Cold Email Playbook for B2B SaaS

A list of 500 highly relevant prospects outperforms a list of 5,000 loosely matched contacts every single time. Cold email success is about relevance, and relevance starts with the list. Use Apollo.io to build targeted lists filtered by: industry, company size, funding stage, technology used (look for companies using Salesforce, HubSpot, or specific tools that signal your ICP), and geography. Add LinkedIn Sales Navigator for deeper filtering and to cross-reference LinkedIn profiles. For each list, define your inclusion criteria in writing before you pull contacts. If you can't write a one-sentence explanation of why this person is on your list, they shouldn't be on it. Bad list: 'VP of Marketing at tech companies.' Good list: 'VP of Marketing at B2B SaaS companies with 50-200 employees that raised Series A in the last 18 months and are currently hiring SDRs (signal they're scaling sales).' Verify email addresses before sending. A bounce rate above 5% will damage your sender reputation and get you flagged as a spammer. Use NeverBounce or ZeroBounce to verify lists — both cost about $0.01 per email verification and will cut your bounce rate by 70-80%. Also enrich your list with job titles, LinkedIn URLs, and company information to enable personalization — Clay is the best tool for this at scale.

Effective cold email is short, specific, and focused entirely on the recipient — not on your company. The template that consistently works: Line 1: A hyper-specific opening that references something real about them (recent news, job posting, LinkedIn post, company announcement). Line 2-3: The problem you solve, stated in terms of their reality, not your features. Line 4: One credibility signal (a specific client win, a data point, a reference). Line 5: A low-commitment CTA. Aim for 100-150 words maximum. Emails longer than 200 words see significantly lower reply rates. Use plain text formatting — no images, no logos, no fancy HTML. Plain text looks like a real person sent it, which is the whole point. Avoid spam trigger words: 'free', 'guarantee', 'limited time', 'act now', 'no obligation' — these trigger spam filters. For subject lines, 3-6 words, no clickbait. Best performing subject lines are either question-based ('Quick question about [company]') or reference something specific ([company] + [specific pain]). Emojis in subject lines: mixed results, test for your ICP. Personalization tokens in subject lines ({first name}, {company}) consistently outperform generic subject lines by 15-25%. A/B test every subject line — your intuition about what will work is usually wrong.

A cold email sequence isn't a single email — it's a coordinated campaign. Research shows it takes 5-8 touches before most prospects respond, but the touches need to add value, not just repeat the same message. Our standard sequence structure for B2B SaaS: Email 1 (day 1): personalized cold email, problem-focused. Email 2 (day 4): short follow-up, add one social proof point ('We helped [similar company] achieve [result]'). Email 3 (day 9): value add — a relevant piece of content, a benchmark, or a quick insight. Email 4 (day 16): slightly different angle — address a different pain point or use case. Email 5 (day 23): breakup email, very short ('I'll leave the ball in your court. Here's our [resource] if it's ever useful.'). The breakup email often has the highest reply rate in the sequence because it creates urgency and feels human. Keep it under 30 words. The common mistake is making the breakup email a passive-aggressive guilt trip — keep it friendly and leave the door open. For LinkedIn integration: add a LinkedIn connection request between email 2 and 3. If they accept, send a short LinkedIn DM reinforcing the email thread. This multi-channel approach increases overall response rates by 20-30% but requires more coordination. Tools like Lemlist or La Growth Machine can automate the LinkedIn steps alongside email steps.

True personalization — researching each prospect individually — doesn't scale beyond 20-30 emails per day. For higher volumes, you need AI-assisted personalization. Clay is the best tool for this: it pulls data from LinkedIn, company websites, news APIs, job boards, and G2 reviews, then uses AI (Claude or GPT-4) to generate personalized first lines based on that research. Here's how to set up a Clay personalization workflow: Pull prospect list from Apollo, enrich with Clay (add LinkedIn URL, recent posts, company news, job postings, funding info), create a prompt like 'Write a 1-sentence cold email opener that references [company news/job posting/LinkedIn post] and connects it to [problem your product solves]', review the AI output and approve or edit. This takes 10-15 minutes to set up and saves 3-4 hours per day of manual research. The quality of AI personalization depends entirely on the quality of your prompt. Generic prompts produce generic output. Specific prompts — with context about your ICP's pain points, your product's value prop, and examples of good openers — produce output that sounds genuinely human. Budget for the SDR to review and lightly edit every AI-generated line, not just bulk approve. The review catches embarrassing errors and adds a human touch that makes a real difference.

Cold email is a game of incremental improvements. A 1% increase in reply rate at 1,000 emails/week is 10 more replies per week — compounded over a year, that's hundreds of additional pipeline opportunities. Build a systematic testing process: one variable changed per test, minimum 200 emails per variant before judging, track positive reply rate (not just open rate or reply rate, which include OOO and negative replies). Things worth testing: subject line variations (question vs. statement, personalized vs. generic), first line approach (company news vs. job posting vs. LinkedIn reference), offer/CTA format (question vs. 'book a call' vs. 'want this resource?'), email length (100 words vs. 150 words), sending time (Tuesday 9am vs. Thursday 2pm — studies show Tuesday and Thursday mornings perform best for B2B). Things NOT worth testing until your baseline is solid: HTML vs. plain text (plain text wins almost always), sender name variations (minor impact), emoji in subject lines (marginal). Use your sequencing tool's A/B testing feature — Outreach, Salesloft, and Instantly all support sequence-level A/B testing. Document every test result in a simple spreadsheet: hypothesis, variable, result, conclusion. This becomes your institutional knowledge and prevents retesting the same things.