The only metric that matters.

In cold outbound, there's only one number that decides whether a campaign worked: positive replies.

Emails go out. Some people write back. Some of those replies are positive. We track plenty of metrics, but the positive reply is the number we live or die by, and ultimately it's the number our clients come to us for. These are real people who've put their hands up to learn more about the event we're promoting.

What counts as positive

Here's how we define a positive reply. Every reply that comes back is read by a classifier and marked interested or not interested. The classifier works from a prompt that encodes hundreds of variations of what a positive reply looks like, and it applies that definition to every reply, in every campaign.

We use tools like LLMs and more recently TypeSafe (aka Jev), which is a decision-making model that only outputs structured text.

Across our campaigns, between 50% and 80% of replies are positive. Most of the rest are some version of "no thanks."

"But why not track opens as well?"

If you come from marketing, where opens and click-through are the standard vocabulary, you might be a bit shocked when you see your first Outstation report sans opens. The answer starts with what an open actually measures today. Basically, opens are a great metric for measuring bot behaviour, and a poor one for measuring human behaviour. And after all, humans are the ones you want at your show, right?

Stripo's 2026 B2B study, covering around 15 billion emails, found that about 49.29% of tracked opens are Apple Mail Privacy Protection loading the message on Apple's schedule rather than a person reading it. Litmus put Apple Mail at around 60% of all tracked opens in December 2025. On enterprise B2B lists, where most inboxes sit on managed Outlook, the Apple share is lower, at roughly 20–30%, but it's still a big slice of the number.

Gmail bends the number the other way too. Since 2013 it has routed every image through its own proxy on googleusercontent.com, so the tracking pixel is fetched from Google's servers (66.249.x.x) before the recipient opens the email. Gmail then caches the image for up to 72 hours, and when a real reader comes back to the message, nothing registers. The same dataset over-counts strangers and under-counts the people who actually read. Security gateways add another layer.

And it isn't free

This is where open tracking stops being merely unreliable. The damage lands on inbox placement: fewer of your emails reach the primary inbox, so fewer get read. The pixel fires anyway, from Apple's prefetch, from Gmail's proxy and from the gateways. So the open rate goes up while readership goes down.

Summing up

Forget about open tracking entirely and favour positive reply rate. It's the only number tied to revenue.