liveBook · Manning

Gross merchandise value

As explained in Machine Learning System Design

The total monetary value of merchandise transactions over a specified period before deductions such as returns or expenses.

Gross merchandise value (GMV) measures the total value of sales over a specified period. For an eCommerce recommender, it serves as the final online business outcome, but becomes measurable only after deployment and production A/B testing. section 5.2.2

GMV fits into a practical metric hierarchy:

  • Teams may optimize click-through rate (CTR) because clicks precede purchases and provide more positive training examples. section 5.2.2
  • Conversion rate (CR) connects clicked offers to purchases, while GMV captures the broader sales outcome. section 5.2.2
  • Proxy metrics are useful only when changes in them reliably predict changes in the online metric they represent. section 5.2.2

Recommender-system evaluation

Suppose 10,000 offers produce a 3% click rate and a 3% purchase rate among clicks. That yields only nine purchases, making GMV experiments slow and purchase labels highly imbalanced; CTR supplies roughly 33.3 times more positive examples. section 5.2.2

Table 5.1 Examples of offline and online metrics
Offline metrics
Online metrics
Recall at given specificity for spam message classification
Number of user complaints about spam messages
Quantiles of 1.5, 25, 50, 75, 95, and 99
Value of expired items, total sales
Mean reciprocal rank, normalized discounted cumulative gain
Click-through rate on search engine result page

Compare offline metrics with the online outcomes they are intended to predict, including total sales. section 5.2.2

How Machine Learning System Design covers it

How deep it goes
Defines GMV and explains its metric-hierarchy role, stopping short of detailed calculation or accounting treatment.
Examples
Uses eCommerce offers, 10,000 impressions, nine purchases, CTR, and conversion rate to illustrate metric tradeoffs.
What you can do after
Evaluate recommender-system metrics by relating offline proxies to GMV and recognizing purchase-data imbalance.
How it connects
Connects GMV to recommender systems, click-through rate, conversion rate, proxy metrics, and A/B testing.
Cover of Machine Learning System Design
liveBookmanning.com
Read Machine Learning System Design on liveBook Get the book at manning.com

This explainer was generated by Manning's knowledge graph from the text of Machine Learning System Design. The section links point to the parts of the book it draws on.