How tracking links and images hurt Gmail deliverability

Receivers score the domains inside your message, not just the domain you send from. When your ESP rewrites a link to click.sometool.net/abc123, every recipient's filter evaluates that host's reputation as well as yours. If it is a shared tracking domain, you are carrying the aggregate behaviour of everyone else who uses it — including the senders who got themselves blocklisted last week.

That is the first cost. The second is structural: tracking pixels and rewritten links make a message look like a campaign, which is how it ends up in Promotions.

The three costs, separated

1. Link domain reputation. URL and domain blocklists are consulted independently of IP and sender reputation. A tracking host that appears in abusive mail gets listed, and your message inherits the verdict. This is the single most under-diagnosed cause of "clean sender, mail still filtered".

2. Message structure. A tracking pixel adds a remote image to a message that would otherwise have none. Rewritten links add a host that only ever appears in marketing mail. Both are inputs to Gmail's categorisation, and both push toward Promotions — where the message is still technically inboxed, since Promotions carries the `INBOX` label, but far less likely to be read.

3. Redirect chains. A link that bounces through a tracker before reaching your real destination looks like cloaking, which is a pattern filters are specifically built to catch. Short, single-hop redirects on a domain that is recognisably yours are much safer than long chains through unfamiliar hosts.

Shared versus custom tracking domains

If you keep click tracking, use a custom subdomain of your sending domain — links.yourdomain.com — rather than the provider's default. It is a CNAME in your DNS and most providers automate the certificate. Use a subdomain, not your root domain, so that a problem with the tracker is contained.

New tracking subdomains have no reputation at all, so introduce one before you need it and let it accumulate history alongside your warm-up rather than switching mid-campaign.

Pixels: worth it or not?

For cold outreach, the honest answer is no. The pixel buys you a number that Gmail has already made unreliable — image requests are proxied through Google's infrastructure, cached, and sometimes prefetched before a human looks at the message, and security scanners fetch pixels on messages nobody opened. So you accept a structural deliverability cost in exchange for data you cannot act on confidently.

For permissioned bulk mail where reporting depends on opens, keep it, but do not use opens as a suppression criterion and do not read them as a placement measure.

A configuration that keeps most of the benefit

1. Custom tracking subdomain on your sending domain, with a valid certificate.
2. One tracked link per message, maximum, in cold outreach.
3. No open pixel on first-touch cold mail.
4. No URL shorteners. Public shorteners are heavily abused and filtered accordingly, and they hide the destination, which is itself a negative signal.
5. Destination domains that host real content. A link to a bare landing page on a week-old domain is scored on that domain's non-existent reputation.
6. Plain text or near-plain HTML in outreach, so the one link is the only machinery in the message.
7. Consistency. Switching tracking domains repeatedly means you never accumulate reputation on any of them.

Diagnose whether links are your problem

Send the same message twice to a set of seed mailboxes: once with tracking as configured, once with the pixel removed and links unrewritten to their real destinations. If placement improves materially in the second version, your tracking configuration is the cause, not your copy or your sender reputation.

Cross-check with what Google publishes. Postmaster Tools v2 gives spam rate, authentication pass rates, delivery errors and compliance status, plus a Deliverability analysis report added in June 2026 — note there is no reputation grade any more, since the domain and IP reputation dashboards retired with v1, which stopped serving senders on 30 September 2025 and was gone by the end of 2025. A near-zero spam rate alongside poor seed placement points at structure and links rather than reputation. Keep the spam rate under Google's 0.30% threshold and aim below 0.10% regardless.

What not to do

• Do not use the ESP's default shared tracking domain if you have any volume worth protecting.
• Do not use a public URL shortener in email. Ever, in outreach.
• Do not point a tracking subdomain at a host you do not control.
• Do not add tracking to warm-up traffic. Warm-up exists to generate clean engagement, not to be measured by pixels.
• Do not change tracking domains to escape a problem without fixing what caused it; the new domain simply has no reputation yet.
• Do not conclude from a healthy open rate that links are fine. Proxies and scanners inflate that number independent of placement.

How MailPilot helps

MailPilot measures placement from seed mailboxes that read folders and Gmail tabs directly, so it reports where a message landed without depending on a pixel — which makes it a practical way to test a tracking configuration against an untracked control. Warm-up over OAuth or SMTP/IMAP builds the sender-side reputation separately, and SPF, DKIM and DMARC are checked on the domain you send from.