Dataset: AI Usage
Entity: AI Usage Measurement
Field ID: suggestions_offered
Type: Number
Description: The number of suggestions the AI assistant offered. Not every provider reports it: acceptance rates therefore have different denominators across providers.
Source: App
Transformation logic: N/A
App Mapping
GitHub Copilot |
|
Anthropic Claude (coming soon) | N/A - Anthropic never reports how many proposals were shown |
Cursor (coming soon) |
|
OpenAI Platform (coming soon) | N/A |
OpenAI Codex (coming soon) | N/A |
Reporting Use Cases
The Suggestions offered field is the denominator of any acceptance rate. Because only two providers publish it, acceptance rates are not directly comparable across all tools.
Acceptance rate:
SUM(suggestions_accepted) / SUM(suggestions_offered)over a period is the standard measure of how useful a tool's suggestions are.Compute rates over a window, not a day: A suggestion offered one day and accepted the next makes a single day's ratio exceed 100%. A weekly or monthly window removes the artefact.
Anthropic has a different denominator: For Anthropic records, compute acceptance as accepted over accepted plus rejected. Never fill this field's absence with that sum.
Two different things in Cursor: Inline completions and applied code blocks are separate populations on separate records. Do not add them together.
