seo-outreach-skill（SEO outreach 技能）

Link Building Outreach Skill

Reviews the full link building outreach pipeline: reads your opportunity list, finds author contacts, analyzes target articles, and generates personalized outreach drafts for manual sending.

Overview

Use this skill to turn outreach opportunities into a structured workflow: article analysis, contact discovery, personalized email drafting, and follow-up planning.

Best For

• SEO teams running link-building campaigns that need better reply quality
• SaaS teams trying to earn placements in listicles, comparisons, and resource pages
• agencies that need repeatable outreach without mass-template language
• operators who want contact research, relevance analysis, and draft writing in one system

Start With

Setup outreach — my product is https://yourproduct.com, my audience is SEO teams, my name is Jane, my title is Founder

Prepare outreach drafts

Prepare outreach drafts for https://example.com/best-seo-tools

External Access And Minimum Credentials

This workflow can touch external services. Use the minimum credentials needed for the stage you are running:

• GOOGLE_SHEETS_TRACKER_URL: optional source for opportunity rows; prefer read-only or exported CSV
• SERPAPI_API_KEY: recommended for article discovery and search-driven contact research
• no mail credentials are required for research and draft generation

If these are unavailable:

• accept pasted tables or CSV files instead of assuming sheet access
• stop at draft generation instead of pretending the skill can send mail
• do not claim private inbox or spreadsheet access without explicit credentials

Access Policy

Safe default: this skill should stop at research, contact discovery, and draft generation.

• spreadsheet input is optional; pasted tables and CSV exports are valid inputs
• do not assume Gmail, Workspace, inbox, or sheet access by default
• if credentials are missing, produce drafts and a send-ready checklist instead of pretending delivery happened
• if the user wants delivery, hand off to a separately approved sender workflow rather than assuming this skill can send mail directly

Core principle: The only reason outreach emails don't get replies is that authors can tell in one second it's a mass template. Every step in this pipeline exists to produce one thing: an email that could only have been written for that specific article.

When to Use

• Before or during a link building campaign
• Processing a batch of outreach opportunities from a spreadsheet
• Finding contact info for article authors
• Generating personalized outreach drafts at scale
• Single-opportunity quick outreach

Quick Start

First time: Tell Claude:

"Setup outreach — my product is [URL], my audience is [description], my name is [name], my title is [title]"

Every time after:

"Prepare outreach drafts" or "Prepare outreach drafts from this CSV / sheet export"

Setup (One-Time Only)

Ask the user for these four things if not already provided:

During Setup, Claude Will:

1. Fetch product page — Extract product name, features, USPs, tone/voice
2. Identify differentiators — What makes this product stand out vs. alternatives
3. Generate base email templates — One per article type (Best/Top, How-to, Review, Resource)
4. Confirm with user — Show a summary before proceeding

⚠️ Note: Competitor analysis requires either a manually provided competitor list or a connected SEO API. Claude will ask if none is available.

Store all setup output in memory for the session. If context is lost, re-run setup.

Outreach Pipeline (6 Steps)

Step 1 — Read Opportunity List

Source: A connected tracker if explicitly configured, or a CSV/table pasted directly.

Required columns:

Auto-detected optional columns: Article Type, Anchor Text, Notes

If a tracker is explicitly configured, filter to rows where Status = pending. Update Status to processed after each row completes. If no tracker is configured, ask the user for a pasted list or CSV export instead.

Step 2 — Detect Article Type

Fetch each target URL and classify based on title keywords and page structure:

Why this matters: The same product needs a completely different pitch depending on article type. For a Top List, the angle is "your list is missing this tool." For a How-to, the angle is "this tool removes the hardest manual step in your tutorial." Same product, entirely different email. Getting the type wrong means the entire email will feel off.

If user pre-filled Article Type, skip detection and use their value.

Step 3 — Find Contact Information

Priority principle: Finding the right person matters more than finding any email. Sending to the actual author or site owner gets 3-5x higher reply rates than sending to a generic contact address.

