Skip to main content
Status: Planned Tier: Free (read reviews) / Pro (write reviews)

Overview

High IQ is built on verified, structured data — terpene profiles, cannabinoid percentages, research papers, and algorithmic scoring. That data tells you what a strain is chemically. But it does not tell you what it feels like on a Tuesday evening when you are trying to unwind after work. It does not tell you that the Gorilla Glue from your local dispensary is drier than usual this batch, or that Wedding Cake is incredible for creative work but terrible for socializing. User Reviews add the subjective layer that structured data cannot capture. They are the community’s lived experience with a strain — the textures, the contexts, the surprises. But unlike Leafly or Weedmaps reviews, which are unstructured walls of text with no connection to verified data, High IQ reviews are structured, accountable, and integrated into the intelligence engine. Reviews are attached to real profiles (not anonymous), linked to purchase history (optionally verified), and factored into the strain scoring algorithm. This is not a review site. It is community intelligence that makes the whole platform smarter. This feature represents a notable philosophical evolution for the product. The original product vision explicitly listed “social features (reviews, likes, profiles)” as something we would never build, citing “noise corrupts intelligence.” We believe reviews can be implemented in a way that adds signal rather than noise — but only if they are structured, verified, moderated, and algorithmically integrated rather than simply displayed as unfiltered opinions. This requires a thoughtful implementation.

What It Does

  • Star ratings — Quick 1-5 star rating on any strain. Takes one tap. This is the minimum viable interaction.
  • Structured review form — Optional expansion with specific categories: effects accuracy (did the listed effects match?), aroma rating, flavor rating, value rating, and potency rating
  • Free-text review — Optional written review with a 500-character limit (encourages concise, useful reviews over rambling)
  • Effect tags — Tag the specific effects you experienced, which aggregates into a community effect profile for the strain
  • Verified purchase badge — If the strain exists in your stash or order history, the review is marked as “Verified Purchase,” which carries more weight in the scoring algorithm
  • Review display on strain pages — Strain detail pages show an aggregate star rating, category breakdowns, and individual reviews sorted by helpfulness
  • Helpfulness voting — Other users can upvote reviews as “helpful,” which surfaces the best reviews to the top
  • Review integration with scoring — Aggregate user ratings factor into the strain’s overall score in the ranking algorithm, alongside terpene data, research quality, and other objective metrics
  • Review moderation — All reviews pass through automated moderation (profanity filter, spam detection, quality gate) before publication
  • Personal review history — Browse all your past reviews in your profile, with the ability to update or delete any review

User Value

The “aha moment” is when you read a review from another user that perfectly describes an experience you had but could not articulate — or warns you about a specific batch issue at your regular dispensary. Reviews add the human context that data alone cannot provide.

How It Works

Writing a Review

1

Navigate to a Strain

Open any strain detail page from search, your stash, or the encyclopedia.
2

Tap the Star Rating

The star rating widget is prominently displayed on the strain page. Tap 1-5 stars for a quick rating.
3

Expand to Full Review (Optional)

After rating, a prompt offers to expand into a full review. The structured form includes: effects accuracy, aroma, flavor, value, potency, effect tags, and free text.
4

Verified Purchase Badge

If the strain is in your stash or order history, the review is automatically tagged as “Verified Purchase.” You do not need to do anything.
5

Submit

The review passes through automated moderation. If approved (typically instant), it appears on the strain page within seconds.

Reading Reviews

1

Aggregate Rating

The strain page shows the average star rating and total review count alongside the High IQ Score. Category breakdowns (effects, aroma, flavor, value, potency) are shown as bar charts.
2

Community Effects

A tag cloud shows the most frequently reported effects from reviewers, with larger tags for more common effects. This complements the database-sourced effects list.
3

Individual Reviews

Scroll to see individual reviews sorted by helpfulness (most upvoted first). Verified Purchase reviews are visually distinguished with a badge.
4

Vote Helpful

Tap the “Helpful” button on any review that adds value. This surfaces the best reviews to the top.

