Methodology
How Matchleaf Works
Matchleaf is a decision tool, not a catalog. You bring the strains you have — whatever is available to you — and tell it what you want to feel. It ranks them by blending a deterministic terpene engine with data on how each strain is actually reported to feel.
The Matching Process
Enter your available strains
Type or paste the strain names from whatever you have — a menu, a home stash, a friend's options. Matchleaf also accepts photos: take a picture of a menu and the tool extracts strain names automatically. You can also scan a single product's own label or COA to score it on its real, measured terpene percentages.
Select your target effects
Choose from 10 effect categories. Select multiple to get separate, independent recommendations per effect — not a single compromise strain. Each effect lane returns its own ranked list.
Get ranked recommendations with terpene reasoning
Matchleaf scores each of your strains against your chosen effects by blending a deterministic terpene model with reported-experience data, ranks them by fit, and shows the terpene mechanism behind each pick. Strains in the database are matched instantly; strains we don't have yet are analyzed in real time.
How Scoring Works
Every ranking combines two independent signals. The first is a deterministic terpene engine: given a strain's terpene profile, it computes a mechanistic score for each effect from a fixed weighting model. The same profile always produces the same score — there is no language model in this loop, and no randomness. The second is a reported-experience score: an aggregate of how the strain is actually described, which reflects THC, the full cannabinoid entourage, and expectation — things a terpene model alone cannot see.
Matchleaf ranks by a weighted blend of the two. In one sentence: we rank by a blend of what people report a strain does and what its terpenes mechanistically support — trusting the terpene side more when we have real lab percentages for it, and less when we only know the terpene names. Neither signal overrides the other; where they disagree, the score lands between them rather than letting one side win. The blend is deliberately conservative today — the reported-experience signal leads while we validate the terpene engine against real user feedback.
How much the terpene engine counts depends on how good our terpene data is for that strain — its data grade:
measuredReal percentages from a lab COA — including a label you scan yourself for the exact product in front of you. Highest trust; the engine gets its strongest say.
estimatedPopulation-average lab percentages for that strain. Real numbers, but not this specific batch.
inferredWe know the terpene names and their rough order, but not measured amounts. The engine falls back to a rank-order estimate, so it counts least. Most strains are here today, and the grade improves as real lab data is added.
When none of a strain's terpenes relate to the effect you picked, the engine has no opinion to give — so it abstains rather than inventing one, and that strain is judged on reported experience alone, or dropped from that lane if nothing supports it. A blank is never scored as a lukewarm match.
For strains not yet in the database, Matchleaf falls back to real-time AI analysis. The output is the same shape — ranked picks with terpene reasoning — generated on the fly rather than retrieved from stored profiles.
The Terpene Model
Inside the engine, each effect is modeled by a set of terpenes tagged by how they relate to it: primary terpenes are the main mechanistic drivers, supporting terpenes reinforce the effect, and some terpenes suppress it — mechanistically working against it and pulling the score down. A strain's score for an effect is the sum of these signed contributions, weighted by how much of each terpene it contains.
Terpenes are weighted above cannabinoid content in the engine: terpene profiles are more consistent across batches than THC percentage, and terpene pharmacology gives a mechanistic basis for effect prediction that the indica/sativa binary does not. THC still applies a modifier — for example, high THC with no calming terpene present nudges anxiety-sensitive effects down. When we have real percentages, absolute concentration matters too: a more terpene-rich profile expresses its associations more strongly, and myrcene's sedation contribution is dampened below roughly 0.5%.
Because the score is just this arithmetic, every pick can show its work: the exact terpenes that pushed it up or down, and by how much. That itemization is the score — not a justification written after the fact.
See the full weighting table — the Terpene Effect Matrix — for exactly which terpenes primarily drive, support, or work against each of the 10 effects. Or open the interactive Engine Explorer to adjust a terpene profile yourself and watch the scores and contributions move in real time.
The 10 Effect Categories
Each category maps to a distinct use case and a set of terpene associations. These are informational descriptions — not medical claims.
Energizing — Uplifting and active. Associated with terpinolene, ocimene, and pinene.
Creative — Divergent thinking and ideation. Often associated with pinene, limonene, and terpinolene.
Euphoric — Mood elevation. Associated with limonene, with linalool and terpinolene in support.
Sexual Amplification — Sensory enhancement. Associated with limonene and linalool. A thin-evidence lane — modeled by few terpenes.
Relaxing — Body and mind ease without heavy sedation. Myrcene, linalool, and caryophyllene.
Sleep — Sedation and sleep-onset associations. Myrcene and linalool; myrcene's contribution is dampened below ~0.5% when we have real percentages.
Pain Relief — Physical-comfort associations. Caryophyllene (a CB2 receptor agonist), myrcene, and humulene.
Social — Conversational ease. Associated with limonene and linalool; low-myrcene profiles.
Focus — Concentration and clarity. Associated with pinene and eucalyptol; myrcene works against it.
Appetite — Hunger associations. Myrcene, with humulene modeled as working against it. Also a thin-evidence lane.
Common Questions
Does Matchleaf use AI to score strains?
The ranking core is a deterministic terpene engine — a fixed weighting model where the same terpene profile always produces the same effect score, with no language model in the loop. AI is used only to read menu and label photos, to phrase the plain-English reasoning under each pick, and as a real-time fallback for strains not yet in the database. The number a strain is ranked by comes from the deterministic engine blended with reported-experience data, not from an AI's opinion.
What makes a strain rank higher for an effect?
A weighted blend of two independent signals: how the strain's terpenes mechanistically relate to that effect (each terpene tagged as primarily driving it, supporting it, or working against it), and how the strain is actually reported to feel. Real lab terpene percentages make the terpene side of the blend count more; when we only know the terpene names, it counts less.
What do measured, estimated, and inferred mean?
They are the data grade of a strain's terpene information. Measured means real percentages from a lab COA — including a label you scan yourself for the exact product in front of you. Estimated means population-average lab percentages for that strain — real numbers, but not that specific batch. Inferred means we know the terpene names and their rough order but not measured amounts. The better the grade, the more the terpene engine is trusted in the blend.
What happens when a strain's terpenes don't relate to the effect I picked?
The engine abstains rather than inventing an opinion. That strain is then judged on its reported experience alone, or dropped from that effect's list if nothing supports it. A strain with no relevant terpenes is never scored as a lukewarm match — a blank is treated as a blank, not a middling result.
Are Matchleaf's effect scores medical claims?
No. Every terpene-effect relationship in the model is a preclinical, mechanistic association drawn from terpene pharmacology research — not a medical claim and not a guarantee of individual effect. Matchleaf is informational only.
What Matchleaf Is Not
- —Not a medical device or health advisor. All content is informational only.
- —Not a store. Matchleaf doesn't sell cannabis or connect to live inventory.
- —Not a catalog. It works with whatever strains you have — not a browsable database of everything available.
- —Not a guarantee. Terpene associations inform probable effects; individual response varies.
Ready to try it? Enter your strains and select an effect.
Open Matchleaf →For informational purposes only. Matchleaf content does not constitute medical advice. Terpene associations described reflect current research literature and Matchleaf's modeling choices, and may not apply to all individuals. Check applicable laws regarding cannabis use in your jurisdiction.