Skip to content

Methodology changelog

Every change to how we score, tier, value and award pre-workouts, newest first: what changed, why, and what moved. The current methodology is v2.10.1.

Major versions change what a tier means; minor versions move some grades, value badges or awards; patch versions change no score.

  1. v2.10.1Current

    The one-line verdict never names a different tier

    • Decision: the one-line verdict on a product page only blames the caffeine safety cap ("Hard cap to C") or the hidden-blend transparency cap ("Tier capped at D") when that cap is what set the tier shown, and it names that tier. When the label math already put a product below the cap, or a stricter cap set the tier, the verdict for the tier it actually earned appears instead. Search results now show a factual summary of each product (brand, name, tier, score, caffeine, cost per clinically dosed serving and testing status) rather than the verdict line.
    • Why: a product over the caffeine safety limit could show F tier while its verdict said "Hard cap to C", which read as if it had scored a C. Search engines also quoted that line as the product's description.
    • Impact: wording only. No tier, score, value or award changes.
  2. v2.10.0

    Bare Rauwolfia extract is not rauwolscine

    • Decision: treat Rauwolfia vomitoria root bark extract as its own questionable stimulant, separate from rauwolscine. Its printed amount is the mass of the botanical extract and does not imply any rauwolscine dose. The extract raises stim level and caps the formula at B, but it earns no dose credit and has no claimed effective dose or safety ceiling. Only labels that explicitly name rauwolscine/alpha-yohimbine or state a rauwolscine percentage map to rauwolscine.
    • Why: a bare botanical extract can contain rauwolscine, yohimbine, reserpine, ajmaline, serpentine and other alkaloids in an unspecified mixture. Operation Supplement Safety says an extract-only label does not reveal which compounds are present, so converting its mass to rauwolscine would manufacture precision the label does not provide.
    • Impact: formulas listing the bare extract now show that exact ingredient instead of Rauwolscine and never report an inferred rauwolscine amount. They retain the conservative questionable-stimulant B cap and stim-level increase. Explicit 90% rauwolscine extracts continue to convert at 0.90 and use the Rauwolscine ingredient.
  3. v2.9.0

    Artificial dyes follow each flavor

    • Decision: each flavor can now record its own artificial-dye count. A blank value inherits the formula's reference label and an explicit zero means that flavor is dye-free. Clean and the no-artificial-dyes filter use the highest effective count across every current flavor.
    • Why: flavors can share the same active formula while using different colorants. Treating one flavor's label as universal could call a dyed product dye-free.
    • Impact: C4 Ultimate's Icy Blue Razz remains dye-free, while Hawaiian Punch Fruit Juicy Red records its FD&C Red 40. C4 Ultimate therefore no longer qualifies for β€œNo artificial dyes,” and its Clean score reflects one dye. Dose, stimulant, nutrition, price and trust rules do not change.
  4. v2.8.0

    T4 is a testing label, not an automatic award

    • Decision: remove the 🧒 Cap Award. Products whose only testing evidence is a manufacturer claim still show T4 Β· Brand Claim Only, keep a trust score of 0 and remain available in the testing filter, but T4 alone no longer creates a Hall of Shame award.
    • Why: the award had no ranking criterion. Every T4-only product received it automatically, duplicating the testing badge instead of identifying exceptional behavior.
    • Impact: the five current tied Cap Award cards disappear after recompute: four BodyTech or BodyTech Elite formulas and Swolverine PRE. No formula score, tier, trust score, value badge or other award changes.
  5. v2.7.2

    Hidden-blend deductions name the blend

    • Decision: the "Where the points went" line for hidden amounts now names each blend, the total it prints and the ingredients whose share it leaves out, e.g. "Undisclosed amounts: Organic Mushroom Powder Blend prints 100 mg but not how much is Cordyceps or Lion's mane mushroom. 1.1% of active mass (0.3 per percent)". It used to say only "1.1% of active mass hidden in blends".
    • Why: the old line didn't say which part of the label was hidden, and next to a printed "100mg" it was easy to read as if nothing were.
    • Impact: wording only. No tier, score, value or award changes.
  6. v2.7.1

    The caffeine warning and stim level say why

    • Decision: the caffeine warning now names the ingredient that needs the extra servings and says the total is over our limit, e.g. "Clinical taurine takes 2 servings = 608 mg caffeine, over the 400 mg limit", instead of "Clinical dose = 608 mg caffeine". Product pages also show what sets the stim level next to it, e.g. "304 mg caffeine + Rauwolscine".
    • Why: the old warning read as if one serving contained that much caffeine, and the stim level looked like it came from caffeine alone even when another stimulant had raised it a step.
    • Impact: wording only. No tier, score, stim level, value or award changes.
  7. v2.7.0

