Deliverability: the unglamorous 60% of the game
Google and Microsoft's 2024–26 sender crackdowns made infrastructure the gate: secondary sending domains (never your primary), full SPF/DKIM/DMARC alignment, 2–4 week inbox warming, volume discipline (under 30 cold sends per inbox daily), and list hygiene with verified addresses and sub-0.3% bounce rates. Spam-rate monitoring in Postmaster tools is non-negotiable — one bad campaign can poison a domain for months.
Budget reality: proper infrastructure for a 3,000-contact/month motion runs $300–800/month in domains, inboxes, warming and verification tooling before a single email is written.
Personalization that survives the slop filter
Recipients now pattern-match AI-generated flattery instantly ('loved your recent post!'). What converts is evidence of actual attention: a specific number from their announcement, a real observation about their stack or hiring, a relevant comparison to a peer company. AI's proper role is research synthesis at scale — enriching each contact with verifiable facts — while templates keep the claim structure human-written.
The test for every line: could this sentence have been written about anyone else? If yes, cut it. Three specific sentences outperform ten generic paragraphs at any volume.
- AI for research enrichment; humans for claim structure
- One verifiable specific per email minimum
- Three sentences beat ten paragraphs
- The ask: one clear, small next step
Sequences, follow-ups and honest benchmarks
Structure that performs: 3–4 touches over 10–14 days — opener with the specific insight, a value-add follow-up (relevant data or resource, not 'bumping this'), a different-angle third touch, and a clean breakup. Multichannel lifts results meaningfully: LinkedIn engagement before email, or a call layered on repliers.
Honest 2026 benchmarks for well-run B2B campaigns: 50%+ opens (with tracking caveats), 5–15% replies, 1–5% positive replies converting to meetings. Chalk Labs' outreach automation retainers build the whole machine — infrastructure, enrichment, sequences, CRM sync — and are measured on booked calls, because opens don't pay anyone's payroll.
Questions we hear about this
Yes for B2B in most jurisdictions with requirements: accurate sender identity, functional opt-out, physical address (CAN-SPAM), and lawful-basis considerations under GDPR for EU targets — B2B legitimate interest is workable with genuine relevance and easy objection. Get specific legal advice for regulated verticals.
Smaller and deeper than instinct says: 500 well-researched, tightly-matched contacts outperform 10,000 scraped ones on meetings booked — and protect your domains. Scale the research process, not the spray radius.
Directionally, with caveats — privacy proxies inflate them. Treat opens as a deliverability signal (sudden drops = infrastructure problem) and judge campaigns on replies and meetings, which can't be faked by a prefetch bot.
In-house needs someone owning infrastructure, enrichment, copy and iteration — realistically a dedicated role. Agencies amortize that expertise; Chalk Labs runs outreach automation from ~$3k/month measured on booked calls. Either way, refuse anyone who won't show you their deliverability practices first.