---
name: copy-fatigue-rotation
description: Plans rotation of cold email templates so no single template is sent until it wears out. Groups templates into structurally different families (opener device, structure, call-to-action type, length), caps daily volume per family, detects fatigue from the step-1 reply rate against the family's own recent peak, rests tired families and adds canary slices of identical leads to separate copy effects from list effects. Use when the user says reply rates are dropping on a template that used to work, asks how often to change copy, how many templates to run, or how to rotate spintax variants.
---

# Copy fatigue and rotation

You help keep cold email copy productive over time. Templates wear out after weeks of heavy sending: replies fall on the same list and the same mailboxes. The fix is structural rotation and rest, measured on a clean metric.

## Step 1. Inventory

Ask for every live template, the date it started, daily volume per template, and daily step-1 sends and replies per template for the last 2 to 4 weeks. Ask whether lists change between days.

## Step 2. Group into families

Two templates are in the same family if they share all four of:

1. Opener device (question, condition, observation, referral, trigger event).
2. Structure (order of offer, proof and ask).
3. Call-to-action type (fit question, interest check, permission to send info).
4. Length band (under 40 words, 40 to 70, over 70).

Synonym swaps and spintax variants do not make a new family. Filters and readers recognise the skeleton, not the words.

## Step 3. Measure fatigue correctly

- Use the **step-1 reply rate** (human replies to email 1 divided by email 1 sends), per family, per day.
- Compare a family only with itself, on a comparable list. A blended reply rate across all steps and lists moves when the mix of lists or steps changes, even if no template got worse.
- Ignore days with fewer than about 1,000 step-1 sends per family. The noise is larger than the effect.
- **Fatigue rule:** a family is tired when its step-1 reply rate is below 50% of its own 7-day peak for 2 days in a row.

## Step 4. Rotation plan

- Keep 4 to 6 families live at a time, with a pool of 10 or more written.
- Cap daily volume per family so no family carries most of the traffic. Spread volume roughly evenly.
- Rest a tired family for 4 to 6 days, then bring it back on a fresh list and watch whether it recovers. If it does not recover after rest, retire it.
- Replace retired families with structurally new ones (step 2), not with a reworded copy.
- New families start at a small daily volume and ramp over a week.

## Step 5. Canary slices

Every day, send a small slice of leads drawn from the same list, split at random, to every live family (for example a few hundred per family). The canary compares families on identical leads, so a drop that shows up in every family points to the list or the infrastructure, and a drop in one family points to that copy.

## Output format

1. Family table: family, templates, start date, daily volume, 7-day peak step-1 reply rate, latest, status (live / tired / resting / retired).
2. Actions for the next 7 days.
3. Canary setup: slice size, list, assignment rule.

## Rules

- Before blaming copy, rule out the bigger causes: bounces, spam placement, a new list source, a broken merge field, a paused mailbox group.
- Do not declare a winner or a fatigue on a handful of replies.
- Keep the family label in variant or campaign names so replies can be grouped.

## Example

Input: 3 templates, all "Quick question about {company}'s IT support" with synonym changes, 3,000 step-1 sends a day each, reply rate down from 1.2% to 0.5% over 3 weeks.

Output: the 3 templates are one family (same opener, structure, CTA and length). Tired: below 50% of peak for 5 days. Rest it for 5 days. Launch 3 new families: a condition-first fit question, a trigger-event opener, and a short two-sentence permission ask. Cap each at 2,500 a day. Canary: 300 leads a day from one list, split across all live families.