Why I Did the Testing Myself
The published literature on alcohol-recovery supplements is good for understanding mechanisms but bad for predicting how a specific combination feels at a real serving size in a real person. The clinical trial gap is wide -- there are very few human trials of multi-ingredient formulas at the doses used in actual consumer products. To get from "this ingredient has a published mechanism" to "this formula at this dose works for me" required actually testing it. I did that on my own body because I was the only test subject I could fully instrument and fully control 1.
The Protocol
Each self-test followed the same structure. I drank at a fixed level set from my own baseline: enough that I would normally expect a hangover. I logged the drinks, beverage type, food, sleep window and timing. In the early tests I varied when I took the test capsules, including after the last drink. Taking them before the first drink worked best, and that became the standard early on.
At first I scored the next morning on a 12-symptom checklist, 0 to 10, modeled on the published Alcohol Hangover Severity Scale 4. I realized early that the data would not be study-grade, the way it was being recorded, so I switched to a blended score built for formulating decisions: did this candidate do the job or not? The Alcohol Hangover Research Group's best-practice statement informed the protocol design 3. (this is the part where I have to be honest that the protocol got tightened up along the way, after the first 30 tests showed me how unstructured I had been initially.)
Drinks were counted as standard drinks of 14 grams of pure alcohol: about 12 ounces of beer at 5 percent, 5 ounces of wine at 12 percent, or 1.5 ounces of spirits at 40 percent5. Counting the alcohol rather than the glasses is what the Alcohol Hangover Research Group recommends, because serving sizes and strengths vary so much3. Every test night's log had the drink type, its strength, the exact volume, the times of the first and last drink, total grams of alcohol, what I ate and when, water before drinking, the sleep window, anything that departed from the plan, and the dose and timing of whatever I was testing. Hangover research also uses a single overall rating beside the symptom checklists, and a 2020 analysis found that one-item rating reflects overall hangover severity better than the symptom average6. Drink Counting Methodology explains why each of those variables mattered.
Acts by promoting aldehyde and alcohol metabolism of foods.†
† These statements have not been evaluated by the Food and Drug Administration. This product is not intended to diagnose, treat, cure, or prevent any disease.
After the first 30 tests, the protocol stabilized into something defensible, in the n=1 self-experiment context at least.
What I Was Measuring Against
About 20 nights across the testing period were no-supplement nights: the same drinking, no test capsule. I always knew which nights they were, so they were not a placebo. They were a reminder of what no protection actually feels like at that drinking level, and the baseline every test was read against.
Helps you feel fresh.†
† These statements have not been evaluated by the Food and Drug Administration. This product is not intended to diagnose, treat, cure, or prevent any disease.
The design was sensitive to direction, not size. With one person you cannot estimate an effect size, but you can tell a candidate that clearly changes the morning after from one that makes no difference against the no-supplement nights 2.
Trade-offs I Knew Going In
Self-blinding is hard. I tried opaque capsules with rotating colors, but that is not blinding, and the testing was not blinded. Order effects (improvement over time as I got better at the protocol) were a real risk, and I tried to control them by re-running no-supplement nights and earlier candidates at intervals. Day-of-week effects, seasonal effects and cumulative alcohol exposure were all confounders I could observe but not eliminate. I called these out in my logs and discounted findings that lined up with them.
Supports balanced consumption of alcohol (from all sources of food and drink).†
† These statements have not been evaluated by the Food and Drug Administration. This product is not intended to diagnose, treat, cure, or prevent any disease.
What This Page Is Not Claiming
This was a personal R&D process, not a clinical trial. The 150 tests were enough to converge on a formula, not to prove anything about other people; that is what the over 400 group tests that followed were for, and they are not proof either. The data were not study-grade and are not published.
For the elimination logic that drove the actual formula decisions, see The Elimination Method. For the drink-counting specifics, see Drink Counting Methodology.