Cold email lives through a strange paradox: volumes grow every year while replies fall. The reason is known — 90% of sequences are built on a template with {first_name} and {company} placeholders. A recipient detects a machine-written email in 1.8 seconds of reading. And deletes it in 0.4.

This breakdown shows how next-gen AI sequences reach 61% open rates and 8+ replies per 100 contacts — not through a “smarter template”, but through real per-contact research.

The death of {first_name} personalization

Template variables aren’t personalization — they’re personalization cosplay. A recipient reads “Hi Dmitry! Noticed Nordwind is growing…” and understands: 4,000 other people got the same email. Trust in the channel is so eroded that even good offers land in the spam folder — and the mental one too.

The humanity check: if the pitch in your email can be sent to anyone on your list without changing a word — it’s spam with nice formatting.

Digital-footprint research

The working scheme of 2026 AI outreach: for each contact the engine compiles a dossier from open sources — recent LinkedIn and X posts, interviews, company job postings, product releases, stack changes (via job descriptions), funding rounds. Out of 30–60 signals, one is selected — the most relevant to your offer.

  • Hiring: 3 SDR roles in a week → the company is scaling sales
  • Release: launched a self-serve tier → they’ll need support and onboarding
  • Migration: hiring Kubernetes engineers → infrastructure move
  • Round: Series B a month ago → fresh budget for growth
  • Content: the CEO posted about the pain you solve → the perfect hook

“Why now” — the only hook that works

A strong first email answers “why now, specifically”. Not “what we do” — “why this is relevant to you this month”. Compare:

— Template: “We help SaaS companies grow pipeline.
   Want to hop on a call?”

+ Why-now: “Dmitry, I noticed Nordwind opened three SDR
   roles this week — that usually means inbound flow can’t
   keep up. We close that gap with organic outreach, no new
   hires. Open to a 12-minute call?”

The second email cannot be sent to anyone else: it’s built on a specific signal of a specific company. That’s the criterion of a true 1:1 sequence — every paragraph is unique.

Deliverability: the foundation under the email

The best personalization is useless from the spam folder. The baseline delivery stack:

  • SPF / DKIM / DMARC — without them any ESP cuts up to 40% of delivery
  • 14-day mailbox warm-up curve: start at 4–6 emails, grow +12%/day with real replies
  • A separate outreach domain (company-mail.com, not company.com) — protects core reputation
  • 40–60 emails/day per mailbox is the ceiling; beyond that, rotate a pool via an SMTP provider like HostUp
  • Bounce hygiene: verify every address before the first send, 2% ceiling

Metrics to watch

  • Open rate > 55% — a sign of healthy inbox placement (below 40% — dig into deliverability, not the copy)
  • Reply rate 5–10% — a healthy 1:1 sequence; < 2% — a targeting or offer problem
  • Positive reply rate — count separately from “unsubscribe me”: target 4–6% of all sends
  • Spam complaint < 0.1% — above this Gmail starts throttling the whole domain

Where humans are irreplaceable

AI researches, writes and varies brilliantly. But the final filter stays human: review the first 30–50 emails of each batch, catch tonal drift and train the model with your edits. In Boostly this loop is built in: drafts arrive in Telegram with “Send / Edit / Skip” buttons, and every edit fine-tunes your persona.

A cold email in 2026 isn’t a blast. It’s 200 individual letters that turned out cheaper to write by machine than to fix with spam filters.

Boostly Outbound Report, Q3 2026