Lead sourcing
Pull raw company lists from databases (Apollo, Clay, LinkedIn scrapes) filtered by industry, headcount, geography. Volume is cheap here; relevance is not.
The six stages of an automated outbound pipeline
Enrichment
Attach emails, tech stack, hiring signals, funding events. Typically 60–75% of raw leads get a valid, verified email. Bounced emails burn your sending domain — verification is non-negotiable.
ICP fit scoring
Score each lead against your Ideal Customer Profile: right size, right industry, an actual trigger (e.g. "just posted a job for the thing you automate"). A strict filter that keeps 30–40% beats spraying everyone.
Personalized outreach
An LLM drafts a first line referencing the lead's specific trigger. Good personalized cold email gets 2–8% reply rates; generic blasts get under 1% and get you spam-flagged.
Follow-up sequence
3–5 touches over 2–3 weeks. Roughly half of all replies come from follow-ups, not the first email. Automation shines here: nobody forgets touch #4.
Booked calls
Of replies, only some are positive (~30–40%), and of those, most but not all show up to book (~60–80%). The output metric that matters: cost per booked meeting.
Funnel calculator: run your own numbers
Set the volume and per-stage conversion rates. The 3D funnel narrows to match, and the pipeline math updates live. Drag the funnel to orbit it.
Drag to orbit · each disc's width = leads surviving that stage (log scale)
Stage-by-stage survivors
Pipeline output
Meetings / month—
Cost per meeting—
Overall conversion—
Emails sent (≈4 touches)—
…
The math is just chained multiplication
meetings = leads × enrich% × qualify% × reply% × positive% × book%
cost per meeting = monthly stack cost ÷ meetings
cost per meeting = monthly stack cost ÷ meetings
Worked example (the "Realistic" preset):
| Stage | Rate | Survivors |
|---|---|---|
| Sourced | — | 10,000 |
| Enriched (valid email) | 70% | 7,000 |
| Qualified (ICP fit) | 35% | 2,450 |
| Replied | 4% | 98 |
| Positive reply | 35% | ≈34 |
| Meeting booked | 70% | ≈24 |
At a $1,500/month stack, that's about $63 per booked meeting — genuinely good. Now notice the leverage: doubling reply rate (better personalization) doubles meetings. Doubling raw leads does too, but burns domain reputation and list quality. The rate levers beat the volume lever.
Where "the agent sells itself" actually breaks
Deliverability physics
Mailboxes cap safe sending at ~30–50 cold emails/day each, need 2–4 weeks of warm-up, and one bad bounce streak can blacklist a domain. Volume has a hard ceiling per domain — automation doesn't repeal it.
The demo-to-close gap
Automation reliably gets you to a booked call. Closing still needs a human (or a very good product-led flow). Funnels that claim "closed deals, zero humans" are usually counting sign-ups, not revenue.
Personalization decay
When every vendor uses an LLM for the "I saw you just raised a Series A" line, reply rates fall for everyone. The durable edge is a real trigger signal others don't have — not nicer wording.