The AI economy runs on content. Yours.
Every article you publish is crawled by models that will answer your readers' questions without sending the click. Viz measures exactly how much of your traffic is machines, lets you decide which ones get in, and shows whether AI at least cites you when it borrows.
Publishing's new bargain, signed without your consent.
Trained on, uncompensated
Training crawlers ingest your archive to build products that compete with you for attention. Most publishers can't even measure how much is being taken.
Answers without clicks
AI search summarizes your reporting directly in the answer. The reader is served, the pageview never happens, and the ad or subscription revenue goes with it.
Analytics that undercount
JavaScript trackers only see browsers. On content sites, a large share of real request volume is crawlers that never appear in any report you look at.
Measure the crawl. Choose your terms. Verify compliance.
Know what the machines take
Server-level counts of every AI crawler on your archive - which bots, which volume, trending over time. The number you need before any licensing conversation. Analytics →
Block training, keep search
Say no to training crawlers with robots.txt plus enforced 403s while Googlebot indexes normally - and a built-in probe verifies blocked bots are actually being stopped, even behind your CDN. AI Control →
Get credit when AI quotes you
Track whether AI answer engines cite you as a source for the topics you cover - and which outlets win the citations when they don't. AI visibility →
The crawl-to-click ratio is the number publishing runs on now.
For twenty years the deal with crawlers was simple: Googlebot reads your pages, and search sends you readers. The exchange rate was good enough that nobody thought to measure it. AI changes the arithmetic. Training crawlers can read your entire archive and, by design, send nothing back. Retrieval bots fetch your reporting to answer a reader's question in place - the value is delivered, the visit never happens.
That makes the ratio between how often machines read you and how often they send you anyone the single most consequential number in your analytics - and it's a number JavaScript trackers cannot compute, because they never see the numerator. Server-side measurement is the only place the full fraction exists.
Measurement is also leverage. The publishers striking licensing deals with AI companies are the ones who can document what is being taken: which bots, which sections, what volume, over what period. A crawl log is an inventory of usage - and robots.txt plus enforced 403s are how you set terms for anyone who won't pay. You don't need to decide today whether to block, meter, or license. You need the data that makes any of those decisions possible.
Common questions, straight answers.
How much of a typical content site's traffic is actually bots?
It varies widely, but public industry estimates consistently put automated traffic at a large share of total requests - often approaching half - and AI crawlers are its fastest-growing component. The honest answer for your site requires measuring at the server, which is exactly what a five-minute Viz install gives you.
If I block AI training bots, will I disappear from Google?
No - as long as you block the right bots. GPTBot, ClaudeBot, and CCBot are separate from Googlebot and Bingbot, and blocking them does not affect search indexing. The one nuance is Google-Extended: it is a robots.txt control that limits AI training use of your content without touching your search presence.
Do AI companies respect robots.txt?
The major operators - OpenAI, Anthropic, Google - publicly commit to honoring it, and generally do. Others have been credibly reported ignoring it. That is why Viz layers robots.txt with server-level 403 enforcement and then probes your site from outside to verify the block actually holds, including through your cache or CDN.
Can I tell whether AI engines at least cite my reporting?
Yes. AI visibility tracking runs the questions your readers ask against the major answer engines and records whether you are cited as a source, how prominently, and which outlets win the citation when you are not. For publishers, citations are the new front page.
Will logging every request bloat my database?
You control retention: set a window that fits your hosting, toggle whether human visits are logged at all, and let old events expire. High-volume publishers can opt into cloud sync for longer history without growing the WordPress database.
Should I block the AI search bots that answer readers without sending the click?
This is the genuinely hard call, because retrieval cuts both ways: the same fetch that powers a click-less answer also powers the citation that sends you a reader. Blocking retrieval removes both. The data-driven approach is to measure your own ratio first - retrieval crawl volume against AI-referred sessions - and decide per section: keep retrieval open where citations send meaningful traffic, and consider gating the content where machines take answers and never send anyone back.