Blog · Visitor analytics
Track your visitors without tracking your visitors
Most of your traffic arrives anonymous and leaves silent. The good news: you don't need their identity to understand their intent. Here's how behavior-only tracking works.
The short version
Behavior is the signal. Identity is optional.
You can capture everything that matters — pages, searches, clicks, returns — without a single name or email address.
Every site has a silent majority: visitors who browse, compare, and leave without converting. Traditional analytics treats them as a page-view aggregate. Modern growth work treats them as individual profiles of intent — anonymous, but deeply understood.
The path is: capture behavior → build a per-visitor memory → let an agent act on it. This guide covers each step, with a focus on staying privacy-safe the whole way.
Step 1
Decide what to capture
Start with the signals that predict intent. You don't need to instrument everything on day one:
Page views
Which pages tell you a visitor is serious? Pricing, features, comparisons.
Searches
What visitors type reveals problems they're trying to solve.
CTA clicks
Free trial, demo, add-to-cart — the moments before commitment.
Return sessions
A second visit is a stronger signal than a tenth page view.
Step 2
Capture it with one script
A single script tag captures all of it. No event taxonomy, no SDK wiring across pages, no data layer to maintain.
// in your <head>
<script src="https://your-site.tradly.io/pixel.js" data-site="your-key">
</script>
// page views, searches, and CTA clicks are captured automatically
pixel.track("search", { query: "delivery software" });
Each event is attributed to an anonymous visitor id — a random string, never an email or identity. Memory accumulates per visitor as they browse.
Step 3
Turn events into memory
Raw events are noise. Memory is a profile: this visitor returned twice, searched for delivery software, and reached pricing but never started a trial.
Customer memory / maya-84f2
2
return sessions
3
searches
Pricing
last page
High
intent score
This is the difference between analytics and memory. Analytics says "pricing had 800 views." Memory says "here's one visitor who is close to buying."
Step 4
Let an agent act on it
The payoff is automation. With memory exposed through MCP, an agent can answer questions and take action:
Agent question
Who should get a recovery email right now?
Memory answer
42 high-intent visitors added a plan to cart and never started checkout.
Agent action
create_recovery_segment({ event: "add_to_cart" })
Why this stays private
Behavior without identity
Nothing here requires cookies, cross-site IDs, or personal data. Each visitor is an anonymous identifier scoped to your site. You learn what they do, never who they are. If a visitor later identifies themselves (like filling a form), you can connect memory to a known profile — but you never need it to start.
See your first visitor profile in minutes
Install the pixel, watch anonymous memory build, and hand it to an agent. No identity data required.
Install the pixel