Your tracker says the email was opened. Was it? Honestly, maybe. In 2026, a big share of "opens" happen with no human anywhere near the inbox.
This piece explains what a real open is, who's faking the rest, and how to tell them apart.
What counts as a real email open?
A real open is simple: a person loaded your email on a real device and looked at it. That's it. That's the thing you actually care about when you check your tracker.
The catch is that no tracker sees people. Trackers see a tiny invisible image, the [tracking pixel](/glossary/email-tracking-pixel), getting loaded from a server. Every load gets counted as an open. And humans aren't the only things that load pixels. Not even close.
Why can't most trackers tell the difference?
Because most trackers count pixel loads and stop there. A load is a load. One open, tick, next.
That worked fine in 2015. It doesn't work now, because three big machine sources load pixels at scale:
Apple Mail Privacy Protection. Since iOS 15, Apple pre-fetches tracking pixels on Apple's own servers. Every Apple Mail recipient looks like an opener, even if they never read the email. And Apple Mail is roughly 58 percent of email client share, per Litmus. That's not an edge case. That's most inboxes.
Corporate security scanners. Tools like Mimecast, Proofpoint, and Microsoft Defender open every link and image in incoming email to check for malware. Somewhere between 15 and 25 percent of B2B recipients sit behind one. Your email gets "opened" seconds after send, by software, every single time.
Gmail's image proxy. Gmail serves images through its own proxy. A proxy open isn't automatically fake and it isn't automatically real. It needs the full signal picture, not blind counting or blind discarding.
Add it up and raw open rates on B2B lists typically run 2-3x inflated. A reported 70 percent open rate is often 25-35 percent real reads.
What does a real open actually look like?
Real people behave like people. They open during normal hours, not at 3 a.m. sharp. They open on real devices with real email clients. Sometimes they come back and re-read your email an hour later, which is one of the strongest buying signals there is.
Machines behave like machines. They fire instantly after send, often within seconds. They often hit many links at once. They come from data centers, not from someone's phone on a Tuesday morning.
Once you know the patterns, the two groups look nothing alike. The trick is checking every open against them, automatically, at scale.
What happens if you treat fake opens as real?
You waste your best follow-ups. It's that direct. You see "opened 3 times" next to a prospect's name, so you call them, and they have no idea what email you're talking about. Because they never opened it. A scanner did.
It cuts the other way too. A prospect on Gmail reads your email twice, but your tracker either drowned that signal in noise or threw it out with the machine traffic. So the one person who was actually interested gets your slowest follow-up, or none at all.
Bad open data doesn't just pad a vanity metric. It reorders your day. It picks who you call first, which subject lines you keep, and which deals you quietly give up on. If the data underneath is 2-3x inflated, a lot of those calls are wrong.
How does Outsolvi decide what's real?
Every open gets graded on a five-tier ladder, from Tier 1 (high-confidence human) down to Tier 5 (bot or scanner). The grade comes from request timing, network origin, and client signature. Opens below a 25 percent confidence floor don't make it into your numbers at all. You can read the full breakdown of how the [confidence scoring](/features/confidence-scoring) works, and the scoring rules are documented openly on the [methodology page](/methodology).
Does it work? The model agrees with human-rated classifications 95-98 percent of the time, and Tier-1 false positives run under 2 percent.
And you don't have to take the grade on faith. The [Diagnose view](/features/diagnose-open-evidence) lets you click any open and see exactly why it counted or why it was filtered. No mystery numbers.
What should you do with this?
First, stop trusting raw open counts. They're not lying to you on purpose, but they're counting machines and calling them people.
Second, don't overcorrect and ignore opens entirely. A graded, filtered open is still one of the best signals you have about who's paying attention.
If you want to see how trackers stack up on this, the [comparison hub](/compare) covers the major ones. Or just test it yourself: Outsolvi has a 14-day free trial, no credit card, and it's $7 a month billed yearly. Send your normal emails for two weeks and see how many of your "opens" were ever real.