Technical Approach

Data Model

Reviews are stored in Convex for real-time sync:
FieldTypeNotes
userIdstringReviewer
strainIdstringStrain being reviewed
starRatingnumber (1-5)Required
effectsAccuracynumber (1-5, optional)Did listed effects match?
aromaRatingnumber (1-5, optional)Aroma quality
flavorRatingnumber (1-5, optional)Flavor quality
valueRatingnumber (1-5, optional)Price-to-quality ratio
potencyRatingnumber (1-5, optional)Potency accuracy
effectTagsstring[]Effects experienced
textstring (optional, max 500 chars)Free text
verifiedPurchasebooleanAuto-set from stash/order data
helpfulCountnumberUpvote count
moderationStatusenumpending, approved, rejected
createdAtnumberTimestamp

Moderation System

Every review passes through three moderation layers:
  1. Automated text filter — Profanity detection, spam pattern matching (URLs, repetitive text, promotional language), minimum quality gate (reject empty or single-word text reviews)
  2. Rate limiting — Maximum 5 reviews per user per day. Maximum 1 review per strain per user (can update existing review).
  3. Community moderation — Reviews with multiple “report” flags are queued for manual review. Reviews that receive zero “helpful” votes after 30 days and have negative sentiment may be automatically deprioritized.

Integration with Strain Scoring

User ratings factor into the strain’s algorithmic score with the following weights:
SignalWeightNotes
Verified Purchase reviews2x weightTrusted signal
Unverified reviews1x weightLower confidence
Minimum review threshold5 reviewsBelow this, reviews do not affect the score
Recency weightingNewer reviews weighted higherReviews older than 1 year are down-weighted
Bayesian smoothingAppliedPrevents strains with few 5-star reviews from topping rankings
This ensures that user sentiment influences scoring without allowing a small number of reviews to distort the algorithm.

Preventing Gaming

ThreatMitigation
Fake accounts leaving fake reviewsRequire email verification + minimum account age (7 days)
Review bombing (coordinated negative reviews)Rate limiting + anomaly detection on sudden rating drops
Self-promotion by brandsNo commercial accounts allowed; flagged language patterns
Review manipulation for ranking boostBayesian smoothing + minimum threshold

Tier Impact

TierAccess
FreeRead all reviews, see aggregate ratings and community effects, vote “helpful”
ProWrite reviews, structured rating categories, effect tagging, verified purchase badge

Dependencies

  • Strain detail pages — built and live
  • User authentication (Clerk) — built and live
  • Stash and order data for verified purchase detection — built and live
  • Convex real-time sync — built and live
  • Convex reviews table schema and mutations
  • Star rating widget component
  • Structured review form component
  • Review display component (aggregate + individual)
  • Helpfulness voting system
  • Automated moderation pipeline (text filter, rate limiting)
  • Strain scoring integration (Bayesian smoothed user rating factor)
  • Review reporting and manual moderation queue

Open Questions

  1. Verified purchase requirement — Should we require verified purchase (strain in stash) to write a review, or allow anyone to review? Requiring verification increases trust but dramatically reduces review volume, especially early on. Compromise: allow unverified reviews but weight them lower and display them below verified ones.
  2. One review per strain per user — This is standard, but cannabis varies by batch. Should users be able to leave multiple reviews for the same strain (one per purchase/batch)? This captures batch variation but complicates the UX.
  3. Anonymity — Should reviewers be identifiable (display name) or anonymous? Identifiable reviews are more accountable but may deter reviews about cannabis usage. Anonymous reviews encourage honesty but enable abuse. Middle ground: pseudonymous (usernames, no real names).
  4. Historical reviews — When a user adds a strain to their stash and writes a review, should we prompt them to review other strains in their collection? This bootstraps review volume but may lead to low-effort reviews.
  5. Product vision alignment — The original vision says “social features (reviews, likes, profiles): noise corrupts intelligence.” This feature explicitly adds reviews. The key distinction is that these are structured, verified, moderated reviews integrated into the intelligence engine — not a social feed. This distinction should be documented and communicated clearly.