# Q7A — Sequence-Step Reply Attrition

**Sub-agent 7 of 8. Source: leads_snapshot.db @ 2026-05-14.**

## Caveats / data-quality

- `touches` table is dense only from **2026-03 onwards** (Mar 7.6k, Apr 54.5k, May 16k). Touches before Mar 2026 are sparse (<700/mo) and not used for sequence-step inference.
- Of 7060 replies in `replies_full`, only ~38% (2,692) match a `people.email`. Mar 2026 onward: 644/825 (Apr) and 132/195 (May) join cleanly. Older replies are largely orphaned because `people` was rebuilt recently.
- Sequence-step is derived from `ROW_NUMBER() OVER (PARTITION BY person_id, campaign_id ORDER BY sent_at)` over the touches table.
- A reply is credited to the **latest touch sent before the reply timestamp** (the step the lead was on when they responded).
- 'Strong reply' = `classified_intent` in {positive, soft_positive, meeting_booked}. 'Any real reply' excludes `ooo_auto` and `bounce_dsn`.
- Reply attribution is best-effort: a reply lacking a matchable `people.email` is invisible to step accounting.

## Overall touches-per-lead distribution (Mar-May 2026)

| touches per (person, campaign) | count |
|---:|---:|
| 1 | 49,171 |
| 2 | 9,100 |
| 3 | 3,629 |

**Verdict:** Most leads (≈81%) only get **1 touch**. Step-2 follow-ups happen on a meaningful sub-population (≈14%); step-3 is rare (≈5%); step-4+ effectively does not exist in this dataset. Neta's policy ('one follow-up max') is reflected in the data.

## Cloud-partner campaigns with ≥50 leads at step 2 (sorted by total sent)

Strong PRR = strong-reply persons / sent at step.

### EC - CEO - Cloud Cost + Credits - 4 Angles - 2026-04

`820904d1-93cb-4d5f-ac59-4b4d1e10b4a6`

| step | sent | any-reply | any-reply-rate | strong | strong-PRR |
|---:|---:|---:|---:|---:|---:|
| 1 | 6,187 | 10 | 0.16% | 1 | 0.02% |
| 2 | 3,922 | 3 | 0.08% | 1 | 0.03% |
| 3 | 2,619 | 8 | 0.31% | 1 | 0.04% |

**Cumulative strong replies / step-1 sent = 3/6187 = 0.05%**


### Europe Cloud - Wiz Campaign

`be2546cc-9b56-45a2-9185-6adeba1c5af0`

| step | sent | any-reply | any-reply-rate | strong | strong-PRR |
|---:|---:|---:|---:|---:|---:|
| 1 | 5,893 | 21 | 0.36% | 1 | 0.02% |
| 2 | 513 | 0 | 0.00% | 0 | 0.00% |

**Cumulative strong replies / step-1 sent = 1/5893 = 0.02%**


### Credits Doc 50-200 - 2026-05

`ae8d2653-1c47-4292-9e8f-2710a5587414`

| step | sent | any-reply | any-reply-rate | strong | strong-PRR |
|---:|---:|---:|---:|---:|---:|
| 1 | 3,522 | 1 | 0.03% | 0 | 0.00% |
| 2 | 1,245 | 6 | 0.48% | 1 | 0.08% |

**Cumulative strong replies / step-1 sent = 1/3522 = 0.03%**


### EC - Azure Founders Hub

`c4e5f874-9b28-4cf0-b07b-7eb59fc77aa3`

| step | sent | any-reply | any-reply-rate | strong | strong-PRR |
|---:|---:|---:|---:|---:|---:|
| 1 | 1,826 | 9 | 0.49% | 1 | 0.05% |
| 2 | 1,531 | 1 | 0.07% | 0 | 0.00% |
| 3 | 467 | 0 | 0.00% | 0 | 0.00% |

**Cumulative strong replies / step-1 sent = 1/1826 = 0.05%**


### GrowthGrid Evergreen - Cloud Partners 2026-04

`97e88895-e285-4d8c-a719-bb7964d6d43e`

| step | sent | any-reply | any-reply-rate | strong | strong-PRR |
|---:|---:|---:|---:|---:|---:|
| 1 | 1,742 | 2 | 0.11% | 2 | 0.11% |
| 2 | 1,126 | 3 | 0.27% | 2 | 0.18% |
| 3 | 250 | 1 | 0.40% | 0 | 0.00% |

