A decade ago, web performance was a vibe. Some sites felt fast, some felt slow, and nobody could agree on why. Then Google shipped Lighthouse and Web Vitals, turned fast into numbers, and the entire industry started optimizing for a score it could finally see. The agent era needs the same thing — for readability instead of speed.
What Web Vitals did
Web Vitals worked because it made an invisible quality measurable. Largest Contentful Paint, Cumulative Layout Shift — these gave teams a target and a shared language. You can't optimize what you can't measure, and once you could measure perceived speed, you got a decade of improvement.
The agent blind spot
Today, almost no one can tell you how readable their site is to an AI agent. Does a fetch return clean content or a wall of markup? Is the important information in the HTML or injected by JavaScript the agent never runs? Is the page one clear document or ten nested layouts? These questions decide whether an agent represents your site correctly, and right now they're completely invisible.
What an AIO Score measures
An AI Optimization score — AIO Score — would put a number on agent readability:
- How much of the page is signal versus noise.
- Whether core content survives a no-JavaScript fetch.
- How cleanly the structure maps to fields an agent can use.
- Whether the page is even reachable without tripping a consent gate.
Web Vitals measured the page as a human perceives it. AIO Score measures the page as an agent reads it.
Why a standard, not a tool
Any one company can ship a checker. That's not the point. Web Vitals mattered because it became a shared standard that publishers optimized toward and platforms rewarded. Agent readability needs the same: a metric stable enough that improving your score actually improves how every agent sees you.
We're going to spend the next few years arguing about how agents should read the web. We'll get there faster if we can agree on how to measure it first.