sharkitect-solutions/outreach-specialist

Outreach Specialist

Claude's default cold outreach advice is generic -- "personalize your emails" and "follow up persistently" -- which produces 1-3% reply rates and deliverability problems. This skill teaches deliverability-first infrastructure decisions, diagnostic positioning over solution-selling, and micro-campaign targeting that achieves 2-5x industry reply rates. The difference is architectural, not cosmetic.

The Outreach Architecture Decision Tree

Route every outreach project through these four decisions before writing a single email.

1. What's the offer type?

| Offer Type | Outreach Style | Email Length | CTA Design | |---|---|---|---| | Free diagnostic / consultation | Problem-awareness → curiosity gap → low-friction ask | 60-100 words | Calendar link or reply-to-book | | Demo / trial | Specific outcome claim → social proof → time-bounded ask | 80-120 words | Direct demo link with fallback reply | | Direct product / service | Pain-agitate → solution hint → qualification question | 80-130 words | Reply with qualifying answer | | Content / event invite | Value preview → exclusivity signal → single CTA | 50-80 words | Registration link, no reply needed |

Key rule: Diagnostic offers outsell direct pitches 3:1 in cold outreach because they lower commitment and reframe the seller as advisor. Default to diagnostic unless the user has a specific reason not to.

2. What's the market segment?

| Segment | Decision Maker | Avg Response Window | Personalization Minimum | |---|---|---|---| | SMB (<50 employees) | Owner / GM | 24-72 hours | Industry-specific (Level 2) | | Mid-market (50-500) | VP / Director | 48-96 hours | Role-specific (Level 3) | | Enterprise (500+) | Individual contributor champion | 1-2 weeks | Account-specific (Level 4) |

Scaling rule: As company size increases, personalization depth must increase and volume must decrease. A 500-email SMB campaign can work at Level 2 personalization. A 500-email enterprise campaign at Level 2 will get 0 replies.

3. What's the volume tier?

| Volume | Emails/Campaign | Infrastructure Needed | Personalization Depth | |---|---|---|---| | Micro | <50 | Single domain, single account | Level 3-4 (role/account) | | Scaled | 100-500 | 2-3 sending accounts, dedicated domain | Level 2-3 (industry/role) | | High-volume | 500+ | Domain rotation, IP warming, multiple accounts | Level 2 minimum + dynamic variables |

Critical: High-volume only works with established infrastructure (30+ days warmed). Never launch high-volume from new domains -- deliverability will collapse within 48 hours.

4. What's the infrastructure state?

| State | Daily Limit | What to Do First | |---|---|---| | Brand new domains | 0 (not ready) | Set up authentication, begin warm-up ramp. See references/infrastructure-setup.md | | Warming (days 1-14) | 10-25/account | Small micro-campaigns only, monitor every metric daily | | Warming (days 15-30) | 25-40/account | Begin scaled campaigns, watch bounce rate closely | | Established (30+ days, clean history) | 40-60/account | Full campaign operations, monitor weekly | | Damaged reputation | 0 (pause all sends) | Diagnose cause, pause 7-14 days, restart warm-up |

Deliverability-First Rules

These are the decisions where mistakes cost weeks or months to recover from. Infrastructure before copy, always.

1. Authentication is non-negotiable. SPF + DKIM + DMARC must all pass before sending a single cold email. Missing any one cuts inbox placement by 30-50%. See references/infrastructure-setup.md for configuration.
2. Warm-up cannot be skipped or rushed. The 30-day ramp exists because mailbox providers build sender reputation over time. Shortcuts trigger spam filters that take longer to clear than the warm-up itself.
3. 50 emails/day/account is the ceiling for Google Workspace. At 51+, Google's rate limiting activates unpredictably. Microsoft 365 allows ~80/day but flags patterns faster.
4. Bounce rate above 4% is an emergency. Stop all sends immediately. Audit the list. Every bounce degrades domain reputation and the damage compounds -- a 6% bounce day can take 2 weeks to recover from.
5. Spam complaint threshold is 0.1%. One complaint per 1,000 emails. Above this, inbox placement drops across ALL campaigns on that domain, not just the offending one.
6. Plain text outperforms HTML in cold email. No images, no HTML formatting, no multiple links. One link maximum. Tracking pixels are acceptable but reduce deliverability by ~5%.
7. Never send cold email from your primary business domain. Use a dedicated outreach domain (e.g., company-reach.com instead of company.com). If the outreach domain gets burned, your business email is unaffected.

