The B2B Prospecting Playbook for Claude Code Users
TL;DR
A step-by-step playbook for B2B outbound prospecting using Claude Code and skill collections. Covers the full workflow from ICP definition through personalized outreach, with specific skill recommendations at each stage.
Why Claude Code Is Now the Best Prospecting Environment
Outbound prospecting has always been a workflow problem as much as a skills problem. The challenge isn't that sales reps don't know how to write a good cold email or research a company effectively — it's that the mechanics of doing it consistently, at scale, across dozens of prospects per day, create enormous cognitive overhead. Looking up firmographics in one tab, checking LinkedIn in another, cross-referencing funding history in a third, opening your email template, personalizing it for this specific prospect, logging it in your CRM — the actual research and writing quality is often good, but it's buried under a sea of manual switching and copy-paste that limits how many truly personalized outreach messages a rep can produce per day.
Claude Code changes this by giving you a single environment where the full prospecting workflow can run end-to-end. Skill files encode your methodology — your ICP criteria, your research framework, your personalization approach, your email structure — so you don't have to re-explain the context each session. Data integrations like Vibe Prospecting give Claude access to current company and contact information without tab-switching. File-based workflows let Claude process a list of prospects systematically, producing research briefs and email drafts you review and send, rather than requiring you to initiate each step manually. The result is a prospecting workflow that's significantly faster, more consistent, and more scalable than the multi-tab manual alternative.
The file-based architecture of Claude Code is specifically well-suited for prospecting workflows. You can store your target account lists as CSV files, your ICP documentation as Markdown, your sequence templates as text files — and Claude can read all of them as part of a session, reference the relevant pieces at each step of the workflow, and write output back to files you can act on. This is qualitatively different from a browser-based chat where every session starts from scratch and context evaporates between conversations.
Skill composability is the other key advantage. A sophisticated prospecting workflow has multiple distinct stages, each requiring different expertise: ICP qualification, prospect research, personalization, copywriting, sequence design, and performance analysis. In Claude Code, you can have a separate skill file optimized for each stage and compose them in sequence within a single session. The ICP skill defines the criteria; the research skill applies them to a specific company; the personalization skill identifies the best hook; the copywriting skill drafts the email. Each skill does one thing well, and Claude coordinates them intelligently within the workflow you define.
Step 1: Define Your ICP with a Claude Skill
The foundation of any effective outbound motion is a precise, documented Ideal Customer Profile. In practice, most sales teams have an ICP that lives in the head of the VP of Sales or that's vaguely described in a one-pager that nobody reads. The ICP is referenced inconsistently, applied differently by different reps, and rarely updated when market feedback suggests adjustments. The result is a prospecting motion that treats 'kind of fits' as the same as 'strong fit' and produces inconsistent results.
A Claude ICP skill solves this by making the ICP definition precise, persistent, and consistently applied. Start by opening a new Markdown file — call it icp-definition.md — and working through your ICP definition systematically with Claude's help. A good ICP skill will prompt Claude to walk through the ICP dimensions that matter most for your product: firmographics (company size, industry, geography, growth stage, funding status), technographics (tech stack requirements, tools they're likely using, tools that indicate a fit), buying signals (events that suggest this company might be in-market now), and negative signals (characteristics that suggest a company is unlikely to buy regardless of apparent fit).
For each dimension, the skill should push for specificity. 'Series B through Series D SaaS companies with 50-500 employees in North America' is a useful ICP definition; 'growing companies' is not. The effort to make the ICP definition specific and quantified pays back in every subsequent stage of the prospecting workflow — Claude can apply specific criteria consistently across hundreds of accounts in a way that a human rep might not.
Once your ICP definition is documented in the skill file, test it by giving Claude 5-10 companies and asking it to score each against the ICP criteria. Review the scores and adjust the criteria where Claude's scoring doesn't match your judgment. The goal is a set of criteria specific enough that Claude's scoring is consistent with how your best reps would evaluate the same companies. After calibration, this skill becomes the authoritative ICP definition that all subsequent prospecting work references — whether you're scoring an individual company during a research session or batch-scoring a list of 200 accounts.
Step 2: Build and Enrich Your Target List
With a documented ICP in your skill files, the next step is identifying which specific companies in the market fit that profile — and this is where data access becomes critical. Claude's training knowledge can provide general guidance on what kinds of companies fit your ICP, but it can't reliably tell you which specific companies currently have the right tech stack, headcount, and funding status to be active prospects today. For list building with current, accurate data, you need a data integration.
Vibe Prospecting's Claude integration is the most direct way to do this within a Claude Code workflow. With the Vibe Prospecting data integration configured, you can ask Claude to search for companies matching your ICP criteria and produce a ranked list of matches with relevant scores and signals. The search can be as specific as your ICP: 'Find Series B-D SaaS companies with 50-200 employees headquartered in North America that are currently using Salesforce and have posted at least 5 sales-related job openings in the past 60 days.' Claude queries the 150M+ business profile database and returns a list of companies that match, ranked by fit score against your ICP criteria.
