The AI Cold Outreach Framework That Gets a 35%+ Reply Rate

The average cold email gets a reply rate somewhere south of 5%. The sequences we run for clients regularly clear 35%. The difference isn't a magic subject line or a better sending tool — it's a framework that makes every message feel like it was written for one person, because in the way that matters, it was. Here's the whole thing.

Why generic outreach fails

Prospects can smell a blast in half a second. "Hi {first_name}, I hope this email finds you well" is an instant delete. Worse, AI made it easierto send bad outreach at scale, so inboxes are more defensive than ever. The bar to earn a reply is now genuine relevance — and that's exactly where a well-prompted AI shines, if you point it at the right inputs.

The R-E-P-O framework

Every message we send follows four beats. Think of it as the skeleton the AI fills in:

• R — Relevance hook. One line proving you did your homework: a trigger event, a recent post, a specific detail about their business.
• E — Empathy. Name the problem someone in their exact role is likely living with right now.
• P — Proof. One concrete, quantified result you produced for someone like them.
• O — One ask.A single, low-friction question — never "hop on a 30-minute call." "Worth a quick look?" converts far better.

The prompt that builds it

Here is the structure of the prompt we give the model for each prospect. Feed it the enriched data, and it returns a draft that hits all four beats:

"You are writing a cold email to [name], [title] at [company]. Here is what I know: [trigger event], [company context], [their likely pain]. Write a 70–90 word email using this structure: a first line that references the trigger naturally, one line of empathy about [pain], one sentence of proof using this result: [your proof], and a single soft question as the CTA. Plain language, no buzzwords, no 'I hope this finds you well.' If the context is too thin to personalize honestly, say so instead of inventing details."

That last sentence is the guardrail that protects your reputation. A fabricated compliment does more damage than a missed send.

Sequencing: the reply lives in the follow-up

Most positive replies don't come from email one. They come from email two or three. Run a four-touch cadence over about ten days:

• Day 1 — the R-E-P-O email. Your strongest, most personalized shot.
• Day 3 — a one-line nudge.Add a single new angle or a relevant resource. Never "just bumping this."
• Day 6 — a different channel. A short, friendly LinkedIn note referencing the email. Multi-channel lifts replies dramatically.
• Day 10 — the break-up."Sounds like the timing isn't right — want me to close the loop?" This polite exit reliably pulls the most replies of the whole sequence.

Protect deliverability or none of this matters

The best email on earth gets a 0% reply rate from the spam folder. Send from a warmed-up secondary domain, keep daily volume modest, verify every address before sending, and strip links and images from the first touch. Boring infrastructure work — but it's the foundation the whole framework stands on.

Put it together

A tight list, the R-E-P-O framework, an honest personalization prompt, a four-touch multi-channel cadence, and clean deliverability. None of it is exotic. The reason almost nobody runs it is that it takes a system to do consistently — which is exactly what turns a 5% reply rate into 35%.