Launch to a qualified list | Academy

The presenter plans to contact 5,000 roofing companies. The final expert rejects that approach and recommends a first batch of 10 to 15 companies with a current, visible reason to consider better video.

The expert replaces a 5,000-company blast with a small, evidence-based test.

Qualify with current evidence

Useful signals include a company running paid ads, relying on old video, or spending heavily while the visible creative has a specific weakness. Record the asset URL, the date you observed it, and the exact gap. Do not infer campaign performance from appearance alone.

The company name and city change, but every recipient receives the same reason for contact.

Copy this asset

Find 10 to 15 [niche] companies in [market] with a clear, current, publicly evidenced reason to test better video ads — for example, actively running ads with outdated or visibly weak video creative. For each company, return the evidence URL, the observation date, the exact creative gap, and one truthful, evidence-specific opening line. Do not infer campaign performance without performance data. Exclude any company you cannot verify from a public source.

Recreate

Check every returned company and opening yourself. Claude can accelerate discovery, but it can also return stale assets or write a stronger claim than the evidence supports.

Decide what the first batch will measure

Define the observation window and the calculation for each rate before sending:

Metric	Calculation
Delivery rate	Delivered messages / sent messages
Qualified reply rate	Qualified replies / delivered messages
Sample rate	Sample-ready profiles / qualified replies
Paid conversion	Paid clients / sample-ready profiles

Keep a launch log with the prospect, evidence URL and date, delivery status, reply, sample status, payment, direct cost, and review time. Do not use reply rate alone to justify scaling.

The expert suggests that a highly targeted batch could receive many replies. Treat that as a hypothesis. Your own locked log is the evidence.

Apply the jurisdiction and compliance checks from Lesson 6. A named human should approve each recipient and message in this first batch, and prior opt-outs must remain suppressed.