    Rauwolscine counts as a questionable stimulant

    • Decision: Rauwolscine (also called alpha-yohimbine or Ξ±-yohimbine) is tracked as its own stimulant, not as yohimbine. Any disclosed amount raises the stim level by one step and caps the formula at B, the same treatment used for yohimbine, synephrine and hordenine. No effective dose or safety ceiling is claimed. Separately, zΓΌmXR Delayed Release is now a distinct caffeine form; when a label prints caffeine mass, all of that mass counts toward the same caffeine total.
    • Why: Operation Supplement Safety identifies rauwolscine and alpha-yohimbine as names for the same stimulant, distinct from but very similar to yohimbine, and reports no human clinical safety or effectiveness studies. Treating it as harmless or folding it into yohimbine would both be misleading. Delayed Release and Extended Release zΓΌmXR describe different release profiles, but both still deliver caffeine.
    • Impact: no current score moves because no stored formula line currently names rauwolscine or alpha-yohimbine. Once such a label is mapped, the formula gains one non-caffeine stimulant and is capped at B. The delayed-release alias is a data correction, not a scoring-rule change: it makes the printed caffeine count correctly without changing how caffeine is graded.
  8. v2.6.3

    Unmapped label amounts follow the graded serving

    • Decision: Label Math now scales a disclosed but unmapped line to the serving being graded. The Label column still quotes the printed base-serving amount; Actual shows that amount at the graded serving. The line remains unscored and is not converted into a different ingredient.
    • Why: Solid Pre prints 500 mg of Pink Himalayan Sea Salt per scoop and is graded at two scoops. Its full nutrition label correctly showed 1,000 mg, but Label Math showed 500 mg and a blank Actual because the salt is deliberately separate from the label's sodium amount.
    • Impact: no score, tier, value badge, or award moves. Solid Pre's salt row now shows an Actual of 1,000 mg at two scoops without adding that salt mass to its 380 mg of sodium. Other disclosed, unmapped lines now follow the same serving display rule.
  9. v2.6.2

    Ingredient type follows the label's claim; only real stimulants count as stims

    • Decision: an ingredient's type (⚑ Energy, 🩸 Pump, 🧠 Focus and so on) now describes what the manufacturer says it's for, so a product's label reads the way it's sold. An ingredient we don't score never counts toward the stim level or the stim count in Clean, whatever its type. Caffeine and the stimulants we do track (theacrine, methylliberine, synephrine and the rest) count as before.
    • Why: a few label ingredients are sold for energy without any evidence that they act as stimulants at the amounts printed. Theobromine, for example, didn't raise alertness at 100 to 400 mg in a controlled trial. Calling them "Energy" is fair to the label, but letting that push a product into a higher stim level would penalize it for a stimulant effect nobody has shown.
    • Impact: none today: every one of the 52 formulas keeps the same tier, scores, caffeine total and stim level (checked before and after). It matters for ingredients marked Energy from now on, such as theobromine, green tea extract and naringin.
  10. v2.6.1

    Salty measures sodium only

    • Decision: Salty scores sodium alone, against the same 300 mg reference. Potassium, chloride, magnesium and calcium stay on the label but are never averaged into Salty. Ingredient pages now say plainly whether each ingredient counts toward the grade, a sub-score, or nothing.
    • Why: we looked hard for evidence that potassium before a workout does anything and found none: no human trial shows it helps performance, hydration or cramping at any dose. It aided fluid retention only when people drank about 1.4 g of it in a litre-plus of fluid *after* exercise, and even then it added nothing on top of sodium (Maughan et al., 1994). Sweat carries far less potassium than sodium, and Ironman athletes who cramped had the same blood electrolytes as those who didn't (Schwellnus et al., 2011). Scoring it would also backfire: Salty averages the electrolytes on the label, so a product that adds 125 mg of potassium would drop from 100 to about 71, while one that leaves it out would keep 100. Chloride, magnesium and calcium have no pre-workout evidence at label doses either.
    • Sodium's source, corrected: the 300 mg reference reflects the American College of Sports Medicine's 0.5–0.7 g of sodium per litre for exercise over an hour, which comes from its 1996 position stand; we had credited the 2007 one. It's roughly a sports drink's sodium in a scoop's worth of water, not a proven pre-workout dose: sodium loading improves endurance in the heat at 1.3–4.4 g with a litre of fluid, and nobody has tested 150–700 mg before a lifting session.
    • Impact: no score moves. Salty was already sodium-only in practice, since sodium is the only electrolyte we score; this makes it the rule and explains it.
  11. v2.6.0

