The most valuable thirty seconds in outbound is the moment a real prospect lands on your site after reading your email. They're curious enough to click through, they're on your page right now, and what they do next tells you more about their intent than any open or click ever could. A person who lands and immediately opens your pricing page is a fundamentally different lead than one who bounces off the homepage in two seconds.
But there's a catch, and it's the same catch that poisons every other email metric: a huge share of your "landings" aren't people. Before a single intent signal is worth anything, you have to answer one question — was this a human at all?
"On-site behavior is the richest intent signal in outbound. It's also the easiest to fool yourself with, because roughly 40% of the traffic hitting your page from an email was never a person to begin with."
First, the bot problem — and the one honest way to solve it
When we first started logging landing events on our outbound pages, the numbers looked incredible. Huge traffic, lots of "engaged" visits. Then we looked closer and found that a large fraction of it — around 40% on some campaigns — was Microsoft's Safe Links scanner, pre-fetching our pages to check them for malware before the human ever clicked. It wasn't traffic. It was a firewall.
Every technique we tried for filtering it out based on behavior failed. Bots that load a page in 50 milliseconds are easy to catch, but the sophisticated scanners render the page, wait, and look human on every behavioral axis. The only signal that reliably separates a person from a machine is who owns the IP address.
Every visit's source IP belongs to an autonomous system — an ASN — and that ownership is the tell. A residential ISP or a mobile carrier is a human. A cloud or hosting provider is a machine. We classify every landing by the ASN that owns its IP and set a single is_bot flag on the event. One flag, computed once at write time, and every report, every intent score, every dashboard filters on that same flag. The 40% never pollutes anything downstream, because it's marked the moment it arrives.
Second, don’t confuse a beacon with a landing
There's a subtler trap. Email tracking pixels and click-redirects fire HTTP requests that can look, in your logs, a lot like a page visit. If you count those as landings, you've double-counted the email signal as an on-site signal and inflated your intent numbers with events that never involved anyone actually viewing a page.
So the pipeline explicitly separates tracking beacons from real page loads. A pixel firing is an open. A redirect resolving is a click. Only a genuine, rendered page view by a bot-filtered human counts as a landing. Keep those three event types in their own lanes and your intent model stays honest.
Now the signal: what a real human’s path tells you
Once you're looking only at verified human landings, the behavior becomes a genuine intent ladder. Here's how we weight it.
| Behavior | What it signals | Intent |
|---|---|---|
| Lands, opens pricing | They're evaluating cost. This is a buying-mode action — nobody reads pricing idly. | High |
| Lands, hits contact / booking | They're one hesitation away from reaching out. The warmest pre-reply signal there is. | High |
| Reads a specific service page | They have a particular problem and are checking if you solve it. Targeted interest. | Medium |
| Returns a second time | One visit is curiosity. A return visit days later is consideration — they're still thinking about it. | High |
| Lands, bounces in seconds | Wrong fit, wrong moment, or the email over-promised. Low signal, but worth noting the page that failed. | Low |
The two signals worth building your follow-up around are the pricing/contact hit and the return visit. A prospect who lands from your email, reads your pricing, leaves, and comes back two days later has effectively raised their hand without typing a word. That's the lead you reach out to personally, today, referencing exactly what they looked at.
"Someone who read your pricing page twice this week is warmer than someone who opened your last five emails. Behavior beats metrics — because behavior is the prospect choosing to spend attention, and attention is the one thing they can't fake."
Turning the path into a follow-up
The intent signal is only useful if it changes what you do next. A high-intent landing should feed the same hot-lead machinery a reply does — flag the record, notify a human, and arm that human with context. The follow-up practically writes itself when you know the page:
Hi Dana,
Following up on my note from earlier this week — no pressure at all. If you were weighing whether this fits the budget, happy to walk through exactly what’s included and where the flexibility is. Sometimes it’s easier to just ask than to work it out from a page.
Want me to put together a quick, specific number for your situation? Takes me five minutes and there’s no obligation.
— Charles
Note what the email does not say: it never mentions that we watched them visit two pages. That would be unsettling. It just meets them where their behavior said they are — evaluating cost — and offers to make that easier. The intent signal informs the angle, not the content.
The whole point: rank by what people do, not what machines trigger
Landing-page intent is the strongest read you get on a prospect short of them replying — but it only works if the foundation is clean. Filter the bots by IP ownership. Separate beacons from real page views. Then weight the behavior of the humans who remain. Do that, and your outreach dashboard stops showing you inflated vanity traffic and starts showing you a ranked list of people who are, right now, quietly deciding whether to work with you.
Those are the ones to call. Not the ones who "opened." The ones who came back to read your pricing a second time.