For teams using other data tools — Apollo, ZoomInfo, LinkedIn Sales Navigator — the workflow is slightly less automated but follows the same logic. Export a list from your preferred data tool using your ICP filter criteria, then bring that list into a Claude Code session as a CSV or text file. Claude can read the list, apply your ICP scoring skill to each company in it, and produce a prioritized version with scores and notes — turning a raw data export into a qualified, prioritized prospect list. The manual export-and-import step adds 10-15 minutes, but the ICP scoring and prioritization Claude applies to the list makes it far more actionable than the raw export.
List building is also an opportunity to surface companies you might not have considered. Claude can help you identify adjacent market segments that share key characteristics with your core ICP, suggest alternative firmographic filters that might surface underserved market opportunities, or flag companies on your existing customer list whose profile most closely resembles your best customers — a useful starting point for identifying look-alike accounts in markets you haven't penetrated.
Step 3: Research Each Prospect with Structured Skills
Once you have a prioritized list of target companies, the research stage is where the skill-based workflow delivers the clearest efficiency gains over manual prospecting. The research task is inherently structured — there are specific things you need to know about each prospect to write effective outreach, and good reps find the same categories of information every time. That structure makes it ideal for encoding in a skill file.
A well-designed prospect research skill structures Claude's information gathering across the key dimensions that drive personalization and relevance. For each target company, the skill should direct Claude to gather: company background (size, industry, founding year, funding status, growth trajectory), recent developments (news, product launches, leadership changes, awards, acquisitions), growth signals (hiring patterns, job posting analysis, expansion into new markets), technology stack (key tools, platforms, integrations — especially those relevant to your product category), leadership profile (the target buyer's background, LinkedIn activity, previous companies), and ICP fit assessment (scoring the company against your defined criteria and flagging the top personalization angles).
With Vibe Prospecting, much of this information is available from the data integration in real time. Without it, Claude can research publicly available information through a web search MCP plugin or work from information you provide manually. Either way, the research skill's value is in structuring the output: instead of raw notes from a web research session, the skill produces a standardized prospect brief in a consistent format that directly feeds into the next stage of the workflow.
The structured output format is what enables scale. If every research brief follows the same format — company summary, top 3 personalization hooks ranked by relevance, ICP fit score with rationale, recommended contact name and role — you can review 20 prospect briefs in 20 minutes rather than reading through 20 pages of freeform notes. The standardization also makes it easy to compare prospects against each other and prioritize your outreach based on a consistent evaluation, rather than relying on which prospect happened to have the most compelling surface-level details.
Step 4: Generate Personalized Outreach at Scale
Personalization is the stage where the investment in structured research pays off. The research brief produced in Step 3 contains exactly the information a personalization skill needs: specific facts about the company, ranked hooks, the target contact's background, and the ICP fit assessment. A good personalization skill takes that input and produces specific, relevant opening lines and hooks that make the outreach feel genuinely researched rather than generated.
The key insight about AI-generated personalization is that the quality of personalization is proportional to the quality of research input. A personalization skill given a rich, specific research brief will produce far better hooks than the same skill given a bare-bones company name and industry. This is why the research step precedes personalization in the workflow rather than combining them — doing the research well, in a structured way, before writing the email produces significantly better outreach than trying to research and write simultaneously.
A well-designed personalization skill distinguishes between different types of hooks and matches them to the appropriate email context. Timing-based hooks (company just closed a funding round, exec just joined) work well as email openers because they're current and specific. Problem-based hooks (their tech stack suggests a common pain point) work better as the bridge between the opener and your value proposition. Role-specific hooks (the contact's background at their previous company) work well in the P.S. or as a LinkedIn personalization angle. The skill encodes these distinctions so Claude produces hooks with the right placement rather than a random collection of specific details.
Personalization at scale — writing 20 or 50 personalized emails in a single Claude session — works best when you batch the research and personalization stages together. Instead of doing the full research-personalization-copywriting sequence for one prospect at a time, load your prospect list, run the research skill across all prospects to produce a batch of research briefs, then run the personalization skill across all briefs to produce a batch of personalized hook recommendations. This batch approach is significantly faster than the one-at-a-time sequence and produces more consistent quality because Claude maintains context across the batch.
Step 5: Design Multi-Touch Sequences with Claude
A cold email is rarely sufficient for a response — modern B2B outbound requires a multi-touch sequence that maintains relevance across 3-8 touchpoints over 2-4 weeks. Sequence design is a skill that most SDRs underinvest in: they write a good first email, then use generic follow-ups that lose the specificity of the original personalization. A Claude sequence design skill helps maintain relevance and escalate commitment logic across the full sequence.
