Measured, not surveyed

How ready is the web for AI answer engines? We measure it.

These benchmarks come out of our own measurement corpus — every site AI Syndicate scores leaves a dated observation behind, whether it belongs to a customer, a competitor we were asked to benchmark against, or someone who ran our free check. Aggregated and anonymized, that record answers a question nobody else can answer with real data: where does the average site actually stand, and what actually moves it?

Loading the corpus…

The short version

Most websites were built for people and search crawlers, not for the machines that now answer questions about them. We can prove that with numbers instead of asserting it: the figures on this page are medians of real scores from real public pages, recomputed daily. Where a breakdown is missing, it is because not enough independent sites sit behind it yet — we would rather show you a gap than a guess.

The state of AI visibility

Each row is one thing we measure, across every domain in the corpus. "Strong" and "At risk" use the same thresholds as the scores inside the product, so these numbers mean exactly what a customer's dashboard means.

Loading benchmarks…

Which engines cite anyone at all

The same question asked of many engines gets very different answers. This is the median share of tracked brands each engine cites — measured from recorded answers, with engines that failed to respond excluded rather than counted as a non-citation.

It depends what you built on — and what you sell

Median AI Access score grouped by the platform a site runs on, and by industry. Platform is detected from the page itself; industry comes from how the business describes itself.

What actually moves the number

For every change shipped through AI Syndicate we record what was done and then measure the score again once it has had time to land. This is the median measured movement per kind of change — outcomes, not promises, and each one needs a real before-and-after on both sides to count at all.

How these numbers are made

Every figure is a measurement

We retrieve public pages and score them deterministically, and we record what named AI engines actually answered. Nothing here is modelled, extrapolated, or self-reported. A missing measurement produces a gap, never an estimate.

Aggregates only, always

This page publishes counts, medians, and percentages. No domain, customer, or account appears, and any statistic without enough independent sites behind it is withheld entirely rather than shown thin.

Dated, and kept

Each observation is stamped with the day it was taken and the version of the scoring method that produced it, so a change in how we score never quietly rewrites history. The record only grows.

What we don't publish

The scoring recipes themselves, per-site results, and anything a competitor could use to reverse-engineer a customer's position. Benchmarks are a public good; the method is not.

Questions

Where does this benchmark data come from?

From our own measurement corpus. Every time we score a website — a customer's site, a competitor we were asked to benchmark against, or a site scanned through our free tools — the resulting score is recorded as a dated observation of that public page. These benchmarks are aggregates of those observations.

Can anyone tell whose sites these numbers came from?

No. Counts, medians, and percentages only. No domain, customer, or account is ever included, and a statistic is withheld unless enough independent sites sit behind it to be meaningful.

Why is a statistic sometimes missing?

Because a median of a handful of sites would mislead. Each breakdown appears only once enough distinct domains have been measured for it.

How often does this update?

The corpus grows continuously as scans run; these aggregates are recomputed daily.

Where does your site sit against these numbers?

Run the same measurement we ran on every site above. It takes about a minute and costs nothing.

Get your AI visibility score