood · the two engines
One check for the day, one for the years
Two layers do the work. The first checks a single day's bowl against every floor. The second follows each nutrient as it builds up and clears from the body over months, so a nutrient that's fine today can't quietly accumulate to a problem later. Both have their own methods write-ups.
The Stochastic Reliability Framework — the single-day engine
Runs the day-to-day Monte-Carlo: allometric scaling for body size, realistic variance on every input, and a worst-nutrient rule scored against NRC — reported as the cautious Wilson lower bound.
Read the methodThe Hepatic Buffering Model — the over-time engine
Follows each nutrient over its biological half-life and models how the liver buffers copper and vitamin A, turning a one-day check into a chronic-safety statement. The tissue check holds at 99.808% for both dog and cat, with nothing falling below the floor.
Read the methodThe scale of the test
A pet's whole lifetime, simulated
Every configuration is validated across 1 Million+ Monte-Carlo simulations, read as a cautious Wilson lower bound at C/R 95/99— the recommended bowls clear every NRC 2006 floor with zero safe-upper-limit breaches, across the full size range. A million-plus simulated feeding days is roughly 2,700 years of one pet's meals, so a lifetime of feeding is thoroughly covered.
Plain-language glossary
Running the same recipe across many thousands of simulated real-world bowls — each one slightly different — instead of checking a single perfect bowl on paper.
A cautious statistical estimate. It reads the results conservatively and reports the pass rate we can stand behind, not the rosiest one.
The nutrient minimums set by the US National Research Council — the international reference vets and nutritionists use for cats and dogs.
The ceiling for nutrients that can harm in excess, like vitamins A and D, selenium and copper. Clearing the floor is only half the job; nothing may cross the ceiling.
Biologically Appropriate, Residually Corrected — a bowl built from diverse real food, then topped up to the floor by the sachet for exactly what grocery meat leaves short.
Every nutrient, every ratio, every test.
It's all here, live on this page — every tracked nutrient, the clinical ratio bounds (calcium-to-phosphorus, zinc-to-copper and the rest), and the full envelope charts. Nothing to download; the proof is on the screen.