Here's an uncomfortable fact: the [open rate](/glossary/email-open-rate) on your dashboard is wrong. Not off by a little. On a typical B2B list it runs 2-3x inflated, which means a reported 70 percent open rate is often 25-35 percent real reads.
That's not a rounding error. That's a different story about your outreach.
Where does the inflation come from?
Three places, and none of them are your prospects.
Apple Mail Privacy Protection is the biggest one. Apple pre-fetches tracking pixels on its own servers, so every Apple Mail recipient registers as an opener whether they read your email or not. Litmus puts Apple Mail at roughly 58 percent of email client share. On a list that skews toward Apple users, this single mechanism can flip most of your open data from signal to noise.
Corporate security scanners are the second. Mimecast, Proofpoint, and Microsoft Defender open every link and image in incoming email to check for malware. Between 15 and 25 percent of B2B recipients sit behind one of these. Every email you send them gets machine-opened, usually within seconds of send.
Gmail's image proxy is the messy third. Gmail serves images through its own proxy, so the request your tracker sees didn't come straight from the reader. Some proxy opens are real humans reading. Some aren't. Telling them apart takes the full signal picture, not blind counting and not blind discarding.
How much does it change the math?
A lot. Say you send 100 cold emails and your tracker reports 70 opens.
If half your list is on Apple Mail with MPP on, those recipients register opens automatically. If a fifth of your list sits behind a scanner, those register too, instantly. Strip the machine opens out and you're usually left with 25 to 35 real reads out of that reported 70.
Now think about what you did with the inflated number. You followed up with "openers" who never saw your email. You wrote off subject lines that were actually fine. You told your team the top of the funnel was healthy when it wasn't. Wrong data doesn't just sit there. It steers.
Which lists get hit hardest?
Lists heavy on Apple Mail users and corporate domains. That describes most B2B lists.
Selling to consumers on personal iPhones? Nearly every recipient has Apple's pre-fetch working against your data. Selling to mid-size and enterprise companies? A big slice of those inboxes sit behind Mimecast, Proofpoint, or Defender, and every send gets machine-opened on arrival.
The frustrating part is that the inflation isn't even. One campaign might land on a scanner-heavy segment and report sky-high opens. The next hits a segment with fewer machines and looks like a flop. Neither number reflects your subject lines or your list quality. You end up "learning" lessons from noise.
Why don't trackers fix this?
Mostly because counting pixel loads is easy and grading them is hard. A tracker that reports every load gets to show you a big flattering number. A tracker that filters has to explain why your open rate just dropped by half, and that's a harder sales pitch, even when it's the honest one.
Our take is simple: no score beats a wrong score. If a tool can't tell you which opens were human, the big number it shows you isn't a feature.
What does a trustworthy open rate look like?
It's a graded one. Outsolvi scores every open from Tier 1 (high-confidence human) down to Tier 5 (bot or scanner), using request timing, network origin, and client signature. Anything below a 25 percent confidence floor is excluded from your counts. The full rules are public on the [methodology page](/methodology), and the [confidence scoring feature](/features/confidence-scoring) page shows how the tiers surface in the app.
The model agrees with human-rated classifications 95-98 percent of the time, and Tier-1 false positives run under 2 percent. And when you want to check a specific open, the [Diagnose view](/features/diagnose-open-evidence) shows you exactly why it counted or got filtered. Click the number, see the evidence.
What should you do this week?
Pull up last month's open rate and ask one question: does this number include machines? If your tracker can't answer, you have your answer.
Then compare. The [comparison hub](/compare) breaks down how the major trackers handle this. Or run the test directly: Outsolvi is $7 a month billed yearly, with a 14-day free trial and no credit card needed. Two weeks of real sends will show you the gap between your reported opens and your real ones. Fair warning: the real number is smaller. It's also the only one worth acting on.