The Three Email Metrics That Stopped Working in 2026

There are three email metrics that almost every B2B sales team still tracks and almost every B2B sales team should stop trusting in the way they currently do. The metrics aren’t broken; what’s broken is how they’re measured.

1. Open rate

The textbook definition: share of delivered emails where the recipient opened the message. In 2025 and earlier, this was approximately what was being measured. In 2026 it isn’t.

Apple Mail Privacy Protection (MPP), launched with iOS 15 in September 2021, pre-loads every remote image (including tracking pixels) on Apple-controlled relay servers before the recipient opens the email. With Apple Mail at roughly 58% global email-client market share (Litmus), MPP affects half or more of every B2B list.

Corporate email scanners (Microsoft Defender for Office 365, Mimecast, Proofpoint) pre-fetch every link and image on every email for malware scanning. Every enterprise B2B prospect generates a fake open within seconds of delivery.

Gmail’s image proxy can register the same pixel as multiple opens. Smaller problem, still real.

The result: a reported 70% open rate typically maps to a 25–35% actual human-read rate.

The fix is confidence scoring: grade each open from Tier 1 (high-confidence human) to Tier 5 (bot/scanner) based on IP block, User-Agent, and timing signature. Filter below a 25% confidence threshold.

2. Click-through rate

The same problem reaches clicks. Microsoft Safe Links, Mimecast URL Protect, and Proofpoint TAP pre-fetch every link on every email for malware scanning. Every link gets clicked automatically by the scanner within seconds of delivery.

Unfiltered click-through rate in 2026 is contaminated 2–3x by scanner auto-clicks. The fix is unique-clicks-per-recipient with scanner-IP filtering and a time-window filter (any click within 5 seconds of delivery is almost always a scanner, not a human).

3. Time-to-first-open

When the open is a machine pre-fetch firing 30 seconds after delivery, the average time-to-first-open metric becomes meaningless. The measure that actually matters is time-to-first-Tier-1-open: when did a real human first engage with the email?

The structural lesson: 2026 email tracking only works if it filters MPP, scanner pre-fetches, and proxy duplicates. Tools that count raw pixel loads are reporting noise, not engagement.

I wrote up the technical breakdown of how MPP actually works here: https://outsolvi.com/glossary/apple-mail-privacy-protection

— Nate Summers Co-Founder, Outsolvi