30+ product attributes scored
Ingredients, dosing, evidence level, contraindications, interactions, synergies, timing, form factor — each a scoring dimension, not just a tag.
Use case — supplement & OTC health brands
A personalized supplement quiz that actually scores your catalog.
Most quizzes match answers to product tags. Blend computes a numeric, deterministic score for every supplement in your catalog — across more than 30 product attributes and more than 20 user signals — with a safety layer on every pick and the specific reasons behind every recommendation.
15-day free trial
Live for supplement brands
White-label
Pre-built scoring logic
A real personalization algorithm, not a decision tree.
Years of supplement-domain logic baked in — RDIs, ingredient interactions, dosing rules, the sub-goal taxonomy. Customize anything; default to what ships.
20+ user signals extracted
Goals, sub-goals, conditions, lifestyle, and nutrient gaps inferred from a food-frequency questionnaire (FFQ). Not just multiple-choice answers.
Personalized RDIs, not generic dosing
Reference intakes adjusted for ethnicity, sun exposure, activity level, and lifestyle — per customer, per nutrient. Dosing recommendations follow the adjusted RDI, not a one-size-fits-all chart.
Safety layer on every pick
Interaction checks, medication conflicts, upper-limit gating against tolerable intakes, synergy pairing. Every flagged limit traceable to a source rule.
Auditable, end to end
Every recommendation, traceable to the rules that produced it.
The scoring engine is fully deterministic. Same answers always produce the same picks, every pick carries the signals that produced it. When a customer or a regulator asks why, you can show the math.
That makes Blend defensible for the categories where it matters most: supplements with structure-function claims, OTC products, regulated markets.
Frequently asked questions
Algorithm questions.
How does the supplement quiz personalize recommendations?
It scores every product in your catalog against the customer using more than 30 product attributes and more than 20 user signals. Each pick comes with the specific reasons that produced it.
Is the quiz algorithm auditable?
Yes. The engine is fully deterministic — same input always produces the same output. Every recommendation carries the reasons that produced it. Important if you operate in a regulated category.
What safety checks run on each recommendation?
Supplement-supplement interaction checks, medication conflict screens, upper-limit gating against tolerable intakes, and synergy pairing. Every flagged interaction or limit is traceable to a source rule.
How is this different from a tag-based persona quiz?
Persona-based quizzes and decision trees don't actually personalize — they segment. Blend computes a numeric, deterministic score per product per customer. See the full comparison →
Ready to ship a real personalization engine?
15-day free trial on every plan.