The First 100 Customers Playbook for B2B AI

The first 100 customers are not a marketing problem — they're a learning problem with revenue attached. Lenny Rachitsky's analysis of how top companies acquired their first 1,000 users shows that nearly every successful company used a single dominant channel in the early days, not a spray-and-pray approach. For B2B AI at seed stage, the playbook is narrower still: founder-led outbound, warm intros, and one community wedge.

Why 100 is the magic number

Ten customers prove someone will pay. One hundred customers prove a market exists. At 100, you have enough data to calculate retention curves, identify your best-fit segment, estimate CAC payback, and justify your first GTM hire. Below 100, every metric is noise. Above 100, patterns become statistically meaningful — even with modest sample sizes.

• 1–10 — validate willingness to pay and core workflow.
• 10–30 — identify ICP and refine positioning.
• 30–100 — prove a repeatable acquisition channel.
• 100+ — scale what works; kill what doesn't.

Phase 1: Customers 1–10 (manual, high-touch)

Your first 10 customers come from your network, not your marketing stack. Every founder, advisor, and investor should know exactly who you're looking for. Offer white-glove onboarding — you personally set up every account, join every kickoff call, and fix every bug within hours. These customers are co-builders, not just revenue.

1. List 50 people who fit your ICP from LinkedIn, past jobs, and investor networks.
2. Ask for intros, not sales pitches. "Who do you know at [company type] struggling with [problem]?"
3. Offer a pilot, not a demo. Pilots create commitment; demos create polite nods.
4. Charge something — even $100/month. Free users give feedback; paying users give signal.

Phase 2: Customers 10–30 (pattern recognition)

With 10 customers, analyze who converted fastest, who retained longest, and who referred others. Double down on that segment. Superhuman's approach to finding PMF — surveying users, segmenting by "very disappointed," and focusing exclusively on users who love the product — applies equally to early customer acquisition. Find the segment that converts at 3x the rate and ignore everyone else.

• Run win/loss analysis on every deal — won or lost.
• Identify your "power user" profile — title, company size, use case, acquisition source.
• Build case studies from your best 3 customers — specific outcomes, not generic praise.
• Start outbound to lookalike accounts — same industry, size, and pain as your best customers.

Phase 3: Customers 30–100 (one channel, repeated)

This is where most seed-stage teams fail — they add channels before mastering one. Pick the channel that produced customers 10–30 and systematize it. Gabriel Weinberg's Bullseye Framework (from Traction) recommends testing channels in order of likely fit, then doubling down on the one that works.

• Founder outbound — 50 targeted emails/week with personalized first lines. Track reply and meeting rates.
• Content + SEO — publish 2 deep articles/month targeting problems your ICP searches for. Slow but compounding.
• Community presence — become genuinely useful in one Slack group, Discord, or forum where your ICP gathers.
• Partnerships — integrate with or co-sell alongside a tool your ICP already uses.
• Events — speak at or sponsor one niche conference per quarter. Quality over quantity.

Outbound that works for B2B AI

Cold outbound isn't dead for B2B AI — bad outbound is. At seed stage, your edge is specificity. Generic "AI-powered solution" emails get deleted. Specific "I noticed your team posted 3 compliance analyst roles — we automate the first-pass review those hires would do" gets replies.

1. Research signal — job posts, funding announcements, regulatory changes, tech stack changes.
2. Personalized first line — reference something specific about their company, not a mail-merge field.
3. Problem, not product — lead with the pain you solve, not your feature list.
4. Clear CTA — "15-minute call to see if this applies?" not "let me know if interested."
5. Follow up twice — 80% of deals require 5+ touches; most founders stop after one.

Metrics to track from customer 1

Don't wait until customer 50 to start tracking. From day one, log:

• Source — how did each customer find you? (Intro, outbound, content, event, other.)
• Sales cycle length — days from first conversation to signed contract.
• Activation time — days from signup to first value moment.
• Retention — are they still using the product at 30, 60, 90 days?
• Referrals — have they introduced you to anyone? (The strongest PMF signal.)

What changes at customer 100

At 100 customers, you should know: your ICP with precision, your best acquisition channel with data, your sales cycle and conversion rates, and your retention curve. This is when you hire your first GTM person, invest in the channel that's working, and start saying no to segments that don't convert. The first 100 are the hardest because you're learning. Everything after is execution — if you built the playbook honestly.