SDRfollow-up-cadence · intermediate

AI-personalized 7-touch follow-up cadence

Last reviewed: 2026-05-23 · saves ~3 hrs/week/run

Our take

Default Outreach/Apollo cadences are templated and prospects know it by touch 3. AI can personalize each touch based on prior touch engagement (opens, clicks, replies). The lift on a 7-touch cadence is typically 1.5-2x reply rate when each touch references the prior context.

Tool stack

Apollo.io

sales-engagement

Lavender

ai-email-coach

Anthropic Claude API

llm-platform

Steps

Build a 7-touch cadence in Apollo with the structure below.
Each touch has a base template + an AI personalization field that pulls from prior engagement.
Run the personalization prompt before send; AI writes the unique opener for each touch.
Lavender quality-checks each draft.
Touch 7 is a breakup — handcrafted, not AI; this matters.

Prompts

Personalize touch N based on prior engagement · GPT-4o-mini

Inputs:
- Prospect basics (name, role, company)
- Previous N-1 touches I sent (full text)
- Engagement on prior touches (opened y/n, clicked y/n, replied y/n)
- Recent LinkedIn activity (last 14 days)

Generate the next touch opener:
- If they opened but didn't reply: reference value differently. Not the same hook.
- If they didn't open: try a sharper subject line + different angle.
- If they clicked but didn't reply: ask a specific question about the asset they clicked.

Constraints:
- Under 75 words total per email
- No "Just following up"
- No "circling back"
- No "checking in"
- Always have a single specific ask (not "do you have time to chat")

Pitfalls

Don't fully automate. AI personalization at touch 1-6 is fine; touch 7 (breakup) should be human-written or it sounds desperate.
Avoid 'I noticed you opened my email' — sounds creepy. Reference content, not behavior.
If reply is positive, immediately stop cadence. Don't let touches 4-7 fire after a positive reply.

Last reviewed 2026-05-23. Independent.