    Beta-alanine is a daily supplement, not a pre-workout requirement

    • Decision: beta-alanine leaves the ideal-pre panel and is treated like creatine. Leaving it out, or under-dosing it, no longer costs Dose points or blocks S, and it never sets the number of scoops behind real servings and cost per serving. The remaining panel weights are unchanged and the score is shared across them. Beta-alanine still shows on the label math, still counts toward transparency (hiding it in a blend costs points like any active), still feeds the endurance sub-score, and still decides Hybrid and the tingle-free filter.
    • Why: beta-alanine works by raising muscle carnosine over weeks of daily use (about 4–6 g a day for 2–4 weeks), not from the dose taken before one workout (ISSN position stand, 2015; Australian Institute of Sport). That's the same reason creatine was never in the panel. The tingle it causes is paresthesia, a harmless skin sensation, not a sign the formula is working. Betaine's evidence is also mostly from daily use; it stays in the panel for now and is under review.
    • Impact (33 real products, offline preview): S4 A1 B5 C12 D7 F4 β†’ S4 A4 B4 C16 D2 F3. - Up: formulas that skipped or under-dosed beta-alanine. Gorilla Mode B 70.8 β†’ A 85.0, Gorilla Mode Nitric C 64.7 β†’ A 84.6, Pre JYM C 66.7 β†’ A 80.0, Alani Nu D 54.2 β†’ C 65.0, Bucked Up D 50.0 β†’ C 60.0. The key-dose cap no longer fires anywhere (it hit 7 products, all for beta-alanine). - Down: formulas whose score leaned on beta-alanine. A bare citrulline + caffeine + beta-alanine label goes C 66.7 β†’ C 60.0; Legion Pulse Stim-Free B 76.5 β†’ C 69.2; Solid Pre and GHOST Legend S 95.8 β†’ S 95.0. Thavage and BULK keep their perfect 100. - Servings: the scoop count now follows the remaining panel ingredients. Pre JYM and JYM High Performance lose their beta-alanine caffeine-limit warning. Bucked Up (taurine), C4 Original (citrulline) and Pre JYM (betaine) now get one, because reaching those doses would mean more than 400 mg caffeine. Moves into C or out of it change value bands, which only apply from C up.
  12. v2.5.0

    The GOAT ranks by formula first

    • Decision: The GOAT goes to the caffeinated formula with the highest tier score; overall score (testing + value) only breaks ties.
    • Why: ranked by overall, price beat formula: after v2.4.0, Solid Pre (95.8, Broke Bro Approved) out-scored BULK (a perfect 100, Fair Price). A headline award must not go to a lesser formula because it's cheaper; value has its own awards.
    • Impact: GOAT Solid Pre β†’ BULK (100; ahead of Thavage, also 100, on overall). The award stat shows the tier score.
  13. v2.4.0

    Value bands move up

    • Decision: Broke Bro Approved ≀ $1.00 β†’ ≀ $1.50; Fair Price ≀ $1.75 β†’ ≀ $2.00; Pricey (≀ $2.50) and Luxury unchanged. Band value scores unchanged (100/75/50/25).
    • Why: with D/F excluded (v2.3.0), the strict $1.00 line left no B-or-better formula able to earn the badge, and the Broke Bro Special winner (Godzilla, $1.42) was only Fair Price. $1.50 is where a typical $30–45 tub lands when one or two scoops is clinical.
    • Impact (C+ catalog): 9 Broke Bro Approved / 4 Fair / 5 Pricey / 4 Luxury. Overall scores of products that changed band shift accordingly.
  14. v2.3.0

    No value band without a clinical dose

    • Decision: value bands apply only from tier C up. D/F formulas show Underdosed, with the neutral value score (50). Cost per effective serving is still stored (Overpriced Pixie Dust uses it).
    • Why: a formula with nothing clinically dosed has a one-scoop "effective serving", so junk looked cheap: Nitraflex and GHOST Pump (both F) were Broke Bro Approved. We never call anything that isn't clinically dosed "approved".
    • Impact: 10 D/F products moved to Underdosed; no Broke Bro Approved below C.
  15. v2.2.0