Never guess an email address. A guessed email that bounces damages your sending domain's reputation and flags your domain as a spam source. An honest "Not Found" is always better than an unreliable guess.

Search in this priority order. Stop as soon as a high-confidence result is found:

1. Article byline
   → Extract author full name
   → Search "author name + site domain" to find their contact

2. Website Contact / About page
   → Visit /contact, /about, /team, /advertise
   → Extract any email directly listed — this is the most reliable source

3. Author personal website
   → If byline links to author's own site, visit it
   → Check their About or Contact page

4. Twitter/X bio
   → If author has a linked Twitter profile, visit it
   → Check bio text and pinned tweet for email

5. Not found
   → Mark as Not Found
   → Recommend Twitter DM or LinkedIn message instead
   → Do NOT attempt to guess or construct an email address

⚠️ LinkedIn: LinkedIn prohibits automated data scraping. If LinkedIn is the only available source, flag the entry as confidence: Low and explicitly recommend manual DM. Do not attempt to extract data from LinkedIn profiles.

Output per entry:

email: john@example.com
source: contact page / byline search / Twitter bio / not found
confidence: High / Medium / Low / Not Found

Confidence levels and recommended actions:

Step 4 — Deeply Analyze Target Article

This step determines email quality. Generic outreach tools stop at Step 3. The difference between a reply and an ignored email is almost entirely decided here.

Fetch and read the full article content. Extract four things:

① Find a specific, quotable reference to use in the email

Goal: Make the author know in the first sentence that you genuinely read their article — not a bot, not a template.

Find one specific thing from the article: a concrete claim, a data point, a recommendation, or a position the author took. This will be the opening of the email.

❌ Without this analysis:

"I really enjoyed your article on SEO tools. It was very insightful!" *(Could apply to any article. Author ignores it.)*

✅ With this analysis:

"Read your SEO tools comparison — you pointed out that most tools have a 24-48hr backlink data lag. That's something we hear constantly from users too." *(References something specific. Author thinks: "This person actually read it.")*

② Identify the content gap — your entry point

Find what the article does NOT cover: a tool category, a use case, a type of user, or a feature area that's missing.

This gap is the reason your product belongs in the article. It's not "our product is great" — it's "your article covers X and Y but not Z, and Z is what [Your Product] does."

Example: An article reviewing 10 SEO tools covers keyword research, technical audit, and backlink analysis — but says nothing about AI Overview tracking. If your product tracks AI Overviews, that's the gap. The pitch writes itself.

③ Assess the author's writing style and audience

Determine: Is this author data-driven or experience-driven? Are they writing for beginners or experts? Is the tone formal or casual?

This directly controls the email's tone and vocabulary. A mismatch in tone — too formal for a casual blogger, too casual for a professional publication — will make even good content feel wrong.

④ Define the one-sentence relevance statement

Combine the above three into a single sentence: what your product does for THIS article's readers in THIS specific context.

This sentence is the core of the email body. Everything else supports it.

Produce this analysis card before writing the email:

article_summary: [one sentence on what the article covers and its main conclusion]
key_reference: [the specific quote, claim, or section to open the email with]
gap_identified: [what topic/tool/use case is missing that your product covers]
author_style: [data-driven / experience / beginner / expert + formal / casual]
relevance_statement: [one sentence: your product helps this author's readers do X]

Step 5 — Generate Personalized Email

Using setup data + the Step 4 analysis card, generate one email per opportunity.

Five non-negotiable rules:

Rule 1: Under 150 words — strictly enforced

Authors are busy. Every word over 150 reduces reply probability. The email must be completable in 30 seconds. If you cannot say it in 150 words, cut — don't expand.

Rule 2: First sentence must prove you read the article

Never open with:

• "Hope this finds you well"
• "I came across your article"
• "My name is X and I work at Y"
• Any sentence that could apply to any article

Open with the specific reference identified in Step 4①. Make the author's first reaction be "this person actually read my piece."