A good sequence skill encodes the structure of an effective multi-touch sequence: the logic of how each touch relates to the previous one, how specificity and urgency escalate across the sequence, when to change channels (email to LinkedIn to phone), and when to exit the sequence for non-responsive prospects. The skill should distinguish between sequences designed for different prospect types — high-priority accounts that warrant more touches and more research investment, mid-priority accounts that get a shorter, more templated sequence, and low-priority accounts that get a single email and then move to a broad nurture track.
The LinkedIn dimension of sequence design is often underspecified in outbound skill collections. LinkedIn connection requests, InMail messages, and post-engagement comments require a different tone and approach than email — they work better when they feel conversational rather than sales-oriented, and they're more effective when they reference shared context (a mutual connection, a post the prospect wrote) than when they lead with a pitch. A complete sequence design skill encodes the LinkedIn touchpoints as distinct from the email touchpoints, with platform-specific guidance for each.
For enterprise outbound where sequences may run over 4-6 weeks and involve multiple contacts at the same account, sequence design becomes a multi-threaded problem. The skill should help Claude design sequences that coordinate across contacts — ensuring the champion, the economic buyer, and the technical evaluator are each receiving relevant, coordinated outreach rather than independently personalized messages that may contradict each other. This multi-threaded coordination is complex to manage manually and is one of the clearest areas where a well-designed Claude skill provides more structure than a rep working without AI assistance.
Step 6: Measure, Learn, and Improve Your Skills
The best outbound teams treat their prospecting playbook as a living document rather than a fixed set of rules — they measure what's working, identify the patterns behind success, and update their approach based on what they learn. Claude skills make this continuous improvement loop faster and more systematic than maintaining a playbook in Google Docs.
The measurement workflow starts with tracking outputs alongside inputs. When you send a personalized email using a specific hook type from your research skill, note in a simple tracking file what hook you used and whether it generated a reply. Over 4-6 weeks of tracking, you'll have enough data to identify which hook types (timing-based, problem-based, social proof-based, role-specific) correlate with the highest reply rates in your specific market. Claude can analyze this tracking data and summarize the patterns — 'timing-based hooks for companies that recently completed a funding round show a 2.3x reply rate compared to your baseline' — which directly informs which research signals to prioritize in your prospecting skill.
Reply content analysis is another high-leverage measurement workflow. When you do get replies — including negative replies and out-of-office bounces — there's information in the response that can improve your skill files. A prospect who replies 'not the right time, try me in Q3' signals a timing mismatch you can encode in your ICP qualification criteria. A prospect who replies with a specific objection about budget reveals a pain point framing that isn't landing — and Claude can help you update your value proposition framing in the skill file accordingly. This systematic extraction of learning from outbound performance is something most teams do informally in their heads; with Claude skills, you can make it explicit and systematic.
Skill file versioning is what makes iteration safe. Because your skill files are in git, every change you make to your ICP definition, research framework, or copywriting approach is tracked. If you update your personalization skill based on performance data and the next batch of outreach performs worse, you can compare the current version to the previous version and identify what changed. This kind of A/B testing at the skill level — running two versions of a skill file and comparing performance — is the most rigorous way to improve your outbound playbook over time, and it's only possible because skill files are version-controlled artifacts rather than institutional knowledge that lives in people's heads.
Frequently Asked Questions
Related Articles
How to Use Vibe Prospecting with Claude Skills for B2B Outbound
Vibe Prospecting integrates directly with Claude Code, giving your AI agent access to 150M+ business profiles and 800M+ professional contacts. Combined with Claude skill files, it creates an end-to-end outbound workflow.
Top 10 Claude Code Skills for Outbound Sales Teams in 2026
The 10 best Claude Code skill collections for outbound sales teams in 2026, ranked by framework depth, ICP specificity, and real-world performance in prospecting workflows.
How to Install Claude Skills: Step-by-Step Guide (Claude Code & Cursor)
Installing Claude skills takes less than 5 minutes. You can clone a GitHub repo, use the Claude Code marketplace, or manually copy Markdown files into your project's skills directory.
Related Skill Collections
ColdIQ GTM Skills
7 master skills with 52 sub-skills and 33 standalone skills for outbound sales. 137 triggers, 34 email templates, and Clay enrichment workflows.
GTM Flywheel
15 battle-tested GTM skills from a $7M ARR agency managing 70+ campaigns. Cold email, ICP research, signal scoring, and sales intelligence.
Supercharge your GTM skills with B2B Prospecting
Power your GTM skills with premium business data. Enterprise-grade prospecting, right where you already work. Access 150M+ businesses and 800M+ professional profiles.
Get Started Free