Outbound

Outbound is the most misunderstood B2B acquisition channel. The default version (buy a list, write a generic intro, blast it through Apollo, refine the subject line) has reply rates under 1% and burns sender reputation faster than it produces meetings. That version is what gives outbound a bad name.

The version that works runs on the same thesis as any high-conversion channel: the right message reaches the right person at the right time. Each of those three is a lever. Get one wrong and the math collapses. Get all three right and outbound becomes the cheapest qualified-meeting channel a B2B company has: significantly cheaper than paid, and faster than content.

This guide is about the channel. For the operational pipeline that drafts your outreach end-to-end, install the outbound-sales skill; that’s the agent that does the job.

The thesis: three levers, multiplicative

Most “outbound improvement” advice picks one lever and optimizes it. Better subject lines (message). Better list segmentation (person). Better sequencing (time). The actual lift comes from working all three together, because their effects multiply:

The arithmetic, conservatively: 5× × 2× × 3× = 30× reply rate vs. spray-and-pray. That’s the difference between 0.5% (delete) and 15% (qualified pipeline). Same channel, same volume, same product. Different system.

The three levers map to three concrete inputs:

• Right time = a signal is firing on the domain now. Signals are the timing trigger. If no signal fires, the account isn’t urgent, so deprioritize it.
• Right person = the title to target depends on which signal fired. Hiring SDRs → VP Sales. Negative support reviews → VP CX. Social spike → VP Marketing. Each signal has its own ICP persona.
• Right message = the signal that fired also tells you the angle. “Hiring 3 SDRs” → talk about SDR ramp. “Negative Trustpilot support reviews” → talk about ticket resolution. The hook is never generic; it’s always the specific reason you’re reaching out today.

How signals work as the channel’s qualification gate

Signal detection answers two questions at once:

1. Is this account worth reaching out to right now? (timing)
2. What specifically do I lead with? (messaging)

The signals our API surfaces, mapped to the implied pain:

The full mapping (with title filters per signal and the hook structure) is bundled with the outbound-sales skill in references/signal-to-angle.md. Run signals first, before any people-lookup spend. It’s the single biggest token-savings lever, and the single biggest reply-rate lever.

The pipeline (channel view)

target accounts
   ↓
[detect_signal]            ← Stage 1: qualification gate (free dispatcher)
   ↓ keep only firing-signal accounts
[map signal → angle]       ← Stage 2: which signal fired = which message + which title
   ↓
[find the right person]    ← Stage 3: LinkedIn lookup with title filters
   ↓
[verify the email]         ← Stage 4: SMTP-verified, catch-all flagged
   ↓
[draft email + DM]         ← Stage 5: multi-channel, both keyed to signal
   ↓
[handoff to your platform] ← Stage 6: structured drafts → Apollo / Outreach / Salesloft / Lemlist / Smartlead

Two design choices in this pipeline matter at the channel level:

1. The skill drafts; your platform sends. B2B teams already pay for an outreach platform (Apollo, Outreach, Salesloft, Lemlist, Smartlead, Instantly, internal CRM). Replacing that surface is hostile to existing workflows. So outbound-sales returns structured drafts (email_subject, email_body, linkedin_dm, signals_fired, primary_signal, angle, ready_to_send: true) per lead, ready to paste into whatever you already use. The channel responsibility splits cleanly: research + drafting in outbound-sales; sequencing + send + tracking in your platform.

2. Multi-channel from the same angle. Email and LinkedIn DM both go out keyed to the same signal, but with different registers (email ≤90 words, formal; DM ≤60 words, conversational). Sending only one channel under-uses the signal; sending the same copy on both is a tell that gets flagged. The skill produces a matched pair per lead.

Channel economics

A single qualified prospect end-to-end:

At a reply rate of 10–15% on signal-keyed outbound (vs. <1% generic), that’s $5–$7.50 per reply. Add your outreach platform’s per-seat license and your SDR’s draft-review time, and a qualified meeting lands in the $40–$80 range. That’s a fraction of the paid-meeting cost for any B2B SaaS at any reasonable ACV, which is why outbound stays competitive against paid even as paid CACs rise.

The same arithmetic kills outbound when you skip the signals gate. Spray-and-pray at 0.5% reply rate produces meetings at $150–$300 even before you factor in deliverability damage. The signals gate is what makes the channel work, not just an efficiency tweak.

Where outbound fits in the channel mix

Outbound complements rather than replaces other acquisition channels:

The teams getting the best blended CAC run outbound and content in parallel: content/SEO builds the long-tail brand surface so that when outbound lands in an inbox, the recipient half-recognizes the name. Each channel makes the other one work harder.

Channel hygiene (the boring rules that matter)

The teams that kill their outbound usually do so at the infrastructure layer, not the messaging layer:

1. Deliverability is upstream of every other lever. A perfect message that lands in spam is worth zero. Warm domains for 4–6 weeks before scaling, use dedicated sending domains for outbound (get.<yourdomain> or similar), never the primary corporate domain. Monitor inbox placement weekly via a tool like Glock Apps or MailReach. None of this happens inside outbound-sales (drafting only); it happens on your sending platform.
2. List hygiene > list size. A 200-account ICP list signal-qualified is worth more than a 20,000-account list sprayed. outbound-sales enforces this by refusing to draft for accounts with no signals firing (SKIP: no signals). Resist the urge to override.
3. Multi-channel cadence, not multi-touch on one channel. Email touch 1, LinkedIn DM touch 2 (day +3), Email touch 3 (day +7, breakup). Don’t blast 5 emails in 10 days: recipients pattern-match on volume and report you.
4. Reply-handling is the conversion step. A 15% reply rate at 0% reply-handling produces zero meetings. Have a human (or a sequence-aware agent) read replies same-day, classify intent, and route warm replies to a real conversation immediately. outbound-sales doesn’t handle replies; your outreach platform’s classifier or a separate reply-handling skill does.

The agent that does this

Everything operational lives in the outbound-sales skill:

• 6-stage pipeline implementation (signal detection → handoff)
• Signal-to-angle mapping (every gtm-tools signal → hook + pain + title filter + persona)
• 5 cold-email templates (≤90 words, signal-keyed, anti-pattern callouts)
• 5 LinkedIn-DM templates (≤60 words, different register from email)
• Skip-when-no-signals discipline baked in
• Structured handoff output for any outreach platform

Install:

$
npx skills add arnaudjnn/gtm-skills --skill outbound-sales

The skill is the operator; this guide is the channel context.

Next steps