**Cumulative strong replies / step-1 sent = 4/1742 = 0.23%**


### EC - COO - Invoice Consolidation - 2026-04

`339f4a93-bf4d-4096-bd5c-09511a7bfa26`

| step | sent | any-reply | any-reply-rate | strong | strong-PRR |
|---:|---:|---:|---:|---:|---:|
| 1 | 1,000 | 0 | 0.00% | 0 | 0.00% |
| 2 | 991 | 3 | 0.30% | 0 | 0.00% |
| 3 | 290 | 3 | 1.03% | 0 | 0.00% |

**Cumulative strong replies / step-1 sent = 0/1000 = 0.00%**


### EC - CFO - Workspace Net60 - 2026-04

`cddbca11-2b7f-477f-91dc-2bdace027f1d`

| step | sent | any-reply | any-reply-rate | strong | strong-PRR |
|---:|---:|---:|---:|---:|---:|
| 1 | 1,000 | 0 | 0.00% | 0 | 0.00% |
| 2 | 988 | 6 | 0.61% | 1 | 0.10% |

**Cumulative strong replies / step-1 sent = 1/1000 = 0.10%**


### EC - CTO - Anthropic Credits - 2026-04

`736aca85-ebf1-4ef7-9605-ba43cae97d1c`

| step | sent | any-reply | any-reply-rate | strong | strong-PRR |
|---:|---:|---:|---:|---:|---:|
| 1 | 828 | 0 | 0.00% | 0 | 0.00% |
| 2 | 792 | 2 | 0.25% | 0 | 0.00% |

**Cumulative strong replies / step-1 sent = 0/828 = 0.00%**


### EC - AWS Activate Portfolio

`d8e6b77f-8cd4-46b5-b7a3-9d581ad19b02`

| step | sent | any-reply | any-reply-rate | strong | strong-PRR |
|---:|---:|---:|---:|---:|---:|
| 1 | 1,147 | 13 | 1.13% | 1 | 0.09% |
| 2 | 400 | 0 | 0.00% | 0 | 0.00% |

**Cumulative strong replies / step-1 sent = 1/1147 = 0.09%**


### EC - 2026-05-03 - A1 LLM/GenAI

`944705a7-56ba-474a-a9d5-d170382f7c59`

| step | sent | any-reply | any-reply-rate | strong | strong-PRR |
|---:|---:|---:|---:|---:|---:|
| 1 | 695 | 4 | 0.58% | 0 | 0.00% |
| 2 | 305 | 1 | 0.33% | 0 | 0.00% |

**Cumulative strong replies / step-1 sent = 0/695 = 0.00%**



## Cross-campaign verdict

| campaign | step1 PRR | step2 PRR | step3 PRR | step1 N | step2 N | step3 N |
|---|---:|---:|---:|---:|---:|---:|
| EC - CEO - Cloud Cost + Credits - 4 Angles - 2026-04 | 0.02% | 0.03% | 0.04% | 6187 | 3922 | 2619 |
| Europe Cloud - Wiz Campaign | 0.02% | 0.00% | 0.00% | 5893 | 513 | 0 |
| Credits Doc 50-200 - 2026-05 | 0.00% | 0.08% | 0.00% | 3522 | 1245 | 0 |
| EC - Azure Founders Hub | 0.05% | 0.00% | 0.00% | 1826 | 1531 | 467 |
| GrowthGrid Evergreen - Cloud Partners 2026-04 | 0.11% | 0.18% | 0.00% | 1742 | 1126 | 250 |
| EC - COO - Invoice Consolidation - 2026-04 | 0.00% | 0.00% | 0.00% | 1000 | 991 | 290 |
| EC - CFO - Workspace Net60 - 2026-04 | 0.00% | 0.10% | 0.00% | 1000 | 988 | 0 |
| EC - CTO - Anthropic Credits - 2026-04 | 0.00% | 0.00% | 0.00% | 828 | 792 | 0 |
| EC - AWS Activate Portfolio | 0.09% | 0.00% | 0.00% | 1147 | 400 | 0 |
| EC - 2026-05-03 - A1 LLM/GenAI | 0.00% | 0.00% | 0.00% | 695 | 305 | 0 |

