Campaign Recycling Playbook

Every account in your TAM sits in a live campaign, and the ones who never answered come back around every 60 days.

• DiscoLike maps every company that fits your ICP, and EmailBison runs the campaigns.
• Each pass is one or two short emails, never a long sequence.
• A contact who finishes a pass without replying hears from you again 60 days later, with new messaging.
• Every two months keeps you in front of buyers whose timing just changed, without wearing out the ones who aren't ready yet.
• Each cycle is a fresh experiment on the same market, never the same email twice.

• DiscoLike
• EmailBison

Every reply is read and sorted, so the next cycle learns from all of them.

• Jev and OpenRouter sort each reply:
  • Interested: routed to a rep straight away
  • Objection: tagged with its reason, such as price, timing or a competitor
  • Not now: set to come back when they said to
  • Wrong person: the right one is found and added
  • Opt-out: never contacted again
• They classify only. Neither one enriches.
• Auto-replies are filtered out before anything is counted.

• Jev
• OpenRouter

New messaging comes from what buyers said, what is changing in their industry, and what your engaged leads have in common.

• Objections and calls: the objections in replies, plus what buyers said on recorded calls in Gong.
• Industry trends: Serper pulls the news, launches and rule changes in your buyers' market, so a new email can open on something current.
• Engaged-lead signals: Sumble shows what your engaged leads share, such as a hiring push or a tool in their stack. That becomes a filter or a new signal.

• Gong
• Serper
• Sumble

Claude Code turns what it learned into a few new experiments for the next cycle.

• Claude Code drafts each experiment with its hypothesis, its segment and the copy.
• Each one changes one thing: the angle, the offer, the segment or the signal.
• Each experiment gets at least 1,000 sends before it's judged on replies and meetings.
• Winners from the last cycle stay in place as the control.

• Claude Code

Everything runs in the background. You only step in when a new experiment needs your judgment.

• A Slack message lays out each experiment: what it tests, why, and the copy.
• Approve it, edit it, or turn it down. Nothing new sends without a yes.
• Buying new domains and inboxes waits for the same yes.

• Slack

Sending infrastructure is replaced before it hurts delivery, and lists that work are grown.

• Deliverability comes before the offer. If emails aren't landing, no test result can be trusted.
• Each sending domain is checked daily against blocklists such as SURBL and Spamhaus, along with its bounce rate.
• A listed or burned domain comes out of rotation, and new Google and Microsoft inboxes take its place.
• New inboxes warm up before they send, and warmup never switches off.
• A campaign pauses on its own if bounces pass 3%.
• When a segment works well, DiscoLike finds lookalikes of the accounts that replied, and they join the next cycle.

• Google
• Microsoft
• DiscoLike

Anyone who replied, booked, bounced or opted out is out of the recycle for good.

• A contact is recycled only if:
  • They finished a sequence
  • They never replied, booked or opted out
  • They never bounced
  • They are not mid-sequence in another live campaign
• Checked across every campaign in the workspace, never one campaign at a time.
• Customers and open deals are removed before each cycle.

People change jobs between cycles, so each contact is checked and replaced where needed.

• Anyone who left the company is replaced by whoever holds the role now.
• New accounts from list expansion get their buying committee found.
• The enrichment waterfall uses IcyPeas, Prospeo, LeadMagic and GetLeads.
• Each vendor only sees the contacts the ones before it missed.

• IcyPeas
• Prospeo
• LeadMagic
• GetLeads

Every address is checked twice again before the next cycle sends, even if it passed last time.

• MillionVerifier checks every address first, then BounceBan checks every address that passes.
• Catch-alls go out only if BounceBan marks them deliverable, from their own campaign so their results read separately.
• Addresses go bad between cycles, so nothing sends on last cycle's check.
• Every 60 days the cycle relaunches with the approved experiments, and the loop starts again.

• MillionVerifier
• BounceBan