Docs
Guides Insights & reporting How open & click tracking works

How open & click tracking works

RadarSend measures who opens your emails and who clicks your links, then filters out the automated “opens” and “clicks” that machines generate so your numbers reflect real people. This guide explains how the tracking works and — importantly — why your open and click rates may look lower here than in a tool that counts everything.

How opens are tracked#

Every email RadarSend sends includes a tiny, invisible tracking image (often called a “pixel”). When someone opens your email and their mail app loads that image, RadarSend records an open.

That image is served from your tracking domain — a branded subdomain on your own domain, like track.acme.com. Open and click tracking don’t work until your tracking domain is live, so this is a required setup step. See Set up your tracking domain.

A couple of things worth knowing:

  • A click also counts as an open. If someone clicks a link without their mail app ever loading the pixel, RadarSend still counts them as an opener in your filtered numbers.
  • Opens are inherently fuzzy. Because opens depend on images loading, they’re the noisiest signal in email — which is exactly why the filtering below matters.

How clicks are tracked#

When RadarSend sends your email, it rewrites the links in it so they route through your tracking domain first. A recipient clicks the link, RadarSend records the click, and then instantly forwards them to the real destination — they just land on the page you linked to, usually without noticing the hop.

This is why links need a live tracking domain too: the same branded subdomain that serves the open pixel also handles click redirects. Using your own subdomain (rather than a shared one) keeps your links on your brand and protects your link reputation.

Why your numbers may look lower than other tools#

This is the big one. If you’ve used another email tool, RadarSend’s open and click numbers may look lower. That’s not a bug — it’s the point.

A large share of “opens” and “clicks” in email aren’t people at all. They’re machines:

  • Inbox privacy features like Apple Mail Privacy Protection pre-load images for the recipient, generating an “open” whether or not a human ever looked at the email.
  • Image proxies and prefetchers (for example, Gmail routes images through Google’s servers) fetch your pixel automatically, often within seconds of delivery.
  • Security scanners at the recipient’s company (Microsoft, Proofpoint, Mimecast, Barracuda, and others) open your email and click every link to check them for threats — before the person has even seen the message.

RadarSend runs a bot classifier that looks at each open and click and sorts it into one of three buckets:

  • High (human) — likely a real person, for example a click followed by a real visit to your site.
  • Low (uncertain) — could be a person, but there’s no proof either way. Most genuine opens land here, because a pixel load on its own rarely proves a person was reading.
  • Bot — almost certainly automated. Removed from your filtered numbers.

Your filtered open and click rates count High and Low activity and leave out Bot. Some uses are stricter:

  • Lead scoring only counts likely-human activity. Uncertain and bot activity earn no points.
  • Sequences branch and trigger on likely-human activity by default.
  • Segments count likely and maybe-human activity by default, with an option to count likely-human only.

A tool that counts every pixel load and every scanner click reports higher numbers. RadarSend shows you that figure too, next to the filtered one, so you can compare.

The classifier runs about every 10 minutes. New opens and clicks are held back for about 20 minutes while scanners are sorted out, and click verdicts can keep being refined for up to an hour.

Reading your engagement numbers#

Open a broadcast to see its results. The Open rate, Click rate, and CTOR tiles on its report each show two figures:

  • The big number is the apparent rate — every recorded open or click, including bots, inbox previews, and scanners. This is the “counts everything” number, comparable to what other email tools report. The line under it says so: incl. bots & previews for opens, incl. bots & scanners for clicks.
  • ✓ Filtered — the rate with machine activity removed, and how many people that is.

While a send is still being checked, a tile may show Verifying and Reviewing activity — scanners are being filtered. until its filtered figure is ready.

The Performance column on the Broadcasts list shows the filtered open, click, and bounce rates. For a full walk through the report, see Read a broadcast report.

If you want to see the filtering itself, go to Settings→Deliverability→Bot classifier. Under Classification status it shows your whole workspace’s activity sorted into Total events, High (human), Low (uncertain), Bot, and Unclassified. To see what was removed from one broadcast, click View filtered activity → on its report.

Common issues#

  • No opens or clicks are tracking at all. Check Settings→Sending→Tracking domains and confirm the domain shows Live, and that your sender is using it. RadarSend won’t send a broadcast from a sender whose tracking isn’t set up, so this usually shows up as a blocked send rather than empty numbers. Also check that the broadcast’s link tracking wasn’t set to Pass-through. See Set up your tracking domain.
  • My open rate dropped after switching to RadarSend. Expected if you’re looking at the filtered rate, which leaves out bots and Apple/Gmail image prefetching. For a like-for-like comparison with your old tool, use the big apparent number on the report.
  • A specific click didn’t register. Some security setups strip or block tracked links, and some clicks get classified as scanner activity. One missing click among many is normal; a pattern of zero clicks points back to the tracking domain.
  • Numbers changed a few minutes after sending. The bot classifier runs about every 10 minutes, and new activity is held for about 20 minutes while it’s checked, so a fresh send’s rates settle as automated events are filtered. This is expected.

Good to know#

  • Website tracking sharpens the picture. Turning on website tracking (part of the setup checklist) lets RadarSend confirm that a click led to a real visit in a real browser and mark it as likely human. That matters most for lead scoring and sequences, which count likely-human activity. Without it, more clicks stay uncertain.
  • The apparent number is never thrown away. RadarSend shows the full total next to the filtered rate, so you always have a like-for-like comparison with tools that don’t filter.
  • Advanced users can tune the filters. The Bot classifier settings page lets you adjust the detection thresholds and Save thresholds, or Run now to reclassify straight away. The defaults are tuned for typical audiences, and most people never need to touch them.
  • A tracking domain is separate from your sending domain. The sending domain authenticates who sends; the tracking domain measures what recipients do.