Targeting Precision Framework

Why micro-campaigns beat blast campaigns: a 50-contact campaign with Level 3 personalization produces 2.76x the reply rate of a 500-contact campaign with Level 1 personalization. The math always favors precision.

Personalization Depth Levels

| Level | What's Personalized | Time per Email | Expected Reply Lift | |---|---|---|---| | Level 1: Spray | Name + company only | 0 (automated) | Baseline (2-3%) | | Level 2: Industry | Industry-specific pain points, seasonal context | 30 sec | 1.5-2x baseline | | Level 3: Role | Role-specific language, title-aware framing, company-size context | 1-2 min | 2.5-3.5x baseline | | Level 4: Account | Company-specific research, recent news/events, mutual connections | 5-10 min | 4-6x baseline |

Decision rule: Spend personalization time where deal size justifies it. $500 ACV = Level 2 max. $5K ACV = Level 3. $50K+ ACV = Level 4.

ICP-to-Campaign Flow

1. Define ICP: industry + company size + role + geography + qualifying signal
2. Build niche list: one industry-role combination per micro-campaign (never mix)
3. Score leads: verified email + decision-maker confirmed + enrichment data available
4. Disqualify ruthlessly: no generic emails (info@, contact@), no unverifiable addresses, no companies outside ICP
5. Cap campaign size: 25-50 for micro, 100-200 for scaled, split into batches for high-volume

Email Copy Architecture

These are structural rules, not templates. Templates are in references/campaign-templates.md.

The 4 rules that separate 3% reply-rate emails from 15% reply-rate emails:

1. 60-120 words maximum. Every word above 120 reduces reply probability. Cold emails are not the place to educate -- they earn the right to a conversation.
2. 3rd-5th grade reading level. Short sentences. Common words. No jargon unless the recipient uses that exact jargon daily. Tools: Hemingway App or readability-score.com.
3. One CTA, stated once. "Would a 15-minute call this week make sense?" not "Book a demo OR reply OR visit our site OR download our whitepaper." Multiple CTAs produce zero action.
4. Subject lines that look internal. 2-4 words that could be an internal memo subject: "quick question", "scheduling idea", "[first name] - timing". Never salesy subjects ("Exclusive offer!", "Don't miss out!").

Follow-up architecture: See references/campaign-templates.md for full sequence frameworks. The key principle: each follow-up must introduce a genuinely new angle (proof, data, different pain point). Never reference the previous email ("Just following up on my last email...").

Anti-Pattern Decision Table

| Anti-Pattern | Why It Fails | What To Do Instead | |---|---|---| | Sending 200+ emails from a 1-week-old domain | Triggers spam filters immediately, domain may be permanently flagged | Follow 30-day warm-up ramp, start at 10-20/day | | "Just checking in" follow-ups | Adds zero value, trained as spam signal by filters and humans | Each follow-up introduces new proof, angle, or value | | HTML-heavy emails with images and buttons | Cold email filters penalize formatting complexity; looks like marketing blast | Plain text, one link max, conversational tone | | Mixing industries in one campaign | Prevents meaningful A/B testing; generic copy underperforms niche copy | One industry-role combination per micro-campaign | | Buying pre-built email lists | 20-40% invalid rate typical; destroys bounce rate and domain reputation | Build lists from verified sources, validate every address before send | | Solution-selling in the first email | Prospect hasn't confirmed the problem exists; feels presumptuous | Lead with problem awareness, offer diagnosis | | Sending same email copy for 30+ days | Mailbox providers detect repeated content patterns and flag as spam | Rotate copy every 2 weeks, maintain 3-4 active variants | | Skipping email verification | Even "good" lists have 5-15% invalid addresses that cause bounces | 3-tier verification: MX check, API verification, catch-all detection | | Including unsubscribe link in cold email | Signals bulk sending to spam filters; cold email is not marketing email | Use breakup email as natural exit; honor opt-out replies immediately | | Personalizing with obviously scraped data | "I saw your LinkedIn post from 2019..." feels surveillance, not relevance | Use timely signals: recent hires, funding, seasonal patterns |

