This is the exact same technology Amazon uses to power "Customers also bought" — Amazon Personalize, running live in your browser. Rate movies, music, books, or games and watch it build a model of your taste from scratch.
👇 Start rating to wake the algorithm
At 0 ratings: Amazon shows everyone the same popular items. At 3+ ratings: It starts detecting your patterns. At 12+ ratings: A personal ML model activates — just for you.
7
Domains
280
Items
0
Recs made
0
Ratings given
Algorithm Awakening
0 / 20 ratings
1
Cold Start — Popular Items
No data yet. Amazon shows the most popular items to everyone — the same list for 300M users.
0+
2
Pattern Detection
Your genre preferences are being extracted and weighted from your ratings.
3+
3
Collaborative Filtering
Finding users with similar taste. "People who liked what you liked also loved..."
7+
4
🔥 Personalize ML Active
Your personal model is now running. Recommendations are unique to you.
12+
5
⚡ High Confidence
Enough signal to make predictions Amazon would stake ad spend on.
20+
Your Taste Archetype
—
Rate more items to unlock your archetype
Same engine · 7 domains
👆
Rate items to wake the algorithm
Click ❤ to like · 👎 to pass · The algorithm learns immediately after each rating
Step 1: Rate 3+ items to start learning
Step 2: Watch recommendations appear
Step 3: See WHY it picked each one
Movies
❤ Like · 👎 Pass — algorithm updates instantly
👤
Generic Visitor
Same for everyone — no personalization
⚡
You — Personalized
Amazon Personalize ML model
—
Items generic list got wrong
—
Items Personalize got right
Waiting for your first rating
What Amazon does: Before you rate anything, everyone sees the same popular items. The moment you rate your first item, Amazon begins building a model unique to you.
ML Behavior Signals
building model...
Session interaction rate—
Like/pass ratio—
Genre affinity score—
Cross-domain pattern—
Predicted LTV segment—
Behavioral Predictions
Purchase probability (next 7 days)—
Churn risk—
Avg order value estimate—
Best contact time—
Recommended ad channel—
Revenue Impact Journey
1
Cold Start
Same items for everyone. No revenue lift.
+$0
2
Pattern Detection
CTR improves as categories emerge.
+$0
3
Collaborative Filtering
Conversion lift from similar user data.
+$0
4
Personal ML Active
Your model runs. Revenue compounds.
+$0
5
High Confidence
Amazon bets ad spend on these predictions.
+$0
Neural Network
TRAINING
Loss
—
Accuracy
—
Epochs
0
Samples
0
12-D Behavioral Vector
LIVE
Hover Time
—
Like Ratio
—
Scroll Speed
—
Hesitation
—
Revisit Rate
—
Price Affinity
—
Genre Breadth
—
Cross-Domain
—
Negative Signal
—
Session Depth
—
Rating Speed
—
Return Prob
—
Purchase Intent
200ms decay
0%
intent score
Churn Probability
LIVE
Risk: —Alert at 70%
Live Cohort Assignment
Rate 3+ items to assign cohort
Price Sensitivity Curve
Sweet spot: —
Next Purchase Sequence
12-Month LTV Projection
—
projected lifetime value
Server ML Validation sklearn endpoint
Browser TF.js
—
Server sklearn
—
Agreement—
Confidence—
Est. LTV—
Churn risk—
Gradient boosting model trained on 10K synthetic behavioral sessions. SageMaker-ready architecture.
Live Context Signals sent with every request
Time of Day
—
Session Length
0m
Active Domain
Movies
Interaction Depth
Cold
Top Genre
—
Confidence
0%
💰 Business ROI Calculator
Based on 10,000 monthly visitors
$0
additional monthly revenue from personalization
Generic conversion rate2.0%
Personalized rate—
Click-through lift—
Extra conversions/mo—
Based on Amazon published 35% revenue attribution and McKinsey conversion lift data.
Customer Segment
—
Return probability
—chance of returning
Your Taste Map
LIVE
🤖
Algorithm is waiting
Rate at least 3 items and watch personalized recommendations appear here — with full explanations of why Amazon picked each one.