Shared Tracking vs Custom Tracking vs No Tracking: What Actually Wins

What Each Option Actually Does to Your Emails

• •Shared tracking (sequencer default): your pixel and every rewritten link load from a domain used by the platform's entire customer base. One spammer's campaign this morning degrades the domain you're sending through this afternoon. Filters blocklist and score these domains constantly, it's the closest thing to a self-inflicted blacklist in cold email.
• •Custom tracking domain: a CNAME (track.yourdomain.com) pointing at the sequencer's tracker. Reputation isolation achieved, but the email still contains a hidden remote pixel and rewritten links, which are content-level marketing fingerprints independent of whose domain serves them.
• •No tracking: the email is just... an email. No hidden content, no redirects, nothing for a content model to flag. Combined with plain-text formatting, it's indistinguishable from the one-to-one correspondence filters protect, the whole game, especially at Microsoft (see the Outlook playbook).

The Decision, By Situation

• •Sending at scale (500+/day): no tracking. The marginal placement cost of pixels is multiplied by volume, and at this scale you have enough reply data to optimize on real signal. This is our fleet standard, the full reasoning and what changed when we switched is in the open-tracking piece.
• •Early testing phase (first few weeks): custom tracking domains, opens read as rough trend only, with a calendar date to turn it off. Defensible, briefly.
• •Microsoft-heavy lists: no tracking, no debate. Segment first with the ESP Segmenter; if a meaningful share of your list is Microsoft or gateway-protected, the pixel is costing you real placement.
• •Still on the sequencer default: fix it today, either configure custom tracking domains or disable tracking entirely. This isn't one of the debatable ones.

If You Run Custom Tracking, Run It Right

• •One tracking subdomain per sending domain, pooling all sending domains onto one tracker rebuilds the shared-reputation problem at household scale.
• •Subdomain of the sending domain, never the main company domain, keep outreach risk off your primary web presence entirely (same isolation logic as sending).
• •HTTPS on the tracker, an insecure redirect domain is its own filter flag.
• •Read opens as weather, not truth, scanners and Apple MPP still pollute them; the honest metrics remain replies and weekly seed tests via the Header Analyzer.