Free playbook · 2026

More leads won’t fix your pipeline. A better system will.

The practical guide to making Claude the operating system of your B2B prospecting. 10 chapters, copy-ready prompts, one checklist.

For founders, sales teams and growth professionals.

The problem

Volume stopped working

1

Inboxes are saturated

Everyone sends 100 “personalised” templates a day. Prospects have learned to ignore them in two seconds.

2

Timing beats volume

One message sent when the need appears is worth a hundred sent at random. The question is no longer “who” but “when”.

3

Ten tools, no system

Lists here, enrichment there, sequences somewhere else. AI only helps once it becomes the interface that connects the whole workflow.

Chapter 01

Define an ICP that actually works

Why it matters

A vague ICP produces vague lists. “Tech SMBs in Europe” is not an ICP, it’s a directory.

The method

  • Start from your 10 best customers: size, industry, buying trigger, who signed.
  • Look for what they had in common right before buying, not just their profile.
  • Also write an anti-ICP: accounts that look like customers but never sign.
Claude prompt

Here are my 10 best customers with, for each, industry, size, the signer’s role and what triggered the purchase: [paste]. Find the common patterns, propose an ICP in 5 criteria that can be checked on LinkedIn, then an anti-ICP in 3 criteria. Flag any criterion too vague to use.

Chapter 02

Understand your personas

Why it matters

The CFO and the Head of Sales at the same account don’t buy for the same reasons. One message for both speaks to neither.

The method

  • For each persona: this quarter’s goals, what makes them look bad, the words they use.
  • List typical objections before the first contact.
  • Keep a short sheet per persona, reusable in every prompt.
Claude prompt

I sell [product] to [persona] at [size/industry] companies. Describe their 3 current priorities, what would make them look bad internally, the words they use to describe their problem, and their 5 likely objections to a solution like mine. Format: one-page sheet.

Chapter 03

Build and prioritise your lists

Why it matters

A list of 2,000 unsorted accounts is 2,000 postponed decisions.

The method

  • Sort into three tiers: work this week, this month, and watch.
  • Score each account on ICP fit and on the presence of a recent signal.
  • Only work tier A by hand. The rest waits for a signal.
Claude prompt

Here is a list of accounts with industry, headcount and latest known information: [paste]. Score each from 1 to 5 on fit with this ICP: [paste ICP]. Sort them into tiers A / B / C and justify each tier-A account in one line.

Chapter 04

Detect buying signals

Why it matters

This is the chapter that changes everything. The right prospect at the wrong time is still the wrong prospect.

The method

  • Strong signals: job change, funding round, hiring in your domain, new subsidiary.
  • Intent signals: engagement on a post about your problem, profile visit, comment.
  • Pair each signal with a specific, pre-written approach angle.
Claude prompt

Here are recent signals detected on these prospects: [paste]. For each, rate intent strength (low / medium / high), explain why this signal makes the timing right, and propose a one-sentence approach angle that references the signal without feeling intrusive.

Chapter 05

Personalise without sounding generic

Why it matters

Swapping {first name} and {company} isn’t personalisation. Prospects spot it instantly.

The method

  • A good opener proves you looked: a signal, a post, a specific context.
  • One message, one idea. No feature lists.
  • End with an easy question, not a request for 30 minutes.
Claude prompt

Write a first LinkedIn message under 300 characters for [first name, role, company]. Context: [signal or post]. My offer: [one sentence]. Rules: no flattery, no lists, direct and human tone, a single open question at the end. Give 3 variants with different angles.

Chapter 06

Qualify faster

Why it matters

A meeting with an unqualified prospect costs your team an hour and returns nothing.

The method

  • Pick a simple framework (BANT, light MEDDIC) and stick to it.
  • Qualify in the written conversation, before the meeting.
  • Write down what’s missing instead of guessing it.
Claude prompt

Here is the LinkedIn thread with this prospect: [paste]. Assess it with BANT. For each criterion, say what is confirmed, what is assumed and what is unknown. Conclude with: book the meeting, ask one more question (which one?), or put on hold.

Chapter 07

Prepare the discovery call

Why it matters

The first 5 minutes decide what follows. Showing up without a hypothesis lets the prospect drive.

The method

  • Prepare 3 hypotheses about their problem, to confirm or reject.
  • Write your most important questions in advance, in order.
  • Decide the next step you’ll propose at the end of the call.
Claude prompt

I’m meeting [first name, role] from [company] tomorrow. What I know: [context, thread, signal]. Prepare a brief: 3 hypotheses about their problem, 8 discovery questions ranked by priority, the 2 main risks to the deal, and the next step to propose at the end.

Chapter 08

Handle objections

Why it matters

“Not now”, “too expensive”, “we already have a tool”. Objections always come back. Improvised answers rarely land.

The method

  • List your 10 most frequent objections and what they really hide.
  • For each: a clarifying question before any answer.
  • Lean on concrete proof, not on an argument.
Claude prompt

A prospect replied: “[objection]”. Context: [persona, thread]. Explain what this objection may hide (3 hypotheses), propose a clarifying question, then a short, natural answer for each hypothesis. Avoid any defensive tone.

Chapter 09

Follow up without pestering

Why it matters

Most replies come after the second or third follow-up. Most follow-ups are “just bumping this”.

The method

  • Every follow-up adds something: a resource, a customer case, a new angle.
  • Space them out and stop after three without a reply.
  • Adapt to what the prospect did in the meantime (profile view, reaction).
Claude prompt

My first message to [prospect]: [paste]. No reply for [delay]. Write a sequence of 3 spaced follow-ups, each adding something different (a resource, a concrete case, a more direct question). The last one should close the loop politely. Under 250 characters each.

Chapter 10

Keep the CRM clean and optimise the pipeline

Why it matters

An outdated CRM lies. And a pipeline that lies leads to bad decisions.

The method

  • After every exchange: status, dated next step, what you learned.
  • Every week: spot deals with no activity for 14 days.
  • Every month: compare reply and conversion rates by signal and persona.
Claude prompt

Here is my pipeline export (stage, amount, last activity, next step): [paste]. Spot at-risk deals, those without a dated next step, and the stages where deals stall. Propose 3 priority actions for the week.

The full system

From signal to meeting

Each chapter is a step. Put together, they form a workflow Claude can run.

  1. 01

    Signal

    An ICP prospect shows intent.

  2. 02

    Score

    The account is scored and prioritised.

  3. 03

    Message

    A personalised approach goes out.

  4. 04

    Follow-up

    The sequence adapts to replies.

  5. 05

    Meeting

    The qualified lead lands in the calendar.

Checklist

Before you send a single message

The best sales professionals won’t be replaced by AI. They’ll be amplified by well-designed systems.

Want this system without doing it all by hand?

Yauda detects buying signals on LinkedIn, scores your prospects, personalises messages and runs follow-ups. And plugs into Claude via MCP.