Rationalization Table

| Rationalization | When It Appears | Why It's Wrong | |---|---|---| | "We need more volume to see results" | Low reply rates after 2-3 weeks | Volume amplifies bad targeting. Fix targeting/copy first, then scale. | | "Our emails are too short, we need to explain more" | Prospects not engaging | Length is not the problem. Relevance and positioning are. Longer cold emails perform worse. | | "Let's add more links so they have options" | Low click-through rate | Multiple CTAs create decision paralysis. One clear ask outperforms every time. | | "We should warm up faster, we're losing time" | Impatience during ramp period | Rushing warm-up risks the domain permanently. A 2-week shortcut can cost 2-month recovery. | | "The list is fine, our copy must be bad" | High bounce rate (>4%) | List quality causes bounces, not copy. Verify the data before rewriting anything. | | "Let's send from our main domain to look legitimate" | Low open rates from outreach domain | Burning your main domain destroys ALL business email. The outreach domain exists to absorb risk. | | "We should follow up more aggressively" | No replies after sequence completes | Aggressive follow-up triggers spam complaints. Test new angles, not more frequency. | | "Personalization takes too long at scale" | Scaling pressure | Then reduce volume to match personalization capacity. Unpersonalized scale produces negative ROI. |

NEVER List

1. NEVER send cold email without SPF + DKIM + DMARC all passing. No exceptions, no "we'll set it up later."
2. NEVER exceed 50 emails/day per Google Workspace account or 80 per Microsoft 365 account.
3. NEVER launch campaigns from a domain less than 14 days into warm-up. Micro-campaigns only during warm-up.
4. NEVER send to unverified email addresses. Every address must pass verification before entering the send queue.
5. NEVER include more than one external link in a cold email. Each additional link increases spam score.
6. NEVER reference previous emails in follow-ups ("per my last email", "just circling back"). Each email stands alone.
7. NEVER use the prospect's first name more than once in an email. Repeated name insertion is an obvious automation tell.
8. NEVER send identical copy to more than 50 recipients. Mailbox providers detect duplicate content and flag it.
9. NEVER continue sending when bounce rate exceeds 4%. Stop immediately, audit, fix, then resume.
10. NEVER pitch a solution before diagnosing the problem. The first email earns a conversation, not a sale.

Performance Diagnostic Framework

When campaign metrics are below benchmarks (see references/outreach-benchmarks.md), diagnose in this exact order. Each layer must be healthy before the next matters.

Layer 1: DELIVERABILITY (are emails reaching inboxes?)
  Check: inbox placement rate, bounce rate, spam folder rate
  Fix: authentication, warm-up status, sending volume, list quality
  |
Layer 2: TARGETING (are the right people receiving emails?)
  Check: ICP match rate, title accuracy, company fit
  Fix: list source, enrichment quality, qualification criteria
  |
Layer 3: COPY (is the message compelling?)
  Check: open rate (subject line), reply rate (body + CTA)
  Fix: subject line testing, body length, CTA clarity, reading level
  |
Layer 4: TIMING (is the send time optimal?)
  Check: open rate by day/time, industry work patterns
  Fix: send window adjustment, timezone targeting, seasonal factors

Critical rule: If Layer 1 is broken, do NOT touch Layers 2-4. Rewriting copy when emails land in spam is wasted effort. Fix infrastructure first, measure again, then move up the stack.