An open notification is a nudge to act. Call now, follow up now, strike while they're reading. But some of those nudges come from your prospect, and some come from a piece of security software that touched your email two seconds after you sent it.
Acting on the wrong kind is worse than doing nothing. "Saw you checked out my proposal!" lands badly when the prospect never opened it. So before you follow up, run the notification through a quick sniff test.
What makes an open notification trustworthy?
Four signals, and any one of them is meaningful. The open graded Tier 1 or 2, it fired at a human hour, it's a repeat open on the same thread, or it came with a click on a real link.
Each of these is something machines rarely do and people often do. Stack two or more together and you can act with real confidence. A Tier 1 open at 10 a.m. their time, on a thread they opened yesterday too, is a prospect with your email on screen. That's the moment follow-up timing actually matters.
Which signals say a real person opened it?
Repeat visits and real clicks. A person who cares about your email comes back to it, and when they click, they click the one link they want.
Repeat opens are especially hard for machines to fake. Scanners process a message once, when it arrives. A person re-opening your proposal on Tuesday and again on Thursday is doing what buyers do: rereading before a decision, forwarding to a colleague, checking a number. Time of day helps too. Opens during the recipient's working hours fit human behavior. Opens in the middle of their night usually don't.
Which signals say a machine opened it?
Speed and bulk. An open that lands seconds after you hit send, or an open where many links got fetched at once, is security software doing its job.
That's exactly how corporate scanners like Mimecast, Proofpoint, and Microsoft Defender behave: they open every link and image the moment mail arrives, checking for malware. Around 15 to 25 percent of B2B recipients sit behind one. Apple adds its own noise, since Apple Mail pre-fetches tracking pixels for every recipient whether they read or not. And some machine traffic carries known scanner signatures that give it away outright. The mechanics of how opens get recorded in the first place are covered in our glossary entry on the [email tracking pixel](/glossary/email-tracking-pixel).
How do you check a specific open?
Click it and read the evidence. Outsolvi's [Diagnose view](/features/diagnose-open-evidence) lets you open any event and see exactly why it counted or was filtered.
Behind that view, every open gets graded from Tier 1, high-confidence human, to Tier 5, bot or scanner, using request timing, network origin, and client signature. Anything below a 25 percent confidence floor never reaches your stats. The model agrees with human-rated classifications 95 to 98 percent of the time, and Tier 1 false positives run under 2 percent. The full grading logic is public on the [methodology page](/methodology), and there's more on the tiers on the [confidence scoring](/features/confidence-scoring) page.
Why does acting on the wrong open cost you?
Because the prospect can tell. A follow-up that says "noticed you've been looking at the proposal" reads fine when it's true and reads terribly when it isn't.
If the open was a scanner, the prospect never saw your first email at all. Now your second one references activity that didn't happen, and you look like you're either guessing or watching them a little too closely. Neither helps. There's a quieter cost too: every follow-up you spend on a phantom open is time not spent on the thread where a real buyer opened twice yesterday. Bad open data doesn't just waste effort. It points your effort at the wrong people.
What if you're still not sure?
Wait for a second signal. One ambiguous open isn't a reason to do anything, but a second open on the same thread, or a click, turns a maybe into a yes.
That patience costs you very little. Real buyers generate more than one signal, and machine noise usually doesn't repeat on the same thread. If your current tracker gives you nothing but a raw count and a ping, that's worth fixing at the source. Our [comparison hub](/compare) shows which trackers grade opens and which just count them, and the 14-day free trial, no credit card, lets you watch your own notifications get sorted into trustworthy and not for two weeks.