Cold email deliverability for link building: staying out of spam in 2026
Link building outreach lives or dies in the inbox. A perfect list and a great pitch are worth nothing if the email lands in spam, and the damage compounds: once a domain gets a reputation, every future campaign suffers.
The rules tightened considerably. Google and Yahoo introduced bulk sender requirements, and Microsoft followed with its own enforcement for high-volume senders to Outlook, Hotmail and Live addresses. Authentication is now the baseline, not an optimisation.
In this guide
Authenticate the domain firstVolume: the number that mattersWarmup and new domainsContent signals that trigger filtersList hygiene and bouncesThe pre-campaign checklistAuthenticate the domain first
Three DNS records, non negotiable on any domain you send outreach from:
- SPF listing every service authorised to send for the domain. Watch the ten DNS lookup limit; exceed it and SPF fails silently.
- DKIM signing every message, with a key that resolves and validates.
- DMARC published at minimum p=none, aligned with SPF or DKIM, ideally both, with a reporting address you actually read.
Send a test to an address at each major provider and read the Authentication-Results header. All three should say pass. If one does not, fix it before a single outreach email goes out.
Volume: the number that matters
Bulk sender enforcement kicks in at high daily volumes, but reputation damage starts far below that. For link building the practical guidance is simple: 10 to 20 new outreach emails per mailbox per day, not more.
If you need more volume, add mailboxes rather than raising the cap. Five mailboxes at 15 a day is dramatically safer than one at 75, and it isolates the damage if one domain gets flagged.
Follow-ups sent inside the same thread are far gentler on reputation than fresh cold sends, which is another reason to thread properly rather than sending a new email each time.
Warmup and new domains
A brand new domain with no sending history that suddenly emails fifty strangers is the clearest spam signal there is. Give a new domain two to four weeks of low, gradually increasing volume, mixed with real conversations that generate replies.
Never run outreach from your primary business domain. Use a separate sending domain so that a mistake cannot take down the email your clients use to reach you.
Content signals that trigger filters
- Links in a first email. One is tolerable, three looks promotional. Save URLs for the reply.
- Tracking pixels and link wrapping. Open tracking rewrites URLs through third-party domains, which filters increasingly distrust. Reply rate is the metric that matters anyway.
- Image-heavy or HTML-heavy emails. Plain text reads like a human wrote it, because a human should have.
- Attachments. Never in a first cold email.
- Money and urgency language. "Guaranteed", "limited time", "cheap" all score against you.
List hygiene and bounces
Hard bounces are the fastest way to damage a sending reputation. Verify addresses before sending, and remove any address that bounces immediately rather than retrying it next campaign.
Handle replies properly too. Bounce notifications from mailer-daemon should close the prospect out automatically, not sit in a reply queue pretending to be interest. And once someone asks not to be contacted, the whole domain should go on a permanent do-not-contact list that every campaign respects.
The pre-campaign checklist
- Separate sending domain, not the main business domain
- SPF, DKIM and DMARC all passing on a live test
- Mailboxes warmed for at least two weeks
- Daily cap set per mailbox at 10 to 20 new sends
- Plain text, one link maximum, no attachments
- Follow-ups threaded, capped at two, stopping on reply
- Bounce detection and a do-not-contact list wired in
Most of that list is configuration you do once. The rest, per-mailbox caps, threading, stopping on reply, bounce handling and the do-not-contact list, is exactly what LinqDesk enforces automatically so nobody on the team has to remember it.