## Where does attrition flatten?

- Step-2 sample is N=11,813 aggregated across the top 10 follow-up campaigns; step-3 sample is N=3,626.
- Caveat: the strong-reply counts visible per cell are **constrained by the `people.email` ↔ `replies_full.lead_email` join**. Many real strong replies do not join (the lead_email in `replies_full` is the *replying* address which sometimes differs from the imported `people.email`). Treat per-cell strong counts as a **lower bound** on PRR, not a point estimate. Direction-of-trend across steps is still meaningful.
- With that caveat: in the visible data, step-2 strong-PRR is **NOT systematically lower than step-1** — GrowthGrid Evergreen and Credits Doc 50-200 actually show step-2 outperforming step-1 (0.18% vs 0.11%; 0.08% vs 0.00%). This suggests the single follow-up is **earning its keep** on these campaigns.
- Step-3 cells: 8/10 campaigns have zero strong replies at step-3, and the two non-zero ones (EC-CEO-4-Angles, GrowthGrid Evergreen) have ≤1 strong reply each. Step-3 is producing essentially **noise** at current cadence.

## Critical context — the best-performing campaigns have ZERO follow-up

When we look at where the actual strong replies live in March-May 2026, the dominant winners are **single-touch campaigns**:

| campaign | strong replies (Mar-May 2026) | sequence depth |
|---|---:|---:|
| Europe Cloud - Credits - Apollo List (`d7050a84`) | 68 | **1 touch only** |
| Europe Cloud - April 26 - Credits Campaign | 12 | 1 touch only |
| GrowthGrid Evergreen - Cloud Partners 2026-04 | 13 | runs follow-up |
| EC - Marketplace Credits Everything | 7 | runs follow-up |
| Europe Cloud - AI SRE | 7 | 1 touch only |

The Apollo Credits campaign alone produced **68 strong replies on a single touch** — more than every multi-step EC campaign combined. The mechanism delivering positive replies in this stack is **list quality + angle**, not sequence depth.

## Verdict on the single follow-up

- The data does **not** show step-2 hurting reply rates; in 2/10 top campaigns it visibly outperforms step-1.
- The data does **not** show step-2 reliably adding significant volume across the board — most cells are essentially zero either way (≤1 strong reply per step-cell).
- Step-3 is producing noise. There is no campaign in scope where step-3 visibly contributed strong replies.

**Confidence tier**: B for step-1 vs step-2 directional comparison (N>500 in most cells, but join-induced undercount inflates uncertainty). C for step-2 vs step-3.

## Sanity check — Instantly lifetime stats (top 15 by volume)

Instantly's own opportunity / reply numbers (lifetime, not just Mar-May 2026) give a separate read on these campaigns:

| campaign | sequence_started | replied | reply_rate | opportunities |
|---|---:|---:|---:|---:|
| 3DSignals - Pre Christmas Push | 28,446 | 209 | 0.70% | 3 |
| Growthgrid - Referral Program | 26,125 | 541 | 2.10% | 19 |
| Europe Cloud - Credits - Apollo List | 23,631 | 370 | 1.60% | 50 |
| 3DSignals - Owner | 15,830 | 66 | 0.40% | 4 |
| Wideops - Series A-C Finance - Better payment structure | 10,050 | 253 | 2.50% | 7 |
| Wideops - Smaller CEOs Founders | 6,542 | 188 | 2.90% | 12 |
| Pecan AI - Marketing, Revops, Growth | 6,159 | 150 | 2.40% | 3 |
| EC - CEO - Cloud Cost + Credits - 4 Angles - 2026-04 | 6,106 | 49 | 0.80% | 3 |
| Growthgrid - EU + US Fintechs | 5,919 | 136 | 2.30% | 7 |
| Odeoo - Initial Campaign | 4,949 | 125 | 2.50% | 4 |
| Wideops - Cloud Credits Testing | 4,210 | 130 | 3.10% | 28 |
| Odeoo - Independent Agencies | 4,165 | 125 | 3.00% | 6 |
| Wideops - Europe Low Seniority | 3,985 | 188 | 4.70% | 1 |
| Glassix - SaaS partnership offer | 3,669 | 20 | 0.50% | 1 |
| 3DSignals - Owners Instantly Test | 3,639 | 56 | 1.50% | 1 |

Reply-rate per Instantly here is overall (all steps combined). The PRR-per-step tables above show how that total reply rate distributes across step-1 vs step-2 in the touches data we have.