    Serving Size Scam at the graded serving

    • Decision: compare the label's container count *at the graded serving* (count Γ· graded multiplier) with clinical servings; require a real gap (> 0 at 0.1 precision).
    • Why: labels declaring a 2-scoop serving ("40 one-scoop / 20 two-scoop servings") were graded at 2 scoops but compared against the 1-scoop count, flagging honest labels.
    • Impact: Solid Pre, Gorilla Mode Nitric and Thavage ("Says 40, math says 20") cleared; GNC Pro Performance and Alani Nu ("Says 30, math says 15") now hold it.
  16. v2.1.3

    Award cards show the deciding stat (no scoring change)

    • Award evidence also stores tier score, overall, pump, focus, endurance and flavor count; each award shows the stat that decided it instead of "Rank #1".
  17. v2.1.2

    πŸ’― Perfect formula badge (no scoring change)

    • Decision: keep S at β‰₯ 90; mark a tier score of 100 with a separate πŸ’― Perfect formula badge (plus a "Perfect formulas only" filter).
    • Why: considered limiting S to 100. Kept S as is (missing only theanine, or a sub-200 mg caffeine choice, shouldn't drop a formula to A) but surfaced "nothing lost" explicitly.
  18. v2.1.1

    Itemized point deductions (no scoring change)

    • The engine records every deduction as a rule with its points (panel shortfalls, hidden mass, calories/carbs); product pages show "Where the points went". Panel dose and the breakdown share one calculation, so the items always add up to 100 βˆ’ D.
  19. v2.1.0

    Disclosure as the baseline; pixie dust earns nothing

    • Decision: FS = D βˆ’ 0.3 Γ— hidden % (replacing 0.70 Γ— D + 0.30 Γ— T); a panel ingredient below 80% of its reference dose earns 0 credit (caffeine exempt).
    • Why: transparency was a bonus on top of dose; full disclosure should be expected and hiding should cost. Token amounts shouldn't earn partial credit.
    • Impact: S4 A1 B2 C13 D7 F3; bare trio = C 66.7; Pre JYM C 66.7.
    • Also: πŸ™… Tingle-free tag (no beta-alanine anywhere and transparency β‰₯ 50; never vouch for a hidden blend).
  20. v2.0.0

    Dose = the ideal-pre panel

    • Decision: Dose is scored against a fixed panel (citrulline 3, caffeine 3, beta-alanine 2, betaine 1.5, taurine 1, tyrosine 1, theanine 0.5); missing or hidden = 0; non-panel ingredients never move the tier. Stim-free drops caffeine + theanine.
    • Also: creatine out of scoring and real servings (daily-loading); tyrosine reference 1 g; questionable stimulants cap at B; new testing tier T3P Β· Potency Tested (score 35) for potency-only reports (Gorilla Mind); "Just a caffeine pill" never applies to stim-free or unrecognised labels; all-blend labels score F (🧚 Fairy Dust, e.g. GAT Nitraflex).
    • Why: an evidence-weighted average of whatever a label contains rewarded short labels; a fixed panel (sources: Garage Gym Reviews, ATHLEAN-X, RP, a DIY-stack video) means a bare trio can't tie a complete formula.
    • Impact: S4 A4 B19 C1 D1 F1 at the time (bare trio = B 76.7), before v2.1.0.
  21. v1.1.0

    Tiers, caffeine and key doses

    • Public wording: "Grade" β†’ Tier (S–F); "Stim tier" β†’ Stim level.
    • Caffeine is dose-scored: r = min(caffeine / 200 mg, 1); 0 mg listed = stim-free, unscored; never drives real servings. Other stimulants stay unscored (stim level only).
    • Stim-free ranked separately (homepage and tier list); GOAT = caffeinated #1, Night Shift Champ = stim-free #1.
    • Key-dose cap: any strong-evidence non-stim active under its clinical dose β†’ max A (89.9). Creatine and caffeine excluded; beta-alanine stays key.
    • Clean dropped from the tier: FS = 0.70 Γ— D + 0.30 Γ— T; Clean is still computed for the dye filter. Dyes and sweeteners are filters, not points.
    • Real servings are driven by strong-evidence actives (moderate as fallback).
  22. v1.0.0

    Launch baseline

    • Grades from label math only: evidence-weighted dose against cited clinical doses, transparency, and safety caps.
    • Citrulline malate 2:1 = mass ratio, 0.6667 yield.
    • Multi-serving labels graded at the declared serving that is effectively dosed.
    • Sodium is the only electrolyte scored: a separate Salty score (sodium against 300 mg), not in Dose. Potassium, chloride, magnesium and calcium are shown on the label but never scored.
    • Real servings capped so caffeine never exceeds 400 mg; never imply an unsafe scoop count.
    • Prices polled weekly with full history; product pages show the lowest price found; Amazon prices only via the API, never manual.