Rule 3: Reference ONE specific piece of content — never generic praise

One precise reference > five vague compliments.

❌ Generic: "Your article was really comprehensive and well-researched." ✅ Specific: "Your point about backlink data lag (24-48hrs for most tools) is something our users bring up constantly."

Rule 4: One ask only — make it impossible to misunderstand

The author must finish reading and know exactly what you want them to do.

❌ Vague: "Would love to explore some collaboration opportunities or a potential partnership." ✅ Clear: "Would it make sense to add [Product] to your list? Happy to set up a free trial account."

One action. Clear outcome. Low effort to say yes.

Rule 5: Match tone to author style

Use the author style assessment from Step 4③. Casual author = casual email. Technical author = technical email. This is not optional — tone mismatch is one of the most common reasons good pitches get ignored.

Full before/after example:

Target article: "Top 10 SEO Tools in 2025" Author style: Data-driven, professional, expert audience

❌ Email without Skill:

Subject: Partnership Opportunity

Hi,

Hope this email finds you well. I came across your article about SEO
tools and found it very helpful and insightful.

I wanted to reach out because we have a great SEO tool called [Product]
that I think would be a great fit for your article. It has many amazing
features including keyword tracking, backlink analysis, site audit,
and much more.

Would love to discuss a potential partnership or collaboration.
Please let me know if you're interested.

Best regards,
[Name]

✅ Email with Skill:

Subject: One gap in your SEO tools roundup

Hi [Name],

Read your SEO tools comparison — your point about backlink data lag
(24-48hrs for most tools) is something we hear constantly from users.

One category your roundup doesn't cover: AI Overview tracking.
It's becoming a significant organic traffic source, and [Product]
focuses on exactly this — showing which queries trigger AI Overviews
and whether your content appears in them.

Happy to set up a free trial if you're considering adding a new category.

[Name] · [Title]

Why the second email works:

Templates by article type (use as base, always customize with Step 4 analysis):

Best/Top List:

Hi [Name], read your [article title] — [specific reference from Step 4①]. One tool missing from your list: [Your Product]. [One-sentence relevance from Step 4④]. Happy to set up a free trial if you're considering an update. — [Your Name], [Title]

How-to / Tutorial:

Hi [Name], your guide on [topic] is one of the cleaner ones I've found — especially [specific section from Step 4①]. One addition that might help your readers: [Your Product] handles [the hard manual step] automatically. Worth a mention if it fits. — [Your Name], [Title]

Review / Comparison:

Hi [Name], solid breakdown of [tools compared] — [specific reference from Step 4①]. One tool that didn't make the list: [Your Product]. Key difference from [competitor they mentioned]: [differentiator from gap analysis]. Happy to share more if you're considering an update. — [Your Name], [Title]

Resource Page:

Hi [Name], found your [page title] while researching [topic]. [Your Product] might be worth adding — [one-sentence relevance]. Let me know if you'd like to take a look. — [Your Name], [Title]

Step 6 — Review, Send, and Close the Loop

This is where drafts become a fully automated pipeline. Step 6 has three parts: review, send, and loop.

Part 1: Review the report

Output all processed entries in a table:

User reviews each row. Confidence-based actions before sending:

Always review emails before sending. This Skill generates drafts — you approve. Speed is the AI's job. Quality is yours.

Part 2: Handoff for sending and reply handling

This public skill stops at reviewed drafts, contact notes, and follow-up plans.

If the user wants actual delivery or inbox handling:

• export the approved drafts and contact table
• hand off to a separately approved sender workflow
• do not assume this skill can authenticate to Gmail, poll inboxes, or run automated follow-ups

Follow-up planning rules

• first follow-up: short reminder with one-line value restatement
• second follow-up: final gentle close-the-loop note
• no aggressive automation
• positive replies should always be flagged for manual handling